The first David Protein bar went on sale 24 months ago. Last week, Medici Brands, the company behind it, raised $250 million at a $2.25 billion valuation, roughly triple where it was a year ago.

My first reaction was that valuation is absurd for a company built around a two-year-old protein bar. Then again, this is Peter Rahal we are talking about. The Medici founder and CEO had already co-founded RXBAR and sold it to Kellogg.

Never bet against Peter.

Drew Fallon, co-founder and chief executive of CPG finance firm Iris Finance, put that speed in perspective in the chart below.

Sure, it is not a perfect comparison. The chart mixes financing valuations with acquisition prices, and some dates and deal terms are estimates.

What I wanted to know was what investors believed they were buying besides the protein bar. Rahal’s exit explains part of Medici’s speed, but not all of it. The rest is how one product claim became an owned ingredient and a parent company with three brands, and why investors are willing to pay a multiple that established food businesses do not get.

David started with the advantages RXBAR had to earn

Rahal and Jared Smith started RXBAR in 2013 in his parents’ basement with $10,000 and no outside investors. The bar was four ingredients: egg whites, almonds, cashews and dates. The idea that made the company was the wrapper. RXBAR printed the ingredients on the front of the pack in large type, with the quantities, so the label read “3 egg whites, 4 cashews, 2 dates”. Nothing to turn over, nothing to look up.

It was a hit. RXBAR did $600,000 in its first nine months and $2 million in 2014, and by 2017 it was doing an estimated $160 million when Kellogg bought it for $600 million. Rahal’s share was roughly $300 million before tax.

The sale came with a five-year non-compete that expired in October 2022. During that time, Rahal set up the family office Litani Ventures, then spent eighteen months as a managing partner at CPG fund Humble Growth. When he returned to operating in 2024, he brought the capital, track record and investor network created by RXBAR.

He used the liquidity immediately. When David announced a $10 million seed in August 2024, Rahal led the round himself with $6 million of his own money, with Valor Siren Ventures, Peter Attia, Andrew Huberman and Layne Norton beside him.

Rahal could afford to fund formulation and inventory before the first bar was sold. That was a massive advantage, but it still left all the operating work. Rahal and his team had to formulate the bar, create demand and win shelf space.

The money gave Rahal a head start. It did not give anyone a reason to buy the bar.

The boiled cod was the strategy

David gave them one: 28 grams of protein in 150 calories. That works out to roughly 75% of calories from protein against about 40% for a Quest bar.

But reaching those numbers meant giving up RXBAR’s defining promise: a short list of familiar ingredients. David’s bar carries allulose, maltitol, and a modified fat called EPG (Esterified Propoxylated Glycerol), which delivers about 0.7 calories per gram where ordinary fat delivers 9.

EPG is what lets both numbers work because it gives the bar the texture and richness of fat with a fraction of the calories. Rahal took the ultra-processed criticism as the cost of a claim a buyer can repeat back. In his own words, “the protein-to-calorie ratio concept is a really powerful message.”

Owning a metric has a strange consequence. In 2025, David started selling $69 frozen Pacific cod fillets and ran cod billboards across New York, on the grounds that boiled cod has a higher protein-to-calorie ratio than a David bar. The company partnered with an Alaskan fisherman to pack it under its own name. Other brands made parodies, but David kept going.

Selling frozen cod to advertise a protein bar is ridiculous, but that is the point. It’s so David. By admitting that cod had the better ratio, the company made the comparison feel more credible and kept attention on the metric it wanted people to use.

Cod may win on protein per calorie, but it does not fit in a gym bag. David could call itself second-best and still sell the more convenient product. According to the company, the cod campaign increased its engagement on X by 14,000% in its first week.

The provocation kept going up. To launch its Bronze bars, Julia Fox appeared on billboards declaring “Men disappoint. David satisfies”, while creators received PR boxes containing bars and vibrators.

For the ice cream launch, David drove a truck with a jacked cow around Expo West. Even its subway ads ran without copy. Each stunt was designed to be photographed, mocked and shared. David did not need everybody to like the brand. It needed them to remember 28 grams and 150 calories.

David had two quarters of sales before it asked for shelf space

David was announced in March 2024 and launched online on 16 September, about six months later, reporting roughly $1 million in its first week. A million bars went out in the first six weeks. Fulfilment ran through a third party, so the company scaled orders without building a warehouse.

The retail push started in early 2025. Doors went from about 3,000 in May 2025 to roughly 16,000 by February 2026 to more than 35,000 today, including Walmart, Target and Costco.

By the time the retail push began, David could show buyers two quarters of demand. It was asking them to scale evidence it already had.

Then David bought the only company that could make its numbers work

On 9 May 2025, David acquired Epogee, the sole producer of EPG, for $75 million, according to filings in the antitrust case that followed. Three weeks later, it announced the acquisition and a Series A led by Greenoaks. Interestingly, the round was also $75 million: exactly what Epogee cost. Coincidence?

Rahal told AgFunderNews that David’s demand was already 120% of Epogee’s capacity when the acquisition happened and eventually reached 150%. By April 2026, AgFunderNews reported a fivefold expansion in EPG capacity, with Epogee courting commercial partners. Rahal said none of the companies in those talks were direct competitors.

The acquisition secured the key ingredient behind David’s numbers, gave Medici control over future capacity and created the possibility of a second business selling EPG outside its own brands.

That is where I stopped thinking of this as a protein-bar company. Medici was buying the technology behind the claim.

The same day David announced the deal, it told three other EPG customers that their supply would be wound down. OWN Your Hunger, Lighten Up Foods and Defiant Foods had each built products around the ingredient.

In June 2025, they sued in the Southern District of New York, claiming stranded R&D, lost sales and specialised inventory. One told the court the acquisition forced it to abandon its business.

Lighten Up has since settled after its founder received a supply agreement to restart his sauce. OWN Your Hunger and Defiant remain plaintiffs and say EPG is still unavailable to them. So far, access has reopened selectively and on Medici’s terms.

HallPass revealed what Medici was building

In February 2026, Rahal used an interview with The New Consumer to lay out the wider plan, and Medici Brands was introduced as the parent. Frozen dessert pints went on sale online in June and into Target over the summer, and ready-to-drink shakes launched in August.

Then came the second brand. HallPass, co-founded with Michael Tierney of Stuffed Puffs, launched nationally at Walmart in late August with peanut cups, crispy wafers and chocolatey candy pieces at 70 calories and 1g of sugar.

Rahal said Walmart would provide “the vast majority of our distribution day one,” and told Inc. he wants HallPass to be the Coke Zero of candy. Rowdy, a chip brand, was revealed soon afterwards and is due before the end of the year.

HallPass drops the protein claim entirely, but EPG remains in its fat system. The nutritional promise changed from protein density to 70 calories and 1g of sugar. The provocation stayed. HallPass shipped product images with visible typos, and Tierney told Business Insider they left them in “for a little bit of internet rage bait.”

Brand two reveals the common system. What carries from David to HallPass, and eventually Rowdy, is not protein or any single category. It is EPG and a way of getting attention.

One Reddit community offers a small but telling example. David is treated as a villain in protein-snack forums, yet the most upvoted HallPass post in a low-calorie food community read: “I know they’re owned by david protein, but these are worth it anyways.” Jenna Movsowitz described the psychology neatly: David’s antagonism can make buying it feel like an act of rebellion.

Yes, I know: Reddit is one community, not the market. But it is an early sign that the product logic and attention strategy may travel even when David’s goodwill does not.

David proved the model. Investors are paying for the repeats.

One limitation before comparing Medici with other food companies: almost every available number comes from the company or an outside estimate.

Sacra estimates that David grew from $8 million in 2024 to $102 million in 2025. David says it passed $100 million in its first year and expects more than $300 million in 2026.

None of those figures appears in audited accounts, and the 2026 number is still a forecast. Medici also does not disclose gross margin, repeat purchase, retail velocity or how much cash its growth consumes.

On the company’s own 2026 projection of more than $300 million, $2.25 billion is about 7.5x revenue. On Sacra’s estimate of 2025 actuals it is closer to 22x.

These are rough headline comparisons and not clean enterprise-value multiples, because Medici’s cash and debt are not public. They still show the size of the premium. Five factors help explain it.

1. Medici is growing much faster: If Sacra’s estimates and David’s forecast are roughly right, revenue grew more than twelvefold in 2025 and could nearly triple in 2026. Chomps is projected to grow about 36% this year. Investors are paying for the rate of change, although David’s larger number remains a forecast from the company being valued.

2. The comparison is not like-for-like: Medici’s valuation comes from a funding round, while Quest, Poppi and RXBAR were sold outright. That difference does not explain the full premium, but it makes the headline multiples less exact than they look.

3. Investors are also underwriting Rahal: He has already built and sold one major protein-bar company. At David, he went from launch to more than 35,000 retail doors, acquired the brand’s critical supplier and introduced a second brand in two years. His execution makes the claim that Medici can repeat the process more credible.

4. EPG may add value beyond a food brand: Medici controls its sole producer, and capacity has expanded fivefold. If outside licensing grows, Epogee becomes another source of revenue. That opportunity is real, but its value is not yet visible.

5. The price includes businesses David has not built yet: Apply the 3x to 4x multiples paid for established food brands to David’s projected revenue and it is worth roughly $0.9 billion to $1.2 billion. The rest has to come from Epogee, new David formats, HallPass, Rowdy, future brands or a premium for the platform tying them together.

Rahal mentioned the same limit on The Peel in July 2026. He said David has a revenue ceiling somewhere between $1 billion and $2 billion, and warned against forcing products into the brand simply to grow past it.

Medici is the structure above that ceiling, holding cash, talent and product development across multiple brands. It is how Rahal plans to keep growing once David reaches its limit.

Now Medici has to grow into the valuation

Assume the $2.25 billion figure is a post-money valuation, the new investors want to triple their money over five years, and no further dilution occurs. That return target is mine, not something Greenoaks has disclosed.

On those assumptions, Medici would need to be worth about $6.75 billion at exit. The revenue required depends on what multiple the eventual buyer is willing to pay.

At the 3x to 4x multiples paid for Quest, Poppi and RXBAR, Medici would need to grow from a projected $300 million to between $1.69 billion and $2.25 billion in five years. That is annual growth of roughly 41% to 50%. If a buyer accepts the platform argument and pays 5x, the requirement falls to about 35% a year.

The round therefore depends on two things. Medici has to compound at a rate very few packaged-food businesses sustain, and the market at exit has to price it as something other than a bar company. Miss either and the numbers above stop working.

Chomps’ valuation leans on performance already visible. Medici’s leans much more heavily on performance yet to arrive. Neither valuation is necessarily wrong. Each places the burden of proof somewhere different.

Medici’s advantages come with their own risks

The EPG litigation tests the ingredient moat

On 4 February 2026, Judge Victor Marrero dismissed the plaintiffs’ complaint and denied their request for a preliminary injunction. They have since filed again. The dismissal turned on their failure to define a coherent market in which David could exercise monopoly power. It did not establish that EPG has ready substitutes.

That leaves an interesting tension. If EPG is easy to replace, the antitrust case weakens but so does the strategic value of owning it. If it is difficult to replace, the acquisition becomes more valuable and control of its supply becomes more consequential.

The labelling cases test whether David can keep making its simple numerical claim

In January 2026, a proposed class action alleged that lab testing found David bars carried 263 to 275 calories against the 150 on the label. David’s position is that its count reflects what the body absorbs. That suit was voluntarily dismissed on 31 March 2026 without prejudice, so it can be brought again, and the company said it remains confident in its labelling.

In late August, a second case challenged the zero-sugar claim on the grounds that allulose is legally a sugar. David is one of several defendants in a wider wave of allulose cases. In a related case against Chobani, the FDA filed an amicus brief arguing that the relevant regulation includes allulose within total sugars.

Whatever the courts decide, the commercial risk is clear. David was built on numbers a shopper can understand in two seconds, but those numbers depend on rules that do not map neatly to how every consumer interprets the label.

The tolerability question tests repeat purchase

In an eight-week randomised controlled study of 139 volunteers, seven predefined gastrointestinal adverse effects were reported more frequently among subjects taking 25 or 40 grams of EPG a day. These included gas, diarrhoea and oily stool. At 10 grams a day, it was reasonably well tolerated.

The study was commissioned to support EPG’s regulatory clearance and did not test a David bar. It does establish that dose matters. Medici does not disclose how much EPG a bar contains, so there is no way to tell from outside how many bars would approach the study’s higher doses. Digestive complaints recur in protein-snack forums, but those reports are anecdotes, not incidence data.

Peter Attia became a reputational problem

Attia, one of the most prominent names in David’s seed round, stepped down as chief science officer in early February 2026 after his communications with Jeffrey Epstein became public. David used experts like Attia to support its scientific credibility. That works both ways when one of them becomes controversial.

Look, none of these risks makes the valuation wrong. But they show how much Medici still has to prove: that David can travel beyond bars, that HallPass and Rowdy can hold shelf space, and that EPG can become more than an ingredient Medici primarily sells to itself.

I came into this wondering how a two-year-old protein bar could support a $2.25 billion valuation. I can now see what investors believe they bought: rage bait that keeps people talking, EPG that lets Medici build products competitors struggle to match, and Peter Rahal’s ability to turn both into brands at unusual speed.

What remains to be seen is whether that combination is a repeatable system or simply one extraordinary launch.

I’m Eshan. An operator-turned investor, backing companies in food, agri, nutrition, and health. I’ve been writing Better Bioeconomy since 2023, as a way to share my learnings and to connect with cool people like you. Thanks for reading!

If you found value in this newsletter, consider sharing it with a friend who might like it.

Or, if someone shared this to you, consider subscribing (it’s free!).

Subscribe now

Disclaimer: The views and opinions expressed in this newsletter are my own and do not reflect those of my employer, affiliates, or any organisations I am associated with.

Longevity has become one of the buzziest words in nutrition and health. I have been trying to work out how much of it is new biology and how much is old products with new labels.

The answer may be found one layer upstream of the brands, with the ingredient suppliers who decide which science is worth turning into a product, and what it takes to get it there.

So I recently sat down with Vincent van Buul, Director Innovation, Application & Formulation EMEA at dsm-firmenich. Vincent’s framing, moving from lifespan and healthspan to cellspan, which he shared on NutraIngredients, is what made me want the conversation.

dsm-firmenich was created in 2023 by the merger of Dutch nutrition group DSM and Swiss flavour and fragrance house Firmenich. The company describes itself as a nutrition, health and beauty business. A worldwide team of nearly 21,000 people works across nearly 60 countries, and continuing operations generated revenues above €9 billion.

Its customers are food, supplement, medical nutrition, and pharma companies, which it supports with ingredients, solutions, and expertise spanning nutrition, health, taste, texture, and formulation.

Vincent has a PhD in behavioural nutrition and an MBA, and his career has moved between the science and the commercial end of it. He was Global Head of Dietary Fibre Science and Innovation, worked on innovation strategy under the Global CTO, and spent time in a venturing role analysing startups before that.

His current remit covers the gap the whole category keeps falling into, between an active ingredient that works in a study and a product someone will eat regularly.

In our chat, Vincent shared:

• How a hallmark of ageing earns a place on the build list

• What cellspan means, and why the category needed a new word

• Why the ingredient is only part of the answer

• What separates a healthspan ingredient from a healthspan claim

• How the company decides which ideas to work on

• Where the next frontier in healthy ageing is

Let’s jump in!

A hallmark of ageing is only relevant if nutrition can influence it

The hallmarks of ageing are the biological processes researchers accept as drivers of ageing, as opposed to symptoms of it. To qualify, a process has to show up as you age, speed ageing up when it is made worse, and slow ageing down when it is treated. A 2013 paper named nine, later a 2023 update took the list to twelve:

1. Genomic instability

2. Telomere attrition

3. Epigenetic alterations

4. Loss of proteostasis

5. Disabled macroautophagy

6. Deregulated nutrient-sensing

7. Mitochondrial dysfunction

8. Cellular senescence

9. Stem cell exhaustion

10. Altered intercellular communication

11. Chronic inflammation

12. Dysbiosis

dsm-firmenich builds against four of them. Why those four? Three conditions have to be true:

• Nutrition has to be able to influence the hallmark.

• The effect has to be mechanistic enough to measure.

• It has to translate into a product that can be developed and sold.

A hallmark that matters a lot in ageing but cannot be reached by what you eat does not earn a programme.

The four are chronic inflammation, gut dysbiosis, cellular senescence and mitochondrial dysfunction. Vincent puts chronic inflammation front and centre, largely because of what it does downstream. Small inflammations in the gut and elsewhere set off secondary effects far from where they started.

Gut dysbiosis is the neighbouring one. “There are roughly as many microbial cells living in and on our bodies as there are human cells,” Vincent said, “with a large proportion of those microbes residing in the gut.” Nutrition passes directly through that system, which puts a nutrient in contact with it by default.

Senescence and mitochondrial function cover the cell itself. A senescent cell is not dead. It stays metabolically active, converts nutrients into unhelpful compounds and feeds the same inflammatory load. Mitochondria run the energy side, and the receptors and pathways around them respond to molecules that arrive through food.

What nutrition can reach is what decides that list. The same constraint shapes everything below.

Cellspan brings lifespan and healthspan back together

Okay, that is a lot of “spans”! Here is how Vincent separates them. Lifespan is how long we live, measured in years. Healthspan is how long we stay healthy. Those two numbers should match, but they do not, which is the problem the whole field exists to solve.

Cellspan runs underneath both. He defines it as how long cells remain resilient, functional, and able to repair, communicate and produce energy. It gives his teams something to design against that is smaller and more testable than healthy ageing, and it is the layer where the four hallmarks operate.

When longevity started showing across supplements, medical nutrition and functional foods, his first reaction was that the industry was relabelling itself. “Yet another term to basically say the same thing that this industry has been doing for decades,” is how he put it.

His own example is vitamin C. It has been sold for decades, it supports immunity, a benefit that can contribute to healthier ageing. What has changed is not necessarily the underlying ingredient, but the way consumers think about health: increasingly proactively, with a focus on maintaining function and resilience before disease occurs.

Vincent’s point is that longevity becomes most meaningful when that shift in mindset is matched by a deeper understanding of the mechanisms involved, rather than simply applying a new label to familiar health benefits.

Cellspan is his way of bringing the longevity story back toward mechanism. It asks what is changing at the cellular level, not whether an existing benefit can be renamed.

The science only translates if the dose fits into a product

The cellspan test raises the question: Once you know which biology to target, what stops the product from working?

Vincent points to gut health as a good example. Beyond traditional dietary fibre, his teams look at more targeted ways to influence the gut environment, including human milk oligosaccharides, or HMOs, and Humiome® B2. These approaches show how nutrition can move towards more precise modulation of gut health and dysbiosis.

But the ingredient is only part of the answer. The real challenge is building it into a broader solution with the right dose, format, taste and consumer experience.

That challenge generalises well beyond gut health. Most nutrients taste bad, arriving with off-notes and other barriers to consumption. This is where the flavour and texture half of dsm-firmenich shines. Pairing an active with a flavour system is the step that decides whether the ingredient gets consumed at all.

“If they’re never consumed by someone, nothing will work,” Vincent said of the many promising nutrients in the literature. Taste has to be right, the format has to be one people enjoy, and the price has to be viable. Miss one and the active is irrelevant.

He describes it as a chain:

• The ingredient has to work

• The human evidence has to be credible

• The product has to be something people consume, trust and understand

• A brand ecosystem has to get it to them

“If that whole chain somewhere is broken, there is no impact,” he said. “And then all the efforts of science get lost.” He is a scientist by training, and he wants his time to produce real impact, not what he calls paper impact.

Credible healthspan evidence comes from multiple layers of evidence

Credible human evidence is the second link in that chain, and it is the one every ingredient on the market claims to have. So what separates a healthspan ingredient from a healthspan claim?

“It’s not about one ingredient, one brand, one biomarker or one claim,” Vincent said. He looks for convergence across mechanism, human evidence, formulation, dose, consumer behaviour and how responsibly the whole thing is communicated. Any single element can be bought, but the whole set is harder to fake.

His example of evidence done properly is DO-HEALTH, run by Professor Heike Bischoff-Ferrari. The trial enrolled 2,157 adults aged 70 and older across seven centres in five European countries and followed them for three years.

DO-HEALTH ran a 2x2x2 factorial, splitting participants into eight groups across three variables. Each group either did or did not get:

• Vitamin D, provided as Quali®-D at 2,000 IU a day

• life’s®OMEGA 60, providing 1,000 mg EPA+DHA per day

• A home exercise programme

That structure shows whether interventions add to each other, which no single-ingredient trial can do.

A later post-hoc analysis of 777 participants from the DO-HEALTH cohort, published in Nature Aging in 2025, applied four DNA methylation clocks to the cohort. Methylation is a chemical tag that attaches to DNA without altering the DNA itself. The pattern those tags form shifts as you age, and it shifts regularly enough to be read as a measure of ageing.

The clocks are algorithms that read the pattern across thousands of sites and estimate biological age. The omega-3 intervention with life’s®OMEGA 60 showed effects across several epigenetic ageing measures, corresponding to an estimated 2.9 to 3.8 months of biological-age advantage over three years. All three interventions together showed an additive benefit on one clock, PhenoAge.

Vincent explained the underlying biology with an analogy of his own. DNA gets read and copied every time a cell divides, and methylation marks attach to it. “Basically, it is the Post-its you put in a book,” he said, marking where to start and stop reading. The trial changed a set of methylation-based estimates of biological age. It did not show DNA being copied more accurately, and as Vincent stated, nothing here changes anyone’s DNA.

More than 90% of innovation starts with a problem worth solving

With all the emerging science and consumer trends, how does dsm-firmenich decide what to work on?

For more than 90% of projects, the first question is whether there is a real pain in the market worth solving, felt by a consumer or a patient. Establish that, and the technology to solve it can be found afterwards.

That ordering is key because the input is effectively unlimited. Ideas are endless, in Vincent’s view, and not confined to his company. Eight billion people, eight billion brains, and now an unlimited amount of compute augmenting them. With ideas that abundant, the hard part is picking between them.

His answer is a stage-gate funnel, where an idea has to clear a defined checkpoint before it gets more resources. He keeps the funnel deliberately loose at the front. That early stage, the fuzzy front end, can stay unstructured, because that is where large and messy sets of ideas live. Each gate downstream demands more concreteness, until an idea has to justify real investment.

The remaining 10% or so is technology push, and he does not take it lightly. Ask people thirty years ago whether they would buy a mobile phone, and they would say no. Some things consumers cannot ask for.

Those bets get held to a different test, one a mentor gave him. “You should smell the money from the start,” he said. If an idea leaves you thinking there are a thousand hurdles and a billion-dollar prize behind them, it goes into a small protected box outside the normal business case. Everything else has to earn its way through the gates.

The next frontier is the format

So where does the science go next? Vincent gave two answers.

The microbiome comes first, and it has been his answer for a while. He still sees it as the place where the most health factors can be modulated.

His second answer is earlier-stage. He is interested in whether flavour perception itself can drive biological effects, particularly around relaxation and the gut-brain axis, using chai and other smooth flavours as the intuitive example. The brain is a part of the body the nutrition industry rarely designs for.

Ask him what is underrated, though, and his answer is format. The assumption that a supplement means a pill, a softgel or a gummy is rapidly changing, and functional foods have come back with it. He pointed to the growth of meal replacements as evidence that the delivery question has reopened.

He sees the same shift in supermarkets, which he visits wherever he travels. The aisle keeps expanding, and it also keeps moving. “It’s not a dietary supplements aisle anymore,” he said.

That shift matters for the hallmarks he builds against. Fibres and lipids take up real space in a product, unlike micronutrients that need only milligrams. A format with room for grams is a format that can carry doses a capsule never could.

A bigger format also has more that can go wrong. It has to taste good, remain palatable over its shelf life, manage moisture and stay safe at the point of consumption. His teams work on those challenges daily, with large brand owners and startups alike.

For a nutrient supplier, that could be where the next decade of work is. Plenty of the industry is still looking for the next active. Vincent argues that for a lot of promising nutrients, the harder problem is downstream of discovery. Getting an effective dose of a credible ingredient into someone, repeatedly, for years, in a format they keep choosing.

Want to connect with Vincent?

Vincent is always interested in connecting with people working on the next generation of nutrition and health innovation, from new science and ingredients to formulation, delivery formats and consumer adoption.

If you are developing a healthy longevity concept, tackling a formulation challenge, or exploring how promising science can be translated into products people will actually use, connect with Vincent on LinkedIn or reach him directly at vincent.van-buul@dsm-firmenich.com

I’m Eshan. An operator-turned investor, backing companies in food, agri, nutrition, and health. I’ve been writing Better Bioeconomy since 2023, as a way to share my learnings and to connect with cool people like you. Thanks for reading!

If you found value in this newsletter, consider sharing it with a friend who might like it.

Or, if someone shared this with you, consider subscribing (it’s free!).

Subscribe now

Unlock Asia’s trillion-dollar opportunity

Don’t miss your chance to attend the Asia-Pacific Agri-Food Innovation Summit and get 10% OFF with my network code. Use my exclusive partner code for 10% OFF your pass: BETTERBIO10. See you there!

Learn more

Disclaimer: The views and opinions expressed in this newsletter are my own and do not reflect those of my employer, affiliates, or any organisations I am associated with.

Last week, I sat down with Jake Berber, Co-founder and CEO of Prefer.

Disclosure before we start: Prefer is a portfolio company of Better Bite Ventures, where, at the time of publishing, I work. Nothing here comes from that relationship (other than the easy WhatsApp access). Everything below is what Jake told me, plus public reporting.

Arabica broke US$4 a pound for the first time in February 2025, after drought and heat cut supply out of Brazil. It spent the rest of the year setting fresh records. Cocoa went through a bigger shock, running to around US$12,900 a tonne in December 2024 as disease and bad weather squeezed West African production.

For the companies that buy those crops, a commodity problem becomes a P&L problem. Pay more, raise the shelf price, absorb the margin hit, or find a way to use less of it.

Prefer helps them do the last one while maintaining high quality. The Singapore company makes coffee and cocoa flavours by fermenting low-cost inputs, with no coffee or cacao in the supply chain. The output goes to large food manufacturers as an extender. Blended into conventional coffee and chocolate, their customers - coffee and cocoa brands and manufacturers - use less of the commodity without changing what the consumer tastes.

Prefer was founded in November 2022 out of Entrepreneur First’s Singapore cohort by Jake and Ding Jie (DJ) Tan. DJ is a former A*STAR scientist who moved from organic chemistry into fermentation. Prefer has raised US$7 million, most recently an oversubscribed US$4.2 million pre-Series A in August 2025.

Before starting the company, Jake studied neuroscience, worked in biotech, started and failed two startups, then took a venture analyst role while getting his MBA.

In our chat, Jake shared:

• Why coffee and cocoa came first

• Why the product moved from replacement to extension

• What a manufacturer is really paying for

• Why a B2B company runs a consumer brand

• What it takes to close deals with big food companies

• How Prefer plans to scale without owning the factories

Grab your coffee and let’s jump in!

Coffee and cocoa are supply problems before they are flavour problems

Prefer’s mission is to recreate the flavour and functional properties of ingredients through fermentation. Why did they start with coffee and cocoa?

“Coffee and cocoa share the same problem. Supply chains are volatile, and prices are rising,” Jake said. “They both grow on a weather-sensitive band of land near the equator, and demand keeps going up. Prices have reached all-time highs, and consumers are sick of the hikes.”

Fermentation was the tool that made it addressable. Coffee went first on commodity prices and market size. Cocoa was pulled in by the customer base. “Our customers who buy coffee often have chocolate too, and they asked if we could make chocolate flavours as well,” Jake said.

Making coffee without coffee means working backwards from the roast

A green coffee bean does not smell like a cup of coffee. The aromas we recognise appear when heat hits the compounds inside the bean. Sugars and amino acids react with one another and throw off hundreds of new browned, roasted flavour compounds. That is the Maillard reaction, and a roaster spends a career learning to control it.

Prefer works that sequence backwards. Rather than growing a plant to produce those precursor compounds, it builds them with microbes, then roasts the result so the same chemistry runs. The company describes it in three steps:

• Ferment food byproducts into flavour precursors

• Roast the precursors to induce Maillard compounds

• Process the material into whatever format a manufacturer needs

The technical half of that came from DJ, Prefer’s co-founder and CTO. Prefer’s first six months of R&D happened in his kitchen.

The feedstocks have changed as the process matured. The early Singapore batches ran on local manufacturing byproducts, including okara from tofu production, day-old bread and spent barley grain from breweries. The current PreferRoast formulation runs on broken rice and chickpeas, chosen for cost, allergens, consistency and industrial supply.

Fermentation is also what makes the output adjustable in ways a farmer cannot match. Prefer can change the microbes, the substrate, the fermentation conditions and the roast profile to steer where the flavour lands. The base product carries no caffeine, so caffeine gets added separately, which is why the range runs from decaf to double strength.

Two formats are in market. PreferRoast, launched in 2024, is the ground product. A soluble powder followed, which goes into instant sachets, ready-to-drink cans and concentrates. PreferChoc is the cocoa line off the same platform, and manufacturers have been testing it at 30% to 50% inclusion in place of regular cocoa powder.

Prefer stopped trying to replace coffee, and set out to complement it instead

The company that first reached consumers was selling bean-free coffee, which sat on the shelf as an alternative to conventional coffee. The company selling today is offering a coffee extender, designed to be added to conventional coffee at up to 40% without a change to the flavour profile.

The repositioning came from the market. Nobody was shopping for an alternative. As Jake told The Spoon, “People wake up in the morning and say, ‘Dang, my coffee is a dollar more expensive today than it was a month ago.’” They want the cup they already like at the price it used to be.

“Customers asked for it; it solves the consumer problem of rising prices. They want the same product, but at the price they used to be able to afford. So by using our ingredients, coffee and chocolate companies are creating products that taste the same and cost less,” Jake said.

A full replacement asks a consumer to give up something they like for something unfamiliar. An extender lets the brand keep selling the same experience and change what sits underneath it. Green Queen has described the move as the blended-meat playbook applied to a commodity, and Jake uses the comparison himself. “We want coffee and chocolate to last, not end,” he told PEAK.

Prefer changed its own products too. The Iced Oat Latte was its debut consumer launch and a full replacement. It now uses PreferRoast as an extender blended with conventional coffee.

The company also changed which argument leads the sale. Its early pitch leaned on the climate case. “We worked out early that sustainability on its own wasn’t going to get a big food company to buy,” Jake said. “Taste and cost are what matter most, so that’s what we focused on.”

Prefer still quotes carbon, with a life cycle assessment putting PreferRoast at 8 times lower carbon intensity than grown coffee. But that number does not do the selling. The cost and consistency advantage opens the door, and the emissions figure becomes a supporting fact once the buyer is already in the room.

Price certainty is the product

So what is a large food manufacturer paying Prefer for? Cost and supply chain stability.

A coffee buyer sits between the commodity market and the consumer. Jake described the position as caught between a rock and a hard place, between what they pay for raw material and what consumers will pay. Absorb the increase for a while, then pass it through, reformulate, shrink the pack, or take the margin hit.

“We run a factory, not a harvest, so we can quote a price and sign long-term contracts on it,” he said.

A harvest is exposed to rainfall, disease, growing cycles and a global trading market, and it sets its own price and tells you afterwards. Sure, a fermentation process still carries input and energy costs, but far more of them can be engineered, forecast and written into a contract.

Prefer’s soluble powder comes in at a lower price than soluble coffee today. For a procurement team, the more valuable part of that is knowing the number will still be there next year.

Price certainty is the product. It is also why the applications skew to formats where the ingredient blends into a formulation such as instant coffee, canned lattes, chocolate drinks, brownies and confectionery.

The consumer brand is a sales tool for the ingredient business

Prefer has built a real consumer footprint at home. “For our range of RTD (ready-to-drink) cans, we’re now in 250+ locations across vending, retail and corporate pantries, all in Singapore,” Jake shared this month. More recently, the cans landed in all 56 Shell forecourt shops and 90 Sheng Siong stores in Singapore.

While that may look like a company that started at the consumer and worked its way upstream, the direction never changed. “We started the business as a B2B company, and we’re still a B2B company,” Jake said. “It’s just taken a few steps and a few years to get where we are today, but the core assumptions for why we started Prefer remain.”

The consumer line is part of how they sell B2B. When Prefer is in a room with an FMCG buyer, the question on the table is whether consumers will buy this. “We can answer that with sales instead of a tasting panel,” Jake said, “which seems to be more effective at taking the risk out for them.”

A panel score says people liked the flavour, but a sell-through rate says they paid for it. Prefer’s canned coffee sales grew tenfold between February and July 2026, and the Shell listing followed that data.

Supply is the other reason. Prefer is sold out of its coffee production every month, so where each kilo goes is a real decision. A can of ready-to-drink uses a small amount of the ingredient compared with a bag of ground coffee, and it still puts the product in front of a consumer. Consumer formats buy more distribution per kilo than the same material sold any other way.

How a big manufacturer decides to adopt a new ingredient

Prefer is now working with larger food companies. What does it take to get a new ingredient adopted inside one?

“One thing that has been surprising is how much of the work they do happens in a black box,” Jake said. Once you win their interest and send application recipes and samples, the work moves to their R&D team to formulate, and the visibility drops away.

The team shows up in person and guides the customer’s R&D with its own suggestions, as far as the customer will allow. The timelines stay long, and the decision stays collective, which in Jake’s view is rational given what a formulation change costs a brand at scale.

“Volume only comes once you get the nod from R&D, QA, procurement and marketing,” he said. Each of the four is solving a different problem. R&D needs the ingredient to work in the formulation. QA needs it to behave the same way every batch. Procurement needs the economics and the supply to hold. Marketing needs to believe the finished product will sell. An enthusiastic innovation team is a long way from a repeat order.

And Prefer seems to have strong demand from those companies. “To date we’ve signed US$26 million in offtake MOUs (Memorandum of Understanding), and the focus now is turning those into recurring orders,” Jake said. The ingredient is already selling into Singapore, Japan and the Philippines, with Thailand next.

In October last year, Ajinomoto launched GRe:en Drop Coffee in Singapore under its Atlr.72 brand, a dairy-free iced latte built on 70% conventional coffee and 30% PreferRoast. More should follow as Prefer works the MOUs toward recurring orders.

The scaling up plan must match what the product promises

Every fermentation company reaches the point where the process has to leave the lab and run at commercial volume.

Prefer is at that point now. “Our pilot plant with 500 tonnes a year of annual capacity for PreferRoast and PreferChoc is commissioning this quarter,” Jake said. The company produces about 10 tonnes a year today, so the new plant is a 50-fold step up, and its first major equipment shipment landed in August.

Jake wants Prefer to own very little of the eventual manufacturing footprint. “In this space we pay per batch and own the process and some of the equipment,” he said. “That’s how we plan to scale, to keep CAPEX low and the potential for diversified manufacturing high.”

Prefer keeps the microbes, the process parameters and the IP. Somebody else keeps the building. For a company whose goal is a lower and more predictable cost per kilo, a capital-heavy factory network would work against the thing it sells.

Building for the mass market makes Asia-Pacific the obvious place to build

Several companies are rethinking coffee through technology, and most of them are building outside Asia, in the US and Europe.

Jake thinks Prefer is playing its own game. “In coffee, most of the others are building for a different customer,” he said. “We’re building for the mass market, with some of the biggest food companies in the world as our customers.”

“Our job is to make the lowest cost per kilo at the highest quality possible.” That single choice explains the extender model, the soluble format, and why cost leads the pitch while sustainability follows behind it.

It also explains the geography. “For manufacturing, APAC helps because CAPEX and OPEX here are lower, and when cost is one of two main value propositions, that’s super important,” Jake said.

Demand fits the same shape. “Asia has famously price-sensitive consumers, which is exactly who we’re building for,” he said, “and a lot of the coffee here is instant and ready-to-drink rather than specialty pour-over.”

Those are the formats a soluble extender was built for, which is why the early commercial work has concentrated in the region, from Ajinomoto in Thailand to 250 locations in Singapore.

The region matches the business model the company picked. If you sell premium coffee at premium margins, manufacturing cost matters differently. If you sell a mass-market brand, a way to take cents out of every serving, the cost of making each kilo is the competitive position.

That is what the next raise is for. Jake has said it would let Prefer “increase production another 10x, bring costs down further to reach profitability, and service customers globally.”

Want to connect with Jake?

Jake is looking for introductions to coffee and cocoa FMCGs, co-packers, flavour houses and private-label manufacturers. If that is you, or you know someone there, he would like to hear from you. You can reach him at jake@prefer.bio.

I’m Eshan. An operator-turned investor, backing companies in food, agri, nutrition, and health. I’ve been writing Better Bioeconomy since 2023, as a way to share my learnings and to connect with cool people like you. Thanks for reading!

If you found value in this newsletter, consider sharing it with a friend who might like it.

Or, if someone shared this to you, consider subscribing (it’s free!).

Subscribe now

Made in APAC 🌏

This is part of a series of conversations with founders and operators across the APAC region doing industry-leading work that shapes how we produce food, nourish ourselves, and improve our health.

Check out my previous conversations:

• How a $2B Agtech Unicorn Builds Product - Halter

• How Physical AI Agents Automate Harvest Timing and Pollination - Polybee

Stay tuned. More to come!

Unlock Asia’s trillion-dollar opportunity

Don’t miss your chance to attend the Asia-Pacific Agri-Food Innovation Summit and get 10% OFF with my network code. Use my exclusive partner code for 10% OFF your pass: BETTERBIO10. See you there!

Learn more

If you found value in this newsletter, consider sharing it with a friend who might benefit from it!

Share

Or, if someone forwarded this to you, consider subscribing.

Subscribe now

Disclaimer: The views and opinions expressed in this newsletter are my own and do not reflect those of my employer, affiliates, or any organisations I am associated with.

I have been increasingly curious about what the rise of metabolic health means for food and nutrition companies. GLP-1s have pulled a lot of attention toward blood sugar, appetite, weight management and muscle health.

But I wanted to work out where the food and nutrition opportunity extends beyond the drugs themselves. Tate & Lyle published one this month called Future-Proofing Nutrition: Designing Products for Lifelong Metabolic Wellness, and it is the latest in a run of consumer research on the subject since the start of 2025. ADM, Kerry, Cargill, dsm-firmenich, Ingredion, Herbalife and OPTAVIA have all published in the same window.

So I read across them to see where their thinking overlaps. Four themes kept coming up that I think say something useful about where metabolic health nutrition is heading.

Every source here is a company with something to sell, and these documents are marketing as much as they are research. I will just say that up front and move on, because the useful part is watching where companies with different products end up agreeing.

Consumers want metabolic health outcomes without asking for a “metabolic health” product

Metabolic health is a cluster of readings. The international criteria agreed in 2009 by the International Diabetes Federation, the American Heart Association and the World Heart Federation, among others, cover five of them:

• Waist circumference

• Blood pressure

• Fasting glucose

• Triglycerides

• HDL cholesterol

Three out of range is metabolic syndrome. Being clear on all five without medication is what a 2019 analysis of US survey data called optimal metabolic health. Clinicians and researchers have those definitions. But shoppers do not, and several of these reports say so directly.

Emma Cahill, global marketing director for sweeteners, fibres and GLP-1 at Tate & Lyle, told: “I have yet to see a brand position boldly around metabolic health in consumer-friendly language.” OPTAVIA’s December 2025 survey puts a number on it from the other side. Fewer than half of Americans, 43%, have heard the term at all.

The behaviour is running well ahead of the vocabulary. Cargill’s TrendTracker research found:

• 76% actively trying to avoid sugar

• 71% actively trying to increase protein

• 35% choosing foods specifically to support digestion and gut health

Those overlap with what metabolic health products are trying to influence. None of them requires the shopper to have heard the phrase.

Herbalife’s May 2026 survey of >10,000 people across 11 Asia Pacific markets asked what people most want to improve. The top answers were:

• Sleep quality: 44%

• Mental wellbeing and stress: 38%

• Nutrition quality: 31%

• Weight management: 30%

Metabolic health was not among the options, so the survey cannot tell us whether people would use the phrase. What it does show is that the priorities people pick out are familiar: sleep, stress, nutrition and weight.

So the problem these reports keep circling is a naming problem sitting on top of behaviours that are already there. That is a better commercial position than the reverse, and it explains why these documents are all partly language exercises.

Cahill’s practical answer is that shoppers have flipped the order in which they read a label, checking the positives like fibre and protein before they check the negatives like sugar and calories. That is a small observation with a lot of formulation consequences.

The opportunity may not be to convince consumers to care about “metabolic health.” They already care about the things underneath it. For food companies, the better use of the term may be behind the scenes, as a way to decide what to formulate for, while the pack talks about outcomes people already understand.

The plate got smaller, and that changed the recipe and the format

For years, nutrient density was a positioning claim. You could put it on a pack. GLP-1 medications turned it into a constraint. ADM found that 74% of medication users feel full more quickly, and once that happens, every gram has to carry more.

Tate & Lyle’s earlier study, fielded among 500 active and former GLP-1 users in North America in May 2025, found users reading labels harder than before, looking for high protein, added fibre and less sugar.

Cargill’s numbers show the same behaviour as a coping strategy. Among GLP-1 users experiencing side effects:

• 25% increased fibre

• 24% increased protein

• 21% moved to smaller and more frequent meals

Nestlé is spending against that shift at scale, with Vital Pursuit built around portion-aligned meals and a Nestlé Health Science platform built around preserving lean muscle mass, gut health, micronutrient intake, hydration and weight rebound control.

On the ingredient side, Arla Foods Ingredients is selling into lean muscle preservation and gastrointestinal tolerance.

The most common answer is protein and fibre. The question is how much of that convergence comes from the biology, and how much comes from what these companies already sell.

Some of it is physiology. Less food means protein matters more for holding lean mass, and fibre matters more for satiety and for the glycaemic response of whatever is left on the plate. But a dairy protein supplier and a chicory fibre supplier arriving at compatible conclusions from different starting points is worth something.

The rest of it is that Cargill, Tate & Lyle, Arla, FrieslandCampina, Beneo and Roquette all sell protein or fibre, and Ingredion’s metabolic health paper names fibre, protein and sugar reduction as the three formulation tools, which is a description of Ingredion. ADM’s report is the clearest case, because it closes with a page mapping each finding onto an ADM ingredient:

• Soy, pea and wheat protein for muscle

• Biotics for digestive discomfort

• Prebiotic fibre for satiety

Every one of these companies has found the white space, and all of them have found it shaped like the ingredient they already sell. Both things are true at the same time, and the convergence worth trusting is the one that crosses portfolios.

The language slips in the same way. “Slows gastric emptying” and “triggers GLP-1 release” are mechanistic claims. “Supports metabolic health” implies a clinical one. Across the reports I read, I did not find a clinical endpoint, and the documents move between those two registers without marking the change.

The smaller plate is changing the format as well as the recipe, and that gets less attention than the ingredients do. ADM’s own conclusion is that what these consumers want is enjoyable, convenient and portion-controlled functional products.

For food companies, the opportunity is bigger than reformulation. If appetite shrinks, what has to change is the portion, the format and the occasion, as much as the protein and the fibre content.

76% of people who stopped taking GLP-1s regained some of the weight

ADM published Where Are They Now? in November 2025, surveying people across the US, UK, France, Germany and Spain who are on anti-obesity medication and people who have stopped. It is the one report here that measures what happened to people after they came off.

Among former users, 60% say food noise has come back. 76% have regained at least some of the weight. Many also report keeping the habits they picked up on treatment:

• 54% still eating more fruit and vegetables

• 49% still holding better portion control

• 47% still cooking more at home

ADM reads the regain as happening despite those habits. The survey reports both and does not show they are the same people. It is also self-reported, and it counts how many people regained something. It does not say how much.

A Cleveland Clinic analysis followed 7,938 patients who stopped semaglutide or tirzepatide. Those treated for obesity had lost 8.4% of their body weight before stopping, and regained 0.5% a year later. That is well under what randomised trials had shown, largely because a quarter switched to another drug and a fifth restarted the original.

Both findings can be true at the same time. And what limited the regain in the Cleveland Clinic cohort was mostly other prescriptions.

Intention data tells a messier story. Kerry’s survey of more than 2,600 US adults, published in February 2025 on earlier fieldwork, found 6% of users planning to stay on the medication for life.

ADM asked a similar question later, across five countries, and found 35% planning to continue long-term. The two questions are not identical, so this is not a flat contradiction. But 6% and 35% point in opposite directions, and both are asking people to forecast their own future behaviour. I would be cautious about building a market thesis on either number.

What people actually did is more useful than what they predicted they would do. Among those who do come off, appetite returns and weight regain is common.

Tate & Lyle describes the same arc as three phases: momentum, maintenance and regression. It also singles out former users, who face a resurgence of appetite and food noise. Kerry adds one figure that has aged well. 90% of users were already buying vitamins, supplements or probiotics while on the drug.

So a good part of the opportunity these companies describe lies in the prescription, not in the drug. Food can sit in three places:

• During treatment, in tolerability and nutrition support

• After it, in appetite and weight maintenance

• Outside it, for people managing the same physiology without ever taking a drug

None of these reports frames it that way, but read together, that is the bet they are making.

Food is unlikely to replicate the pharmacological effect of the drugs head-on, and the pharmacokinetics explain why. Endogenous GLP-1 has a half-life of one to two minutes before DPP-4 degrades it, and the long-acting analogs reach days, several of them by binding to albumin.

So the realistic opportunity sits next to the drugs, in products that reduce side effects, improve tolerability and support adherence.

And then there is everyone else. Cargill’s sugar-avoidance number has no drug gate on it at all. People managing weight or blood sugar without a prescription sit outside almost all of the segmentation work in these reports.

The drug creates an obvious consumer segment, but the underlying metabolic needs extend much further.

Weight management is the entry point, and healthy ageing is the broader opportunity

The Tate & Lyle report is called Designing Products for Lifelong Metabolic Wellness. The frame is in the title.

dsm-firmenich’s 2025 Global Health Concerns study with Ipsos shows the same absorption from the consumer side. Among the people who are both concerned about healthy ageing and already doing something about it, 46% are targeting metabolic health as part of that. Kerry’s Health and Nutrition Institute puts food for health and longevity at the top of its 2026 trends.

The clearest overlap is muscle:

• ADM lists soy, pea and wheat protein for maintaining muscle mass

• Arla is selling lean muscle preservation

• Nestlé’s GLP-1 platform leads on preserving lean muscle mass

Each of those is a response to a drug side effect. Protein for muscle is also the standard answer to age-related muscle loss. So a company formulating for someone on a GLP-1 and a company formulating for a seventy-year-old are reaching for the same ingredient to solve closely related problems, and calling it two different things.

That suggests more overlap between the two than the category labels imply. The same ingredient platform may be able to serve both, even where the format, the evidence and the positioning have to differ.

When I looked at the healthspan shift a few weeks ago, the category turned out to contain a lot of separate things. Healthspan is “mobility, sleep, cognition, metabolic control, the muscle you keep.” Two of those five, metabolic control and muscle, are what these reports already sell into.

So the widening is not a leap for anyone. A company with a protein for lean-mass preservation and a fibre for glycaemic response is already holding two components of healthspan. Entering healthy ageing may cost it less in new technology than in a different evidence package, a different format and a different reason for the shopper to buy.

For food and nutrition companies, that makes weight management an entry point but not the end market. The larger opportunity is helping people hold onto metabolic function for longer.

What this means if you are building or investing in this category

Build for the physiology that outlasts the prescription

The durable products are unlikely to be the ones whose whole identity is “for GLP-1 users.” Appetite, blood sugar, lean mass, digestive tolerance and nutrient density all matter before someone starts on a drug, while they are on it, and after they stop.

That broadens the need beyond current users to people considering medication, people who have stopped, and people managing the same problems without a prescription.

A product built around those needs has a chance to stay relevant across more of that journey. One built around the medication is relevant for as long as someone stays on it.

Sell the outcome in the words people already use

Two vocabularies, doing different jobs. Metabolic health can be the internal one, for deciding what to formulate for and which outcomes are worth testing. The one on the pack has to survive a shopper reading it in four seconds.

So the naming problem belongs to the industry. The shopper does not have it. The brand that wins here may be the one that keeps the phrase off the pack and sells steady energy, or fewer afternoon crashes, or holding onto muscle into your sixties.

Spend on outcomes, because the mechanism is shared

If the same handful of ingredients turns up in every report, the ingredient stops being what separates one company from another. Once several competitors can tell a plausible mechanistic story, the mechanism becomes a weaker differentiator.

What none of these reports tells you is what happened to the people who ate the finished product over time. That evidence is expensive to produce, and a competitor cannot get it by reformulating around the same ingredient.

Separate exposure to GLP-1 adoption from exposure to metabolic health

These look like one bet in a pitch deck, but they are two. A company whose growth needs more prescriptions written is underwriting a drug cycle, a reimbursement decision and a discontinuation rate, none of which it controls.

A company solving satiety, lean mass, glycaemic response or post-medication maintenance is underwriting a physiological need that exists either way. The test is what happens to demand if prescriptions stop growing tomorrow. If the need is still there, it was never a drug bet.

GLP-1 accelerated this category and made the opportunity much easier to see. On the evidence of these reports, it is not going to define its boundaries.

I’m Eshan. An operator-turned investor, backing companies in food, agri, nutrition, and health. I’ve been writing Better Bioeconomy since 2023, as a way to share my learnings and to connect with cool people like you. Thanks for reading!

If you found value in this newsletter, consider sharing it with a friend who might like it.

Or, if someone shared this to you, consider subscribing (it’s free!).

Subscribe now

Unlock Asia’s trillion-dollar opportunity

Don’t miss your chance to attend the Asia-Pacific Agri-Food Innovation Summit and get 10% OFF with my network code. Use my exclusive partner code for 10% OFF your pass: BETTERBIO10. See you there!

Learn more

Disclaimer: The views and opinions expressed in this newsletter are my own and do not reflect those of my employer, affiliates, or any organisations I am associated with.

A phone can now show you last night’s sleep stages, this morning’s readiness score, your glucose curve after lunch, and a panel of blood markers from a commercial lab. Every one of those readings is real, but they are not equally well understood. Still, they show up on the same screen looking as though they are.

Whether we can measure more of our biology is clear. But what any one of those numbers is telling you is less clear, because a single reading can be answering three different questions at once:

• What happened? Your glucose rose after lunch.

• What does it mean for me? Whether that rise is unusual for you, or sits inside your own normal range.

• What should I do? Change the meal, change the timing, or do nothing.

My reading is that consumer health is moving fastest into the question it can answer least well. Measuring what happened is getting easier. Knowing what one reading means for one person is harder. Knowing what to do about it is where these products are heading, and where the evidence runs out first.

Oura, WHOOP, Google and Dexcom are all adding data they did not collect

The leading consumer health platforms all shipped the same feature this year: an import button.

Oura sells Health Panels with Quest Diagnostics: book a blood test in the ring’s app and fifty results come back into it, and you can now upload a PDF of a test taken anywhere else.

WHOOP made the same move earlier, also with Quest, with 65 biomarkers reviewed by a licensed clinician and free uploads of old bloodwork.

Google went further and took the medical record itself: its Fitbit health coach connects lab results, medications, visit history and records from your own doctors, and its worked example is a food question, asking how that slice of pizza affected your glucose.

Dexcom has gone the other way, adding interpretation to a single signal, with a redesigned Stelo app doing pattern recognition, AI coaching and photo meal logging.

Two competing wearable companies went to the same laboratory business, and the headline feature on both accepts data the device never measured. So the wearable is turning into the place you keep everything else as well.

Thirty people ate the same meal twice and got two different readings

Start with the first question: What happened after this meal? Glucose is the fair test here, because it is the signal with the most evidence behind it.

At the US National Institute of Diabetes and Digestive and Kidney Diseases, Kevin Hall’s laboratory ran two inpatient feeding studies in which thirty adults without diabetes were served the same meals twice, about a week apart. The results came out in the American Journal of Clinical Nutrition in January 2025.

A research kitchen prepared and presented every plate. Continuous glucose monitors captured 1,189 responses to those duplicate meals, using sensors from both Abbott and Dexcom.

The question was how closely one person’s two responses to an identical meal match. Not very closely, apparently. Researchers score this on a scale from 0 to 1, where 1 means a measurement repeats itself perfectly. Hall’s team got 0.28 for the Abbott sensor and 0.17 for the Dexcom.

Eating the same breakfast twice generated about as much variation as eating two different breakfasts. They concluded that dietary advice based on these readings needs many measurements before a person’s response can be estimated reliably.

And this was inpatient, with every plate weighed and served by a research kitchen, which is about as controlled as this gets. In our day-to-day life, more things are moving that could change a single meal’s result.

Hall struggled to reproduce the response to a single meal. That is also what consumer glucose products put on the screen. Stelo shows the curve after a meal and now offers to photograph the plate first. Oura’s Meals feature does the same. The sensor is accurate, and the reading is real. But it is describing one meal, and one meal moves around a lot.

That said, it is thirty people, and the variability is not all biology. The two sensors did not agree with each other either, at 0.28 and 0.17 on the same meals. One post-meal response is not repeatable enough to base personal dietary advice on, which is what the study set out to test.

Averaging many meals finds the signal that one meal loses

The second question is the one people want answered: What does this mean for me?

Later that year, the same journal published a study that gets closer to answering it. At Westlake University in Hangzhou, a team working with José Ordovás and Ju-Sheng Zheng ran a study of 176 healthy Chinese adults, treating each person as their own experiment and feeding them standard meals under a glucose monitor.

Individual meals still varied. But when they combined many meals into a single score for how strongly a person's blood sugar responds to food in general, that score held steady. It came out at 0.73 on the same reliability measure that gave the single meal 0.17, and it was still recognisably the same person two years later. It is one cohort of healthy Chinese adults eating standardised meals, so whether the same score holds in other populations, or outside a research kitchen, is not established.

Hall asked whether one meal predicts itself. Westlake asked whether a person has a general tendency that shows up across many meals. The second team did what the first said was needed, and found something that held for two years. But the app puts the individual meal in front of you, because that is the thing that happened today.

That choice comes up everywhere, not only with food. A single meal, a single night’s sleep and a person’s average over months are different kinds of information. Some signals hold up one night or one meal at a time. Others only mean something once they have been averaged.

Glucose monitoring worked best where a diagnosis, a target and a treatment already existed

That leaves the third question: Does acting on the information help?

In June, Dexcom presented the CONNECT trial at the American Diabetes Association’s annual meeting. It randomised 283 adults with type 2 diabetes who were not on insulin at 22 US primary care practices, and ran for 26 weeks against routine care with finger-prick testing.

Everyone started with an HbA1c of about 8.8%. HbA1c is a blood test that reflects average blood sugar over roughly the previous three months, and it is one of the main measures clinicians use to judge how well diabetes is controlled.

The monitored group’s HbA1c fell 1.6 points against 0.7 in the control group. This is Dexcom’s own trial. The company sponsored it, several authors report consulting or research support from it, and the results have been presented at a conference but not yet published as a peer-reviewed paper. None of that undoes a randomised trial that hit the outcome it set out to test.

HbA1c at 26 weeks is itself an average of about three months of blood sugar, so what CONNECT tested was an aggregate outcome in a diagnosed population with an established target. It never rested on any single reading. Hall’s duplicate meals said one reading is not stable enough to rest on, and CONNECT did not ask it to.

Two months later, JAMA Internal Medicine published a review concluding that for people without diabetes, there is no good evidence that continuous glucose monitoring improves health or prevents diabetes.

The authors add that the meaning of glucose fluctuations in that group is uncertain, and that chasing spikes may lead people to cut out healthy foods such as fresh fruit.

What I think separates them is that in diagnosed type 2 diabetes, doctors already knew what was wrong, which number they were trying to bring down, and what to do about it, long before any sensor arrived.

A healthy person watching a curve rise after lunch has none of that, and no treatment that has been shown to help with the rise itself. The same pattern shows up elsewhere: wearable signals shifted up to seven weeks before an inflammatory bowel disease flare in 309 people, again in a group that already had a diagnosis.

The companies mark this boundary themselves. Dexcom’s Stelo site sells an AI Coach and promises to show how food, exercise, stress and sleep affect your glucose, and states that the user is not intended to take medical action on the device output without consulting a clinician. Those disclaimers draw a line between an insight a company can generate and a decision it will stand behind.

Products have to turn measurements into something people can act on

A consumer health product does more than show a number.

Stelo draws the curve, attaches it to the meal you photographed, and offers up to three suggestions off the previous day. Oura compresses a night of physiology into a readiness score. WHOOP returns a recovery percentage. An AI coach takes whichever of those is on the screen and writes a recommendation from it.

Each of those steps turns a measurement into something a person can act on. And each one needs a choice about what to show: the meal, the night, the day, or the three-month average.

The industry already knows baselines matter. Oura takes about two weeks to learn your averages and WHOOP about thirty days, and both use that history to judge whether today is unusual for you. That is a baseline for reading today's number, and it is a different thing from knowing how you respond to a particular food.

Glucose products do aggregate, but over time rather than over meals. Stelo reports daily and weekly summaries, and CONNECT's HbA1c is three months of blood sugar averaged. Both of those average the clock. Westlake averaged the meals, which is the score that would answer the food question. What the app shows you is still the meal.

The screen gives both the same weight. A glucose curve after lunch shows up looking as definitive as a three-month average.

The numbers are still useful. A review of 39 systematic reviews found that wrist-worn trackers reliably increase physical activity, even where their effects on cardiometabolic markers stay inconsistent. Prompting attention is a real function. But it is a different function from telling someone what to do.

How the industry could get better at this

Four of these strengthen the path from a measurement to an action. The last one is about what to do where it cannot be strengthened yet.

Run the trial

Elevance Health and the University of California, Irvine, worked with Apple on a twelve-month randomised trial in 901 adults with asthma across 41 states. It ran against usual care, and the result is in JAMA Network Open.

Among those whose asthma was uncontrolled at the start, symptom control improved 4.6 points against 1.8. It is the same shape as CONNECT: a consumer device pointed at a diagnosed group with an established target and a known action.

Pay for the outcome

The US Food and Drug Administration has selected Dexcom’s Glucose Health Program as the first participant in its TEMPO pilot. The pilot is tied to the Centers for Medicare and Medicaid Services’ ACCESS Model, which pays on demonstrated improvement in patient health instead of on services delivered.

WHOOP has entered the same pathway. An institution that pays for outcomes has to say what counts as one, and it will not be a glucose curve after lunch.

Aggregate before you personalise

The Westlake score is stable across two years in a research setting, and a product could compute something like it. Whether anyone would value a number that is dull to look at and updates slowly is less clear.

A product that shows a steady, person-level number instead of today’s fluctuation is telling you which of the two it thinks matters.

Add a human, or the intervention itself

In June, Dexcom agreed to acquire Nutrisense, which pairs continuous glucose data with registered dietitians. Viome went further recently, buying Circulate Health and its plasma exchange clinics, which I wrote about here.

One adds human interpretation. The other adds the ability to act on the measurement inside the same company.

Know where to stop the claim

In January, the FDA reissued its general wellness guidance, and the line it draws is not about the sensor.

Some non-invasive wearables can estimate blood pressure, heart rate or glucose and stay outside device regulation. The measurements have to be validated, and the product cannot use them to guide a specific clinical action. It may tell you to see a doctor. It may not tell you what for. That is what the disclaimers above are made of.

Closing thoughts

Nutrition is where this gets difficult. Food touches glucose, sleep, energy, appetite and mood. We control it directly, and we get feedback within hours. That combination makes it tempting to act on a single reading, even though the evidence does not support firm conclusions from one meal.

What would matter is deciding when not to turn a measurement into an insight. Adding another signal is an engineering problem, and the industry has become good at it. Deciding what to do with each signal is a different problem. Some deserve a score, some deserve an explanation, and some should stay a measurement with no recommendation attached. It asks a product to put the evidence ahead of the interface.

The useful question about a consumer health product is no longer how many readings it carries. It is which readings it is prepared to leave off the screen, and what evidence it has for the ones that stay.

I’m Eshan. An operator-turned investor, backing companies in food, agri, nutrition, and health.

I’ve been writing Better Bioeconomy since 2023, as a way to share my learnings and to connect with cool people like you. Thanks for reading!

Unlock Asia’s trillion-dollar opportunity

Don’t miss your chance to attend the Asia-Pacific Agri-Food Innovation Summit and get 10% OFF with my network code. Use my exclusive partner code for 10% OFF your pass: BETTERBIO10. See you there!

Learn more

If you found value in this newsletter, consider sharing it with a friend who might benefit from it!

Share

Or, if someone forwarded this to you, consider subscribing.

Subscribe now

Disclaimer: The views and opinions expressed in this newsletter are my own and do not reflect those of my employer, affiliates, or any organisations I am associated with.

Last week, I sat down with Siddharth Jadhav, founder and CEO of Polybee.

Polybee is a National University of Singapore (NUS) spin-off that puts small autonomous drones to work inside greenhouses and over open fields. They do two jobs: pollinating crops, and scanning them so the platform behind them can forecast yield and time the harvest.

In February, the company raised a $4.3 million seed round led by Paspalis Capital and elev8.vc, with SEEDS Capital and Blue River Technology founder Jorge Heraud participating. More recently, the company signed a channel partnership with Vegetables by Bayer to take the platform to Australian vegetable growers.

Siddharth started the company in 2019 as a side project while working as an aerial robotics researcher at NUS. He had admission offers for a PhD in the US and turned them down. Polybee now runs drone fleets across seven to eight countries, and its customers include Perfection Fresh, Fruitist, PMFresh and Comfresh.

Back in Issue #140, I mentioned that the ag robotics categories reaching commercial scale share a structure. A narrow decision, a measurable outcome, and a task that repeats across growing conditions. I named harvest timing as one of the next candidates. Polybee is that company, so I wanted to see how it looks from inside it.

In our chat, we spoke about the two things every produce grower is missing, how a drone company earns the right to run across a whole farm, what “physical AI agents” means when the term is being used loosely, and why the same product has a different customer in every market.

Let’s jump in!

Most harvest decision in produce today is made by eyeballing a few plants

A grower walks the field, samples a handful of plants by eye, forms a view of how the crop is coming along, and sets the harvest date on it. Accuracy is not a metric anyone can quote, because there is nothing objective to measure it against.

As Siddharth put it in his AgFunder interview, growers would be the first to admit this is not how they would choose to run the business. Produce makes it worse by being stubbornly non-uniform. Even in a greenhouse, with a controlled environment and advanced genetics, plants develop at visibly different rates.

Polybee attacks both halves of that: the accuracy and the sample size. A fleet of off-the-shelf drones flies pre-programmed routes and scouts every plant rather than a handful. They launch themselves, recharge themselves, and need no user touchpoints at any stage.

What they capture depends on the crop. On a fruiting crop, the system reads the fruit load, how much has ripened, and how fast ripening is tracking against the weather. On a vegetable crop, it reads growth rate and whether the plants are inside the buyer’s quality spec.

Cameras scan at millimetre precision, and the models count every plant, size every unit, and project a growth curve per block with confidence intervals. Polybee puts stand counts above 95% accuracy, measurement error under 3mm, and yield and sizing forecasts above 90% accuracy a week out.

Everything the fleet collects lands in Polybee’s Farm Intelligence Platform, which is where a scan turns into a decision. It runs as three modules:

• Forecaster maps plant establishment, yield and harvest readiness down to the row, for every planting and every block, every week.

• Harvest Optimiser turns that into a weekly cutting schedule, balancing expected yield against packhouse capacity, weather, and what has already been sold.

• Planner feeds the same signal back into production planning through the season rather than only at the start of it.

The loop those three describe is plan, forecast, execute, verify. Scan to insight runs about 12 hours, the forecast horizon is seven to ten days, and since the Bayer partnership, the same scan carries disease detection through near-infrared cameras.

Forecasting is live on spinach, broccoli, blueberries and lettuce, with tomatoes and strawberries in development. Polybee says growers save 30 to 50 hours a crop cycle and lift pack-out, the share of the crop that meets spec and gets sold, by 15%.

Pollination happens either without bees or without hands

Pollination runs on the same airframe and a different trick. The drone uses the downwash off its own propellers to vibrate a flower at the right frequency, roughly what a bumblebee does when it buzzes one. The vibration shakes the pollen loose, and nothing touches the plant.

The drones launch when the micro-climate is right rather than on a fixed clock, treat every row repeatedly, and log a timestamp and pass count for each plant. When the battery runs low, they dock, recharge, and resume where they stopped. Coverage runs one to three drones per hectare. Because no insect moves between crops, there is no disease transmission risk, which matters in a sealed greenhouse. It is live on tomatoes and strawberries, with blueberries and peppers in development.

Where bumblebees are available, Polybee complements them rather than replacing them, picking up the shoulder seasons when hives are hardest to manage. Where they are not, which includes Australia and much of South America, the job is done by hand. People walk the rows with leaf blowers, or whack the plants with sticks and hope. That is the incumbent Polybee is displacing.

“What initially seemed like a pond of opportunity is a sea of opportunity”

In 2018, Siddharth was three years into a research career at NUS and stuck on a decision. He had the PhD offers, and a clear enough picture of academic life to wonder whether he wanted five more years of it. Around the same time, he had started reading about agriculture, and what struck him was how little had changed since industrialisation and the breakthroughs in genetics and breeding.

Then his advisor roped him into NUS Enterprise’s Lean LaunchPad programme, a Singapore run of the Stanford course, wanting a hand with the market research. It carried no obligation. You went through it, you came out with a business plan, and what you did with it was your business.

Singapore in 2018 was talking about vertical farming, because indoors was the only way to grow much of anything on the island. That is the slice Siddharth went at, and he went at it as an engineer. Nobody in the industry had an obvious answer to the pollination problem, so he framed it as a technical question. Could you automate the job with drones?

So he built prototypes, called growers, showed up at events, and kept pulling. The signs were promising, and the adjacent problems kept appearing. “What initially seemed like a pond of opportunity is a sea of opportunity,” he said. That is when he spun the company out.

Visibility and control are the two things produce growers lack

Polybee did not stay indoors for long, and the customers are what moved it. Greenhouse growers came into the picture early, then open-field operations, and each of them kept putting the same question back to him. You are going to fly an autonomous platform around my whole farm with good cameras on it. What else can you tell me?

So Polybee went back and asked those growers what information they wanted and did not have. Across every crop and every growing environment, the answers converged on two things: visibility and control.

Produce farming has been unpredictable for centuries, and his thesis is that the root cause is the missing visibility and control. “We are building physical AI agents that turn unpredictable produce farms into data-driven factories,” is how he puts it.

Forecasting is the visibility product, because a grower who cannot see what is coming cannot plan against it. Pollination is the control product, because it lets a grower act on fruit set rather than hope for it. Both were picked for the same reason. They shift a grower’s numbers soonest, and the ROI is clear.

A working product is not enough. It has to work on the grower’s farm under their circumstances

Polybee now has a repeatable sales motion, built on case studies from early partners who let the company take their results to market. Their work proved three things. That the technology functions, delivers value, and has a real ROI.

“You still want to see it work on your farms under your circumstances,” he said, and he thinks agriculture is right to insist on it. Even a seed from an established breeder gets trialled before anyone buys at scale.

So the motion runs in two stages, and stage one is proof of value. Polybee’s advice to a new customer is to put the product on 20% to 30% of their production area and run it across three or four adjacent production weeks.

In developed markets, a grower is harvesting almost every week of the year, at close to the same volumes. That is what makes three or four consecutive weeks a workable test window. Follow the forecasts, follow the harvest timing, then compare those blocks against the status quo blocks next door.

The target Polybee sets itself is at least a 10% yield uplift, and it is charging against that number. Pricing is a fixed fee per hectare, with no drones to buy, and the promise of a 3x to 5x return inside the same crop cycle.

Commercially, the company reports 10% to 15% higher yields in spinach and broccoli when its harvest timing is followed, and up to 15% from pollination in strawberries and tomatoes. Its baby leaf spinach and broccoli case studies claim a 3x profit improvement.

Siddharth knows how that lands in a sector where input companies fight for single percentage points. Salad and leafy vegetables are extremely sensitive to when they are cut, and their spec moves with the crop stage. A couple of days either way moves yield by that much. That sensitivity is why Polybee started here rather than with easier crops.

The pollination number has a second layer. Those gains concentrate in the shoulder seasons, when bees are hardest to manage, and prices are at their highest. A 15% uplift on the most valuable fruit of the year is worth more than the percentage suggests.

Stage two is the whole farm, and that is where the money is. A grower moving from one kilogram per square metre to 1.05 or 1.1 is not just selling more product. They are spreading the same fixed costs over more of it. “That’s where my biggest margin uplift will come from,” he shared.

“Physical AI agents” means the job gets done end to end with nobody lifting a finger

Everyone is calling themselves an AI company right now, and the term has stretched far enough to mean very little. So what does Polybee mean when it says it builds “physical AI agents”?

Siddharth is aware the vocabulary is being used loosely. Polybee uses AI to get jobs done on the farm end to end. Right now, that is two jobs, pollination and forecasting, and the list grows as the company goes deeper into an account.

What that looks like in practice is fleets in seven to eight countries flying dozens of drones a day. Flights are pre-programmed, the drones recharge themselves, and they synchronise their own data. The only time a person walks into the field is to set the drone up and configure it the first time. After that, the system runs itself, and Siddharth describes what it becomes once it does: the central nervous system of the farming operation.

Farms are among the most unforgiving environments to automate, so the industry gets some grace for how hard scaling has been. But what he does not excuse is the design habit underneath it. Too many products have asked farmers to reorganise around the product.

That trade is fine for a trial and fatal for a rollout. “If you want to really scale a product, you have to fit into a farm’s operations and not the other way around,” he said. Some of the failure was immature technology. In 2026, he does not think that excuse holds.

The moat is the stack that keeps the fleet flying, and the data

Polybee combines high technical defensibility with immediate bankable ROI, as one of its investors described it, and that is a strong pair to hold in ag tech. Where does the company’s durable advantage come from? It starts with the unglamorous half of it.

Building this took autonomous navigation software that can orchestrate fleets. It also took data pipelines with the throughput to process what those fleets collect, reliably and at an economics that works for Polybee. Growers get results back within about a few hours, and that turnaround is itself an engineering achievement.

Three more moats sit on top of that stack. First, Polybee is close to patent grants across several regions on its pollination method. It uses turbulent airflow from the drone propellers to self-pollinate crops like strawberries and tomatoes, with blueberries coming.

Second, he believes Polybee is among the top few companies globally for produce crop data. That trove was built over years across a wide spread of environments, crops and varieties, which is what makes the models transfer.

Third, the partnerships. Polybee works with DJI Enterprise for priority stock, equipment discounts and a direct line to their R&D team. The stack is hardware agnostic, so in regions where it may not be as straightforward to use DJI products, another off-the-shelf drone does the job.

The problem is the same everywhere, but market structure decides who you sell to

Polybee is running in Australia, the US and the UK, on crops from spinach to strawberries. Across that spread, what travels and what has to be relearned?

The common denominator everywhere is the missing visibility and control. Polybee walks into a new region or a new crop, starts talking about forecasting, and hears a version of the same line every time. “Where has this been all this time? We have been looking for something like this for years.” Siddharth still finds the uniformity of that response surprising.

What differs is market structure, and it changes who you sell to. In the US, grower-shipper-packers hold much of the share. They run some of their own production, contract out the rest, process it, and deal directly with retailers.

That makes the shipper-packer the customer, not the contracted farmer. Polybee’s rule is to find whoever suffers from a bad forecast and can also act on a good one. A shipper-packer influences harvest timing across its contracted growers, so it can pull the lever. The contracted farmer is paid a fixed price per hectare and carries the production risk, so an accurate forecast leaves him nothing to do with it.

Elsewhere the answer inverts. A vertically integrated grower selling direct to a retailer holds the harvest decision and the retail relationship in the same hands, and that grower is the target. The product does not change. The ideal customer profile does, region by region, and getting it wrong means selling an accurate forecast to someone who cannot use it.

Advice for first-time founders: “If it’s not a resounding yes, it’s a no”

Siddharth is a first-time founder and had to learn fundraising, hiring, strategy and customer discovery at the same time. Two lessons came out of it that he thinks travel beyond agriculture to any B2B business.

The first is to sketch the customer’s ROI early, before you have earned the right to be precise about it. What does someone walk away with if they adopt this? Get it right to an order of magnitude at the start, then spend the pilot proving it.

That work becomes the proof-of-value stage and, later, the case study sales runs on. One condition: do it with development partners willing to walk the hard part of the road with you, and reward them fairly when the product commercialises.

Australia is where that played out. Polybee went there first for the proximity to Singapore, the R&D infrastructure and the tech workforce, and treated it as a sandbox. Growers including Perfection Fresh, Flavorite and Boratto Farms let the team onto their farms to test an unfinished product and work it out together.

The sandbox became the company’s second home. Polybee opened its ANZ headquarters in Melbourne in 2025, and Australian vegetable growers are the first market for the Bayer partnership. “Our early adopters are equal innovators alongside us,” he has said of them. “We wouldn’t have been able to build this product without their feedback and their involvement.”

The second is hiring, learned the way most founders learn it. “If it’s not a resounding yes, it’s a no,” he said. Do not talk yourself into an 80% yes with a 20% reservation, and do not hire because the opening is open and the candidate looks decent. Delaying is the cheaper mistake.

Want to connect with Siddharth?

Both products are commercially launched, early customers are scaling, and the channel partnerships are there to keep the pipeline fed. Polybee knows what the product is, what it is worth, and who makes a good customer. What comes next, in Siddharth’s own words, is growth.

Siddharth is open to hearing from growers, channel partners, and talent looking to join Polybee’s next stage. You can reach the team at info@polybee.co and Siddharth on LinkedIn.

Made in APAC 🌏

This is part of a series of conversations with founders and operators across the APAC region doing industry-leading work that shapes how we eat, grow food, and nourish ourselves.

Check out my first conversation with New Zealand ag tech unicorn Halter.

Stay tuned. More to come!

Unlock Asia’s trillion-dollar opportunity

Don’t miss your chance to attend the Asia-Pacific Agri-Food Innovation Summit and get 10% OFF with my network code. Use my exclusive partner code for 10% OFF your pass: BETTERBIO10. See you there!

Learn more

If you found value in this newsletter, consider sharing it with a friend who might benefit from it!

Share

Or, if someone forwarded this to you, consider subscribing.

Subscribe now

Disclaimer: The views and opinions expressed in this newsletter are my own and do not reflect those of my employer, affiliates, or any organisations I am associated with.

Today, I want to pick up where the last one left off. Last week I mapped the seven positions companies are taking in healthspan and spent one section on what the incumbents were buying. I was midway through drafting this one, on how those incumbents are playing it, and a few days ago that section became a priced fact.

Procter & Gamble (P&G) agreed to acquire Thorne for $3.8B in cash, with closing expected in the fourth quarter. L Catterton took the same company private in October 2023 at $680M. That is 5.6x in under three years.

While the number is the headline, I found the list of interested parties more interesting. Reuters reported in late June that Haleon had bid. The Financial Times reported around the same time that Unilever was considering a bid at up to $4B, though how far that went is not clear from the public record and Unilever has not commented. Shailesh Jejurikar, P&G’s chief executive, declined when asked to confirm that P&G had come through a competitive process at all.

Even on that thinner record, three companies from three different industries were circling the same asset: a consumer health business spun out of GSK, a household products giant, and a food and personal care conglomerate midway through divesting its food. They were not looking at the same thing.

What follows is my attempt to work out what each of them saw. The short version is that every company in this story wants something it cannot grow on its own. They are short of different things, and the thing most of them are short of has got twice as expensive in nine years.

P&G reported 1% growth six days before agreeing to pay for 30%

On 29 July, P&G reported fiscal 2026 results: net sales of $87.0B, organic sales growth of 1%, and volume flat for the year. Fourth-quarter organic sales were zero. Health Care took the steepest volume decline of any segment at 3%, with organic sales down 1% in the quarter, on full-year net sales of $12.5B and full-year organic growth of 1%.

Guidance for fiscal 2027 is 1% to 3%. Six days later, the company committed $3.8B to a business compounding above 30% a year and tracking ~$650M of 2026 revenue. That is close to 5.9x forward sales, paid by a company that, based on recent data, could not generate volume growth in its own health business last year.

Haleon had the same problem but, in my view, in a more uncomfortable place. Its FY2025 organic revenue grew 3.0% to £11.0bn. Its VMS division grew 1.9% to £1.69bn, behind the group, and the weakness was specifically in North America, where the company pointed to a soft multivitamin market, heavier competitor promotions and distribution losses it says it has since addressed. Thorne is a US premium brand sitting on a US practitioner base, which is the hole in Haleon's map. It bid and did not win.

Unilever, reported to be considering the same asset, was looking at it from a seemingly different position. Its first-half 2026 underlying sales grew 4.8% on underlying volume growth of 4.2%, with second-quarter volume at 5.5%, its strongest since 2010.

So the same asset was being weighed from different starting positions. The two companies documented as pursuing it are the two whose own numbers say they cannot produce that growth internally. Unilever’s numbers look nothing like either of theirs, and what it did with that is not on the public record.

Evidence can be bought, credibility has to be inherited

Three assets sit behind these deals, and each behaves differently. Evidence is a clinical file: trials, endpoints, a safety dossier, a novel food authorisation. Credibility is what exists once somebody qualified has read that file and is willing to attach their name to it in front of a patient. Permission is the right to make a claim in public without a regulator or a retail buyer objecting.

Evidence can be bought outright, and it usually is. Permission can be assembled in a few years with clean manufacturing, third-party certification and good lawyers. But credibility is the slow one.

Thorne is not worth $3.8B because its trials are better than everybody else’s. It is worth that because tens of thousands of practitioners have already turned those trials into trust, and trust is the part that does not come attached to the file.

What each buyer already owns decides what it has to buy

Underneath the deal announcements, the acquirers in this story fall into four groups, and they differ in which of those three they are shortest of.

These are tendencies rather than rules. The largest companies span more than one group, and the exceptions are real. Nestlé sits squarely in the first group and has owned a leading practitioner brand since 2017, which I come back to. What holds across all four is the pattern in what each has been paying for.

Food and nutrition majors

They hold shelf space, manufacturing scale and a brand-building machine. What most of them are shortest of is clinical credibility and a relationship with the people who recommend products for a living.

Their largest cheques have gone to businesses people consume on a schedule. Unilever paid $1.2B for Grüns in April, a company founded in 2023 that had passed a $300M annualised run rate by October 2025, which is ~4x sales. Danone agreed to buy Huel at around €1B, a deal still working through a CMA phase 1 inquiry with a decision due 11 September.

Danone also bought The Akkermansia Company, a strain with a decade of academic work behind it, for a sum neither party disclosed.

Personal care and household groups

They own something the food majors do not, which is a defended slot in the daily routine and the retail leverage to hold it. What is hardest for them to manufacture is credibility with anyone who reads a label closely.

P&G is not new to this aisle. It bought New Chapter in 2012 for an undisclosed sum, and paid €3.4bn for Merck KGaA’s consumer health business in 2018, picking up vitamin brands including Seven Seas and Bion3. What eight years and two acquisitions did not buy was a practitioner base, which is the specific thing Thorne has but Seven Seas does not.

Kimberly-Clark is running the same logic at a different scale with its $48.7B acquisition of Kenvue, a tissue company buying the maker of Tylenol and Band-Aid.

Consumer health companies

They hold the pharmacy shelf, the regulatory apparatus and the pharmacist relationship. What they are short of, based on recent trends, is growth.

Alongside Haleon’s 1.9%, which I mentioned earlier, Bayer’s Nutritionals sales fell 3.9% in 2025 on a currency- and portfolio-adjusted basis. This is the class with the deepest category credibility and the weakest momentum, which puts its members on both sides of the table. T

They need to buy growth, and their own credibility makes them worth buying. Haleon bid for Thorne, and Kenvue is being acquired, inside the same twelve months.

Ingredient houses

They hold the thing everyone else is trying to rent, which is the dossier: the trials, the novel food files, the manufacturing know-how. What they are furthest from is the consumer. Their moves have been to buy evidence and shed commodity volume.

dsm-firmenich paid €275M for the postbiotics producer Adare Biome, which brought with it Lactéol and the postbiotic Lactobacillus LB. Over the same period it agreed to divest animal nutrition at a €2.2B enterprise value. Buy the file, sell the tonnage.

All four groups already have factories, formulation scientists and shelf space. What separates them is which of the three assets they cannot make for themselves, and that is what sets the price they end up paying.

A dossier has a knowable cost. Any of these buyers can work out roughly what it would take to run the trials in-house, and most of them could afford to, so nobody pays far above that number. That is why dsm-firmenich got Adare Biome for €275M. A practitioner network has no equivalent figure, because no budget shortens it. Thorne has been building its clinician base since 1984, and a buyer who wants one either buys Thorne or starts in 1984.

So the question that sets the price is whether a competitor could get what you have by hiring for it. If they could, you are worth what your growth is worth. If they could not, you are worth whatever a buyer will pay to avoid waiting. Grüns went at 4x sales. Thorne went at nearly 6x. The same question decides who will lend against you, which I will come back to.

The clinician network survived the last time a giant bought one

One thing I wanted to know when the news broke was whether P&G can scale Thorne without destroying the clinical credibility that built it. The answer turns on what that credibility is for.

Thorne is used by tens of thousands of healthcare practitioners, and ~60% of its revenue now comes from consumers under 40. Colin Watts, Thorne’s chief executive, described a market that has moved from boomer-driven prevention to a Gen Z and millennial base buying for energy, sleep, anxiety and workout performance. The clinicians are not the customer. They are what makes the brand credible enough to charge a premium to people who will never sit in a clinic.

Which is the point I made in the last issue, before the price was known: a network of clinicians who vouch for you takes years to assemble and does not transfer with an acquisition, and the channel moat reaches a large outcome when it sits underneath a consumer business.

Thorne at $3.8B is that sentence being priced. The clinician base and the consumer business were both necessary. A practitioner brand without a consumer business has not fetched anything close to this, and a consumer greens brand without the practitioners fetched 4x sales instead of nearly 6x.

That reframes the dilution risk, and it turns out this experiment has already been run once. In December 2017, Nestlé paid $2.3B for Atrium Innovations, a business doing close to $700M of sales, whose portfolio included Pure Encapsulations. Nestlé’s own announcement called it the number one recommended brand in the US practitioner market.

Eight years on, Nestlé’s 2025 results name Pure Encapsulations as one of the premium brands driving growth in its supplement business, against sales declines in its mainstream and value brands. When Nestlé decided this year which of those assets to keep and which to sell, the practitioner brand was on the keep list, and the supermarket brands were not.

So the network does transfer. What I had wrong in the last issue was the condition attached to it. A clinician base survives acquisition by a company that has never held one, provided the buyer does not try to run it as a mass brand, and in Nestlé’s case, it did not.

Whether P&G runs Thorne as a mass brand is a question about distribution strategy rather than about whether clinician trust is portable, and it is something I will be watching between now and the close.

What buyers pay tracks what they cannot build

If credibility is the asset that cannot be built to a deadline, that should show up in what people pay for it. Deal activity supports that, and the gap has been widening for nine years.

Thorne went at ~5.9x forward sales. Grüns went at ~4x, and Huel at closer to 3.5x. On the valuation, Jejurikar told CNBC that the price “is a good price for the growth rates they have” and “kind of in line with the industry benchmarks we’ve seen.” A chief executive describing nearly 6x forward sales as a benchmark tells me the repricing is no longer an exception.

The closest comparable transaction is eight years old. Nestlé’s $2.3B for Atrium against close to $700M of sales is a shade over 3x. It is not a clean like-for-like, since Thorne is one brand and Atrium was eleven, and Garden of Life rather than Pure Encapsulations was the largest of them. It is still the nearest benchmark in this category, in my view.

Two more numbers sit in the same place. Nestlé paid $5.75B for the Bountiful core brands in 2021 at 3.1x trailing sales. The mainstream part of that portfolio, now classified as assets held for sale, carries ~CHF 1bn of annual revenue and has been put by market sources, not by Nestlé, at a potential value of EUR 3bn to 4bn. Around 3x again.

Atrium in 2017, Bountiful in 2021, the Nestlé mass tail today. Three points across nine years, all at ~3x sales, against Thorne at nearly 6x. The mass end has not got cheaper. The premium end has got twice as expensive.

This is not just a European or North American thing. Kirin paid A$1.88bn for Blackmores in 2023, Australia’s largest vitamin company, against A$649.5M of FY2022 revenue. That is 2.9x, and the buyer was a Japanese brewer looking to reduce its dependence on beer.

The floor under the mass end is private label, and it is rising. Private label reached 21.3% of US sales in 2025 and grew 3.3% against 1.2% for national brands, nearly triple the rate. Costco told its 2026 annual meeting that Kirkland Signature sold $90B in 2025. A commodity multivitamin has no answer to that, which is why I doubt anybody will pay more than 3x sales for one.

Nestlé is a good illustration because it holds both ends. Nature’s Bounty, Osteo Bi-Flex and Puritan’s Pride are going. Garden of Life, Solgar and Pure Encapsulations are staying, on Nestlé’s stated grounds that its capabilities in science and brand-building give it a competitive edge in premium. One company, two multiples, and it has decided which one it wants to own.

Kirin was not an outlier in its region. Asian buyers have been acquiring Australian and New Zealand supplement brands for a decade. Biostime took 83% of Swisse at an A$1.67bn enterprise value in 2015, Shanghai Pharmaceuticals took Vitaco in 2016, and By-Health took Life-Space at A$690M in 2018.

What they were buying is the same category of asset P&G bought, in a different form. Australian provenance and TGA oversight work as a trust signal in Chinese and Southeast Asian markets in a way domestic brands have not been able to replicate. It is credibility that comes from elsewhere and cannot be assembled at home on a schedule.

The difference is who owns it. Australian provenance belongs to every Australian brand, so a buyer who wants it can have it by acquiring any of them. That is a category attribute, and it prices like one, which is why Blackmores went at 2.9x. A clinician network belongs to one company. That is the whole reason Thorne went at twice the multiple.

Which leaves every incumbent in this article chasing the same short list. The mass shelf trades at a multiple that makes it a poor use of an acquisition budget, and the number of brands with a real clinical base is small. That is how a single supplement brand ends up drawing interest from three industries at once.

Capital is pricing predictability

Three pools of money are active in this category, and they are buying three different things.

Strategic acquirers pay for credibility attached to a habit, which is most of this article so far. Private equity pays for growth it can engineer, which is what L Catterton did at Thorne, entering at a 94% premium to the unaffected share price and exiting at 5.6x the entry valuation under three years later.

It is not an isolated trade. Bain Capital agreed to acquire Vitabiotics, the UK’s largest vitamin company, at a reported £900M on 25 July, ten days before the Thorne announcement. Two supplement deals above a billion dollars landed inside a fortnight, and a sponsor was on one side of both.

The third pool buys something the other two do not, and it gave me a different way of looking at which positions in this category can raise money.

On 14 July, direct-to-consumer wellness brand IM8, co-founded by David Beckham and owned by Prenetics, secured $1B in non-dilutive growth financing from General Catalyst’s Customer Value Fund. The structure funds up to 70% of the company’s marketing spend in exchange for a capped share of income from the customer cohorts that spending acquires. Once the fund recovers its investment and its capped return on a given cohort, everything after that belongs to IM8. The company reported around $17M of revenue in June and projects a $300M annualised run rate by year-end.

In the last issue, I put IM8 in two of the seven positions at once, the narrative play and the format play, because as far as public records show, it markets on the literature behind its ingredients rather than trials of its own formulations, and sells the result as a daily routine. What is hard to replicate and what is easy to finance are close to inversely related here.

A clinical trial has no cohort payback curve and no repayment profile, so there is very little for a lender to underwrite. A customer on a monthly reorder has both. A company that owns a molecule and the trials behind it holds the asset that is hardest to replicate and hardest to borrow against. A brand built on licensed ingredients and a daily habit holds the one that is easiest to copy and easiest to fund.

Closing thoughts

The three assets this category runs on are not priced the same way, and the gap between them is the thing I did not see last week. Evidence changes hands for a few hundred million because any serious buyer can work out what it would cost to produce the same file. Credibility went for $3.8B because none of them can work out how to produce it faster than Thorne did.

What I also did not have was the shape of the buyer set. A household products company, a consumer health business spun out of a pharma group and a food conglomerate midway through divesting its food were all circling the same asset.

The industry boundaries that sort these companies no longer sort the assets they are chasing, and the ones that moved hardest are the ones whose own numbers leave them the least room to wait.

What sits underneath that is harder to square. Credibility is manufactured out of evidence. Practitioners recommend Thorne because somebody ran the trials, and kept running them for forty years. So the market has decided that credibility is the expensive asset and evidence is the cheap one, while evidence is the raw material credibility is made from. The category has repriced the finished product of a process it is not paying to run.

The obvious reading is that the premium corrects. High prices pull in supply, and supply brings the price back down. I do not think that happens here, because supply is the one thing this asset class cannot produce to order. A clinician network cannot be accelerated by spending more on it, and if nobody funds the evidence that new practitioner brands would have to be built on, the ones that already exist get scarcer.

As always, my understanding keeps evolving here. This piece corrects something I wrote last week, and the deal at the centre of it has not even closed yet. If you spot something I have missed or got wrong, let me know.

And if you are someone investing in this space, I would love to connect. Thanks for reading!

Unlock Asia’s trillion-dollar opportunity

Don’t miss your chance to attend the Asia-Pacific Agri-Food Innovation Summit and get 10% OFF with my network code. Use my exclusive partner code for 10% OFF your pass: BETTERBIO10

Learn more

See you there!

If you found value in this newsletter, consider sharing it with a friend who might benefit from it!

Share

Or, if someone forwarded this to you, consider subscribing.

Subscribe now

Disclaimer: The views and opinions expressed in this newsletter are my own and do not reflect those of my employer, affiliates, or any organisations I am associated with.

The conversation in human health has been shifting from lifespan extension to healthspan. I have spent the last few weeks working out who in the food and nutraceutical space is capturing that shift and how.

The two words describe different things. Lifespan is how long you live. Healthspan is how long you live in good health, before chronic disease and functional decline take hold.

Put simply, healthspan extension is not about adding years to your life. It’s about adding life to your years. That to me is a much more compelling proposition than being promised you will see 120.

A 2024 analysis in JAMA Network Open covering all 183 WHO member states put the average gap between them at almost 10 years, and found it had widened by 13% since 2000. Most of us can expect to spend roughly a decade at the end of life unwell, and that decade is what the industry has started selling against.

What follows is a map. I want to lay out the distinct positions companies are taking, which ones look defensible, what the incumbents are doing about it, and what this shift does to the rest of food and nutraceuticals.

Let’s dig in!

The demand shifted from clean labels to clinical proof

The demand is not in doubt. McKinsey’s latest wellness research, previewed in June, has 84% of consumers ranking wellness a top priority, with more than half saying they prioritise it more than they did a year ago. Longevity is one of the growth pockets it names, and cognitive health is one of the gaps, with consumers reporting they could not find products to help them maintain it.

What the supplement aisle converts is narrower. Lumina Intelligence finds 57% of consumers recognise the term healthspan, 63% believe anti-ageing supplements can extend it, and 77% say they are interested in products supporting cellular repair. Only around 20% buy an anti-ageing supplement in a given year. Interest runs three to four times ahead of purchase, which is the gap every company in this article is trying to close.

What changed is what people want proved before they buy. McKinsey describes consumer preference moving from clean and natural ingredients toward clinically proven efficacy, and finds the shift sharpest in supplements and over-the-counter products. Doctor recommendations are regaining ground as a trusted source at the same time. Both of those are requests for somebody else to have checked the work first, and they are the demand-side reason the rest of this article looks the way it does.

There is an awkward gap underneath all of it. Healthspan has no agreed measure. At the population level there is one, the health-adjusted life expectancy figure the WHO uses, which is where that 9.6-year gap comes from.

At the level of a person or a product, there is nothing settled. A systematic review published last September screened over 14,000 papers and found 113 primary definitions still in circulation, concluding that the absence of a standard is now hampering comparison between studies.

So people are measuring more about themselves than ever, and none of it is a validated measure of the thing they are trying to improve. Blood panels, continuous glucose monitors and epigenetic clocks like GrimAge and DunedinPACE all produce numbers.

None of those numbers has an agreed relationship to how many healthy years you get. That is the situation in which “clinically proven” becomes the shortcut consumers reach for, and in which a clinician’s recommendation regains value.

GLP-1 drugs are what turned all of this from a preference into an expectation. Around 18% of American adults now use one, up from roughly 14% a year earlier. Users eat less, so every bite has to carry more, which has turned nutrient density into a formulation requirement and pushed protein, fibre and portion-controlled formats across the industry.

Muscle preservation has become a named target rather than a later worry. The larger effect is that millions of people have taken something and watched a measurable change happen in their own body over a few months. That resets the bar for everything else they consume.

Demographics sit behind all of it. The WHO projects that one in six people worldwide will be 60 or over by 2030, that the over-60 population reaches 2.1 billion by 2050, and that the over-80 group triples to 426 million over the same period.

That combination changes the design constraints. A food company optimising for taste, convenience and cost now has to think about clinically meaningful dosing, daily compliance, ingredient provenance and long-term safety. Precision fermentation earns its place here on consistency, because it produces a defined molecule at the same concentration every batch, which is what you need when the dose is the claim.

Which brings me to the problem. Euromonitor, via senior global insight manager Nick Stene, finds that half of all newly launched dietary supplement SKUs since July 2024 carry at least one healthspan claim. NAD+ messaging alone appears on just under 6%. When a claim appears on half of everything new, it stops helping anyone sort.

The seven positions companies are taking

Underneath the marketing, I count seven different positions. They differ in what the company owns and how easily someone else could come to own it, too.

The mechanism owner

They hold a molecule mapped to a named biological process of ageing, backed by its own human trials. Timeline is one example. Niagen Bioscience, formerly ChromaDex, is probably the purest one, with nicotinamide riboside tied to NAD+ decline, more than 35 published human trials and EU novel food authorisation. Its eight-week study in 132 overweight adults showed a 51% rise in whole-blood NAD+ at 300mg a day.

The proof burden is the heaviest in the category: randomised human trials on your ingredient at your dose, plus a safety dossier clearing novel food or new dietary ingredient review. The molecule itself is copyable. The trial record and the regulatory file are the durable part.

The strain holder

They own a living organism with clinical provenance. Pasteurised Akkermansia muciniphila is the cleanest case, characterised by Willem de Vos and Patrice Cani, given a positive EFSA safety opinion in 2021 and authorised in the EU in 2022 as a food supplement at up to 5 × 10¹⁰ cells a day.

Probiotic and postbiotic effects are strain-specific, so a competitor cannot substitute a related organism and inherit your data. The moat is the patent estate, the proprietary trial record and the manufacturing know-how, since deposits in public collections can often be obtained by third parties.

The production play

They make biologically what was previously extraction-limited, animal-derived or prohibitively expensive, and increasingly compete on delivery as well as supply. Helaina’s precision-fermented human lactoferrin sits here, alongside fermentation-derived ergothioneine, mycoprotein platforms, and the bioavailability work around liposomal and phytosome formats.

The proof burden rests on equivalence, safety and a cost curve. These companies win by becoming default supply, which holds only as long as nobody makes the same molecule more cheaply.

The measurement and personalisation play

They own the diagnostic and sell the intervention. Bioniq formulates from blood biomarkers, Viome from microbiome and gene expression, Elo Health from panels combined with wearable data.

The defensible layer sits underneath the assay, since competitors can run the same markers. It is the protocol that decides who gets flagged, what they are told to change, what gets rechecked and when.

The caveat is that no biomarker of ageing has been clinically validated, and there is no agreement yet on how one should be, so the scoreboard everyone is being judged against is not itself settled science.

The commercial logic still points one way. In April, Infinite Epigenetics, TruDiagnostic’s parent, acquired Tally Health and moved into supplements, so the company running the biological age scoreboard now sells products the scoreboard rates.

The practitioner channel play

They sell through gatekeepers who require evidence before they will recommend anything. Thorne, Pure Encapsulations, Designs for Health and Xymogen all operate here, alongside the dispensary platform Fullscript. The proof burden is different, weighted toward third-party certification through USP or NSF, clinical dosing, GMP compliance and practitioner education.

The channel is 9.2% of US supplement sales and growing over 7% a year. A network of clinicians who vouch for you takes years to assemble, which makes it one of the harder positions to attack. Worth noting that Thorne is not a pure channel play, since it serves 47,000 practitioners and more than five million direct customers. The channel moat reaches a large outcome when it sits underneath a consumer business.

The format and adherence play

They own a slot in the daily routine. Complete-meal and routine-based products like yfood, Huel, Grüns and Nestlé Vital sit here. The asset is repeat consumption, tapping into consumers’ habits. A greens powder somebody drinks every morning outperforms a validated ingredient they forget in a drawer.

Habit is durable while it lasts and it compounds, and it is the one position here a company with no scientific programme at all can build from scratch.

The narrative play

They assemble third-party-validated ingredients into a story. Avea and SRW Labs are examples. As far as public records show, these companies market on ingredient-level literature rather than published trials of their own finished formulations. IM8 is kind of in between this and the format play, and I would put it in both.

Incumbents are buying capability at different price points

The acquisition record is the most legible thing in this category. It splits by what is being bought, and in one case by what is being sold.

Brands and habits command the largest cheques

Nestlé paid $5.75B for The Bountiful Company's core brands in 2021, covering Nature's Bounty, Solgar, Osteo Bi-Flex and Puritan's Pride, at 3.1 times the acquired brands' trailing sales. In June, it completed its purchase of the remaining 51% of yfood Labs at a reported $523M valuation against €150M of 2025 sales.

Danone agreed in March to buy Huel for around €1B, a deal still working through UK competition review, and Unilever bought Grüns at a reported $1.2B in April, roughly thirteen months after a Series B valued it at $500M and about thirty months after launch.

All three targets are formats people consume on a schedule. Grüns joins Unilever's Wellbeing Collective alongside Liquid I.V., Nutrafol, SmartyPants, Onnit, Olly, Equilibra and Welly, a group whose combined sales already top $1.1B against a stated ambition past $3B. Unilever has spent four years assembling that portfolio.

Science assets trade quietly and considerably cheaper

dsm-firmenich bought the postbiotics producer Adare Biome at a €275M enterprise value, bringing in Lactéol and the postbiotic Lactobacillus LB. Danone acquired The Akkermansia Company in 2025 for an undisclosed sum, buying a strain with a decade of academic work behind it.

Nestlé's original stake in Amazentis was never reported. The one deal here with a public number sits an order of magnitude below the brand cheques, which is suggestive rather than conclusive. Undisclosed does not mean small, and acquirers leave bolt-on values unstated regardless of size. That said, Nestlé, Danone and dsm-firmenich are all listed companies with disclosure thresholds, so a large purchase would likely have had to surface.

The clearest example of what changes hands in these deals is Ingredion’s June acquisition of Benicaros from NutriLeads, an upcycled carrot pomace extract that works at a 300mg daily serving. The asset deal transferred the intellectual property, the trademarks, the human clinical trials and the manufacturing know-how.

The practitioner channel is being repriced

L Catterton took Thorne private in 2023 at $680M. The Financial Times reported in June that Unilever is exploring a bid valuing it at up to $4B, with Haleon also among interested parties, though no deal has been struck.

Thorne serves 47,000 healthcare practitioners alongside more than five million customers. An asset that changed hands at $680M three years ago is now being discussed at roughly six times that.

The portfolios are being repositioned toward consumer nutrition

dsm-firmenich is exiting Animal Nutrition and Health to CVC at a €2.2B enterprise value, which the company describes as the final strategic step in becoming a fully focused consumer company in nutrition, health and beauty.

IFF is selling Food Ingredients, its largest division, to CVC for $4.3B, and its Pharma Solutions business to Roquette for $2.85B, sharpening its focus on flavours, fragrances and health. Both are moving out of tonnage and into function.

Unilever has probably gone furthest. On 31 March it signed a $44.8B combination of its entire food business with McCormick, taking $15.7B in cash and $29.1B in shares and leaving itself with 9.9% of the combined company. Just over a week later, it announced the Grüns acquisition. A company that spent a century building Knorr and Hellmann’s handed them to a spice business and bought a greens gummy for a reported $1.2B in the same fortnight.

What incumbents build for themselves is worth reading alongside what they buy and sell. Nestlé Vital, launched in February for adults in midlife, runs its Morning Routine formulation on taurine, magnesium and B vitamins in a daily-routine format. dsm-firmenich is reframing its innovation story from lifespan to cellspan around four modifiable hallmarks of ageing, with its Humiome Post LB postbiotic at preclinical data and ongoing human trials. Distribution now, expensive science later, and a story broad enough to cover both.

So what does this mean?

The second-order effects reach well past the companies making the claims. I can think of seven consequences that follow from all of this, and they land unevenly across food and nutraceuticals.

Small producers get priced out of their own claims

A properly powered randomised trial with functional endpoints runs from the high hundreds of thousands into the low millions, before the regulatory dossier. Regional ingredient companies cannot self-fund that.

Their options narrow to licensing material into somebody else’s clinical programme, selling into a private-label or commodity position, or keeping the ingredient and dropping the claim.

Consolidation follows, and it consolidates around whoever can afford the evidence rather than whoever has the better ingredient. That is a different selection pressure from the one this industry ran on before.

A market forms for evidence itself

PLT Health Solutions partnered this year with the Alethios decentralised clinical trial platform, specifically to give ingredient customers faster and cheaper substantiation. Once trials become a purchasable service, the barrier drops for everyone building a dossier from here.

The businesses selling substantiation capture the upside of the shift and carry none of the claim risk, which makes them a structurally comfortable place to sit. Biomarker panels, manufacturing quality systems and regulatory consultancies are being pulled into a category that did not previously need them at this volume.

Value moves upstream to whoever holds the dossier

When the evidence sits with the ingredient supplier rather than the brand, the supplier holds the asset that survives a brand rotation. A finished-goods company that licenses its actives is renting its differentiation.

FrieslandCampina Ingredients frames the same shift as protein becoming the baseline rather than the headline, which is a supplier-side way of saying the value is migrating into the formulation and away from the label.

Mainstream food reformulates, and commodity nutrition gets re-rated

Protein and fibre are now the leading claim in categories that never used them, increasingly formulated together for satiety and nutrient density, and healthy ageing is showing up across snacks, bakery, dairy and frozen.

The GLP-1 numbers underwrite it. Body composition substudies put lean mass somewhere between a quarter and 40% of total weight lost, and the drug-by-drug ordering is still being argued over in 2026 data. Lean mass includes water, organs, and connective tissue, so it is not the same as muscle, and these proportions are broadly typical of rapid weight loss of any kind.

The commercial effect is that protein, fibre and leucine-rich formats picked up a clinical rationale they did not have five years ago, and companies sitting on decades of ordinary nutrition science suddenly hold relevant evidence.

The channel mix shifts toward both ends and hollows out the middle

Direct-to-consumer is now 29% of supplement sales, up from 16% in 2020, while the practitioner channel reaches around $6B this year on roughly 6% growth. Those are the two places where a proof story can be told at all, either by a clinician who reads the data or in a direct relationship long enough to explain it.

An undifferentiated retail shelf is the hard place to sell a claim nobody has time to evaluate. The boundary between the two ends is also blurring, with practitioner brands moving into direct and retail channels alongside the clinics.

Jurisdiction decides what a product is

In the EU, almost none of the marquee longevity ingredients carry an authorised health claim, even where the ingredient itself is an authorised novel food. In the US, the FTC’s finalised order against TruHeight this year specifies randomised, double-blind, placebo-controlled testing on the product itself or an essentially equivalent product.

The same formulation can be a compliant product in one market and an enforcement target in another, which means the claim strategy has to be built market by market.

The word gets used up

At half of all new launches, healthspan has already stopped sorting anything for a shopper, a retail buyer or an investor. Consumers are moving in the same direction, with NielsenIQ describing a self-directed, results-driven consumer who checks claims for themselves.

Natural and clean label went through the same arc and came out as table stakes that nobody can charge for. When it happens here, the companies holding real dossiers inherit a category that nobody else is allowed to describe.

Closing thoughts

Consumers have started asking for proof, and the thing they want proved has no single measure. Consumers continue to buy healthy ageing products, and they increasingly want clinical evidence behind it.

Healthspan turns out to be several things: mobility, sleep, cognition, metabolic control, the muscle you keep. The review that found 113 definitions was pointing at a missing standard, since plenty of people define healthspan and nobody has agreed which definition counts. So each company claims a piece of it, and none of them can claim the whole.

Reading back across the seven, the question that separates them is what a competitor would have to build rather than buy. On that test, the practitioner channel is, in my view, the hardest to attack, because a network of clinicians accumulates one relationship at a time and does not transfer with an acquisition.

The strain holder is close behind. The production play holds default supply only until somebody makes the same molecule more cheaply, and the narrative play owns the assembly and nothing underneath it.

That ranking sits against where the money has gone, and working out why changed my mind about it. Replication difficulty sets defensibility. What sets the price is scarcity in the buyer's own hands. Unilever can fund a trial and license a strain, and it has built its Wellbeing Collective to $1.1B in combined sales across seven brands. It still paid a reported $1.2B for thirty months of Grüns. That tells you what habit at that velocity is worth to a company that already knows how to build brands.

There is a ceiling problem too. A dossier sold into other people’s formulations is capped by what formulators will pay, and it is sold to ingredient houses that know what building the same file would cost. dsm-firmenich bought Adare Biome while building Humiome in-house, which tells you what a company with its own research organisation thinks a dossier is worth. The most defensible positions in this category have the smallest buyer set and the best-informed buyers.

The money is going to habit, and it makes a lot of sense. A product somebody takes every morning for three years is worth more than one with a good trial and no repeat purchase. Strains and dossiers change hands quietly while formats change hands loudly, and Thorne is an early real sign of the market repricing one of those positions.

As always, my understanding keeps evolving here, and this is a category where the science and the commercial reality are moving at different speeds. If you spot something I have missed or got wrong, let me know.

And if you are someone investing in this space, I would love to connect. Thanks for reading!

Don’t miss your chance to attend the Asia-Pacific Agri-Food Innovation Summit and get 10% OFF with my network code. Use my exclusive partner code for 10% OFF your pass: BETTERBIO10

Learn more

If you found value in this newsletter, consider sharing it with a friend who might benefit from it!

Share

Or, if someone forwarded this to you, consider subscribing.

Subscribe now

Disclaimer: The views and opinions expressed in this newsletter are my own and do not reflect those of my employer, affiliates, or any organisations I am associated with.

Below are the 10 most interesting I came across last week, paired with my thoughts on what they mean for where the industry is heading.

PS: If you are an investor or working at a corporate in food, agriculture, or biomanufacturing, do reach out. I would love to exchange notes on what I am seeing across the sectors. Reply to this email or drop me a message.

Let’s dig in.

🧈 Savor raised $32M to scale its carbon-derived fats, partnering with AAK on dairy and bakery applications

• California-based startup uses a thermochemical process to convert point-captured carbon, green hydrogen and methane into agriculture-free fats that replace butter and palm oil. It has already commercialised its CO2 butter in San Francisco, where it appears in chocolate truffles at Michelin-starred One65 and baked goods at Jane the Bakery.

• Savor’s Carbon Crafted platform turns gases into carbon chains called alkanes, then into fatty acids through controlled temperature and pressure, producing short-, medium- and long-chain triglycerides that the company says are chemically identical to conventional fats. It needs no fertile land, fertilisers or freshwater, and Savor claims the fats cut carbon emissions by 98% and require 1,000 times less land than conventional counterparts.

• The round takes Savor from a few tonnes of product a year to a couple hundred tonnes, ahead of a planned Series B to build a 10,000-tonne facility. A two-year joint development agreement pairs Savor’s conversion technology with Swedish speciality fats leader AAK’s formulation expertise and commercial infrastructure, targeting dairy alternatives and baked goods across the US and European markets.

Investors: AAK

Source: Green Queen

Thoughts 🤔

AAK’s involvement fits a broader pattern. It has been building exposure across multiple next-generation fat platforms, with partnerships already in place with Checkerspot on microalgae-derived oils and Arzeda on computationally designed ingredients.

Thermochemical, algal, and fermentation-enabled are different technical routes to producing fats without traditional agriculture, and AAK now holds a position in each. Each carries different technical and capital risks. So AAK is placing modest, structured bets (equity in Savor plus JDAs elsewhere) that give it a close view into each platform’s progress.

In return, it supplies what these companies typically lack: mature formulation capability, regulatory and quality infrastructure, and access to food manufacturers who already buy specialty fats.

This effectively positions AAK at the integration layer. However fats get produced upstream, they still have to be functional, consistent, and integrated into existing food systems, and by sitting at that interface AAK keeps the early-stage R&D risk with the startups.

🌾 Rize raised $31M series B to scale low-emission rice farming across Southeast Asia

• The Singapore-based startup runs an agronomy platform that helps smallholders adopt alternative wetting and drying (AWD), a method of cycling paddies between wet and dry phases rather than keeping them flooded. Over the past 2 years it has grown tenfold, reaching 17,000 farmers across 50,000 hectares in Vietnam and Indonesia and shipping 1,500 tonnes of low-emission rice to Europe, Canada, Australia, and Singapore.

• Rize champions AWD by buying seeds, fertilisers, and other inputs in bulk and reselling them to farmers, and it ensures the rice meets the Maximum Residue Limit standards that premium export markets require. The startup says its approach cuts emissions in half and water use by up to 30%, while raising farmer incomes by 30% without reducing yield.

• The round splits into $20M of equity and $11M of debt. The capital will scale AWD into new markets, advance carbon certification, and build AI tools for farmers and field teams. Rize aims to reach more than 150,000 farmers across 300,000 hectares by the end of the decade and to eliminate 500M tonnes of CO2e by 2040.

Investors: BNP Paribas Asset Management, the Rockefeller Foundation, Temasek, and Breakthrough Energy Ventures

Source: Green Queen

Thoughts 🤔

Rize buys inputs in bulk, handles market linkage, and ships verified low-emission rice into premium export markets. Switching farmers to AWD takes upfront capital, and Rize appears to be carrying a good part of that cost itself.

That is a working-capital-heavy model, which is why the round pairs $20M of equity with $11M of debt. Lenders lend against cashflows and assets. Impact and climate investors buy the upside.

That mix tells you what kind of business this is. One of the hardest barriers to smallholder practice change is financing. Someone has to bridge the gap between paying for inputs now and earning the low-emission return later.

Rize seems to be absorbing a chunk of that gap, and as the network scales toward 150,000 farmers, the capital it ties up should grow with it. That is the kind of cost debt is built to carry.

🥛 TurtleTree signed an exclusive deal with Novonesis to scale its precision-fermented lactoferrin and raised funding

• The Singapore-based startup makes LF+, a recombinant bovine lactoferrin produced through precision fermentation. The ingredient suits sports nutrition, women’s health, adult and elderly nutrition, and functional foods. Lactoferrin regulates iron, supports immunity, and aids gut health, and has long been supply-constrained as a scarce dairy-derived protein.

• Under the agreement, Danish biosolutions leader Novonesis will scale, manufacture, and commercialise the protein through its Human Health division, using its US production facility. Novonesis is targeting the early-life nutrition market while securing selected commercial rights for dietary supplements. The company also made a minor equity investment in TurtleTree alongside the deal.

• TurtleTree has already commercialised LF+ through its consumer brand Intentional, whose first supplement, IronKind, pairs lactoferrin with prebiotics for iron regulation, energy, and gut health. The ingredient features in Cadence Performance Coffee espresso shots and a partnership with animal-free dairy firm Strive Nutrition.

Investors: Novonesis, 21Catalyst Ventures (the VC arm of Mitsui Chemicals)

Source: Green Queen

Thoughts 🤔

TurtleTree spent years building its lactoferrin strain, regulatory dossier and early brand. Handing exclusive early-life nutrition scale-up and commercialisation to Novonesis is a clear choice about which layer of the stack it wants to own.

It keeps the strain and the pioneering GRAS win, and effectively rents an incumbent’s industrial fermentation footprint and customer relationships instead of trying to build both off a venture-backed balance sheet. In a tough funding market, this looks like the pragmatic version of a broader shift in biomanufacturing away from owning every step of the value chain.

The equity piece is the part that stands out. Novonesis is taking a minority stake in the startup whose molecule it will exclusively manufacture and sell into early-life nutrition, with Mitsui’s venture arm alongside.

For a large ingredients incumbent, that is a way to buy optionality on a leading precision-fermented lactoferrin candidate in a competitive field, cheaper and faster than building a strain from scratch or acquiring the company outright. Of course, an eventual acquisition sits in the background as one potential outcome if the partnership delivers the performance Novonesis wants.

As the production side of precision fermentation gets more crowded and making the molecule becomes table stakes, getting it into early-life nutrition brands and formulations is the harder, more valuable part.

🥚 The Every Company teams up with ADM to scale up US production of its animal-free egg protein

• The animal-free egg protein maker will produce its OvoPro ingredient at ADM’s Iowa facility at commercial scale, with the plant set to come online next year. Every says it marks the first US site with precision fermentation capacity at this scale.

• OvoPro is a recombinant ovalbumin that replicates the functionality of egg whites, working as both a binding agent and a protein-boosting ingredient across baked goods, bars, snacks, confectionery, and pasta. It has FDA clearance for sale.

• The deal comes weeks after Every quadrupled its European output through Huvepharma, producing OvoPro at a 9M-litre fermentation site in Bulgaria that runs the same feedstock and process. Every favours co-manufacturing because wholly owned plants can exceed $200M in CAPEX and carry multi-year lead times.

Source: Green Queen

Thoughts 🤔

Every is building something closer to a distributed manufacturing model than a single-site biotech operation. With Huvepharma in Bulgaria and ADM in Iowa, it is replicating the same organism, feedstock, and process across partners in two regions.

That is significant because formulators adopt OvoPro to reduce exposure to a concentrated, volatile egg supply chain and a single production site would reintroduce that risk. Multi-site capacity starts to address it, even if true interchangeability takes time as facilities are qualified.

ADM’s role adds another layer. OvoPro is landing inside an existing biomanufacturing and fermentation services business that already converts commodity inputs into tailored proteins and bio‑based materials at industrial scale. ADM has signaled that precision fermentation is part of how it plans to serve future protein demand across food and feed, not just a sustainability add-on.

This deal is a good example of a scaling model where startups standardize biology and process, and incumbents provide manufacturing infrastructure. It is a practical path for bringing precision-fermented ingredients into mainstream supply chains.

🍄 Syngenta and Groundwork BioAg partner to bring biologicals and soil carbon solutions to farmers

• The partnership has Syngenta commercialising mycorrhiza-based products and soil carbon solutions under its own brand, starting with corn, soy, cereals, and sunflower across Latin America and Europe. Groundwork BioAg takes on manufacturing, supply, digital tooling, and the full carbon program development.

• Mycorrhizal fungi form symbiotic relationships with crop roots, improving nutrient and water uptake while catalysing durable mineral-associated organic matter in the soil. That organic matter is what locks carbon away over the long term, so the same input that improves yields also builds a measurable sequestration asset.

• Farmers get two returns from one application: higher, more resilient yields and a new revenue stream from carbon credits sold under Syngenta’s label. For a leading biological crop protection company, the deal pushes biologicals further toward the centre of its portfolio and pairs them with a carbon program most input suppliers cannot yet offer.

Source: Syngenta

💪🏾 dsm-firmenich teams up with Nous to launch a caffeine-free botanical energy ingredient globally

• The ingredient, called Koncentra, is a caffeine-free botanical blend of ashwagandha, ginkgo biloba, holy basil, and oats. Nous makes it using a proprietary natural extraction process it says preserves the bioactive compounds while keeping quality consistent and scalable. It targets energy, focus, and mood.

• dsm-firmenich is positioning Koncentra as an alternative to caffeine, offering sustained performance without the jitters, anxiety, and crashes consumers associate with stimulants. In powder form it is soluble and works at a low effective dose, which the company says lets formulators drop it into food and drinks with minimal impact on taste, colour, or turbidity.

• dsm-firmenich recently invested in Nous through its ventures arm and will now fund the ingredient’s development, claims substantiation, and market validation. It plans to design clinical trials to back Koncentra’s energy and cognitive and physical performance claims, moving it from clinical development to commercialisation.

Source: Nutrition Insight

Thoughts 🤔

The Koncentra blend is familiar botanicals. Ashwagandha, ginkgo, holy basil, and oats are off-patent and buildable by many formulators. The scarce inputs are clinical substantiation, multi-market regulatory clearance, and global distribution, and those are what dsm-firmenich is contributing rather than Nous.

This is a classic dsm-firmenich Ventures-style play: take an early stake, then plug the asset into the parent company’s clinical, regulatory, and market-development machinery. The ingredient majors are increasingly buying optionality on external innovation once the science looks promising, rather than trying to invent every ingredient in-house.

The sequencing is key here. dsm-firmenich says it will now help design the clinical trials needed to support Koncentra’s energy, cognitive, and physical-performance claims, which means the commercial story is still ahead of the proof package. In a category where a competitor like Nektium’s Zynamite already carries nine human trials, that gap matters.

If the eventual data only matches what better-known botanicals already deliver, Koncentra’s edge may come down to formulation convenience and extraction know-how more than a truly new efficacy story.

⚙️ Tetra Pak launches its first industrial bioreactor for fermentation-derived food ingredients

• The Bioreactor RF targets producers using yeast, bacteria, and fungi to convert raw materials into food ingredients and is built to hold stable conditions from pilot validation through to full-scale manufacturing. It comes in sizes from 10 to 50,000 litres, with larger vessels on request, and follows Tetra Pak’s recent acquisition of Bioreactors.net.

• At the centre is a patented magnetic agitation system that removes mechanical seals, a common source of contamination risk and instability in conventional designs. Tetra Pak says it maintains power and oxygen transfer across the full size range, addressing a scale-up limitation in traditional magnetic agitators and lowering the risk of batch loss.

• It says the preassembled, factory-tested unit can cut operating costs by up to 12% and investment costs by up to 8% vs conventional industrial bioreactors. Every unit shares the same design platform, control logic, and operating philosophy, letting producers standardise operations as they scale across sites.

Source: Food Ingredients First

Thoughts 🤔

Scaling fermentation cleanly from lab to commercial is still one of the category’s hardest problems. Process conditions shift as vessels get bigger, and contamination or seal failures can still wipe out batches.

That is what makes Tetra Pak’s Bioreactor RF notable. What it offers is standardization: the same control logic, operating philosophy, and magnetic agitation platform from 10 litres to 50,000 litres, with claims of faster commissioning and lower cost.

The real value is continuity between stages. If the system transfers performance across scales, it could cut re-validation pain and process risk for producers moving from pilot to industrial production.

🥔 Aardaia raised €5M seed round to breed wild plants into new crops, starting with a protein-rich tuber

• The Netherlands-based startup invents crops from first principles rather than re-engineering commodity species. It identifies wild plants already carrying useful traits and breeds them into viable crops, using no genetic modification or gene editing. Its flagship, the aardaker, is a protein-rich tuber that pairs the productivity of a root crop with the nitrogen-fixing biology of a legume.

• Aardaia builds large genomic datasets and combines whole-genome sequencing with detailed phenotyping to predict in silico which genetics will perform before anything is planted. The startup claims the aardaker needs no synthetic nitrogen fertiliser and could produce up to 5x more protein per hectare than any current crop, a target it is still working toward.

• This year Aardaia is screening 750,000 unique aardaker genotypes, and it aims to push toward 2 million next year with the new funding. The company notes that ~95% of global calories come from just 30 species, despite more than 400,000 plant species on Earth, most never domesticated for food.

Investors: Point Nine, Astanor, Grey Silo, FoodLabs, and angel investors

Source: EU-Startups

🧠 MeNow’s AI uncovers functional ingredients from waste with two bioactives revealed as potential disruptors in the GLP-1 arena

• The startup scans the molecular makeup of botanicals and their by-products to predict the bioavailability, safety, and health benefits of compounds, drawing on a library of 60,000 organisms and more than 2 million natural molecules built over the company’s first year.

• Commercial traction is already there. A collaboration with Colgate identified an optimal plant combination for treating gingivitis and cut development time by 75%. The platform has since surfaced a compound from discarded sorghum crop residue predicted to activate SIRT1, one of the most sought-after targets in longevity science for its links to anti-ageing and cellular protection.

• The founders see the greatest weight management potential in a brown seaweed byproduct, which activates the CCK1 receptor, a satiety regulator they claim the pharma industry could never stabilise as a drug. MeNow says it acts faster and stronger than the GLP-1 pathway while boosting nutrient absorption.

Source: NutraIngredients

Thoughts 🤔

For an ingredient startup, the molecule is often not the hardest part. Novel-food approval is, and in the EU that route now averages ~2.5 years to a safety opinion, sometimes over five, before anything reaches a shelf.

That is the lens I would read MeNow’s news through: its standout candidate is a GLP-1 agonist the company says is already cleared for food use, so it starts far closer to market than a novel molecule would, even if a new use still needs signing off.

That changes what the AI is for. As the discovery layer gets more crowded, with other companies mining botanicals and side streams on the same premise, scoring candidates for regulatory readiness goes after the part of the funnel that gates revenue. Surfacing an active molecule is not novel, but doing it with a compound that starts far closer to market, from a near-free waste stream, can help it stand out.

💰 Japan’s draft $2.3T public-private investment roadmap earmarks $6.2B for new foods within a $60B food tech push

• The Japanese government has released a draft public-private investment roadmap projecting ¥370T ($2.29T) in cumulative financing across 17 strategic sectors through 2040. Four of these verticals are food tech, together accounting for ¥9.7T ($60B), with ¥1T ($6.2B) ringfenced for new foods.

• The new foods vertical, covering animal-free proteins and functional nutrition, is the smallest of the four allocations. Indoor farming leads at ¥4.6T ($28.4B), followed by land-based aquaculture at ¥2.9T ($17.9B) and food machinery at ¥1.2T ($7.4B). The plan places strong emphasis on Japan’s existing strengths in fermentation, seasoning technologies, rice flour products, and marine ingredients.

• Japan is targeting ¥3T in new food sales across domestic and export markets by 2040, first courting health- and climate-conscious consumers in Europe and the US before expanding into Asia. The roadmap follows an earlier $8B biomanufacturing fund and the country’s stated goal of becoming the most advanced bioeconomy society by 2030.

Source: Green Queen

What else I’m reading and listening

• David Beckham’s IM8 longevity supplement brand secures US$1B in funding (Nutrition Insight)

• Fuel for active nutrition: AI and tech power personalized product innovations (Nutrition Insight)

• Biotech M&A on track for best year since pre-Covid (CNBC)

• China’s Bioeconomy Is Brewing Its Biggest Wave Yet (Cam Watson)

• NutraCast: How biotechnology is powering the lifemaxxing movement (NutraIngredients)

Don’t miss your chance to attend the Asia-Pacific Agri-Food Innovation Summit and get 10% OFF with my network code. Use my exclusive partner code for 10% OFF your pass: BETTERBIO10

Learn more

Better conversations

The sharpest thinking in agrifood tech often happens off the record. My interview series brings it on the record:

• How a $2B Agtech Unicorn Builds Product - Halter’s Toby Hurley

• Ag and Climate Tech Don’t Have a Capital Problem, They Have a Capital Stack Problem - Renaissance Philanthropy’s Joshua Elliott

• Why 2026 Is the Great Shakeout Year for Food and Ag Tech - EcoTech Capital’s Adam Bergman

• Agtech’s Real Bottleneck Is the Translational Layer, Not the Technology - Beanstalk AgTech’s Justin Ahmed

• Engineering the Exit: How to Get Acquired in Deep Tech - SOSV’s Cyril Ebersweiler

Browse the full archive. More conversations dropping soon. Stay tuned!

Thanks for reading!

If you found value in this newsletter, consider sharing it with a friend who might benefit! Or, if someone forwarded this to you, consider subscribing.

Share

Subscribe now

Disclaimer: The views and opinions expressed in this newsletter are my own and do not necessarily reflect those of my employer, affiliates, or any organisations I am associated with.

For Issue #153 of Better Bioeconomy, I sat down with Toby Hurley, Director of Product at Halter, one of the most talked-about companies in agtech right now.

If you follow this space at all, you have heard of Halter, and lately for one reason. In March this year the New Zealand company raised a US$220 million Series E led by Peter Thiel’s Founders Fund at a US$2 billion valuation. That roughly doubled its valuation in nine months, after a US$100 million Series D in June 2025 at around US$1 billion.

It is one of the largest agtech rounds ever, landing while the sector’s funding has collapsed from its highs. For an APAC-born company in a category most people wrote off as niche a few years ago, that is a remarkable run.

For anyone who needs the primer: Halter builds solar-powered, GPS-enabled smart collars that let farmers virtually fence and shift cattle from a smartphone, no physical fences required. Founded in 2016 by Craig Piggott, a former Rocket Lab engineer who grew up on a Waikato dairy farm, it now serves more than 3,500 farmers and ranchers across New Zealand, Australia, and the United States, and has sold over a million collars.

Toby is the person who runs product on top of all of it. He joined in 2021 from corporate litigation, with no background in product, agtech, or farming, and today leads an org of more than 100 engineers and designers. I came to this conversation from a technical background of my own, curious about how a product like this actually gets built, the thinking behind each upgrade, and how the pipeline holds together as it moves from one market to the next.

In our chat, we spoke about what carries across very different markets, when to pile features on and when to strip them back, why owning the data matters and owning the execution layer matters more, and the whole-farm model he is chasing next.

Let’s jump in!

Underneath the collar, Halter is four systems working as one

Before the product decisions make sense, it is important to understand what Halter really is, because it is more than a collar. A team of engineers and designers builds it as four parts that work as one system: the collar, the connection, the app, and the intelligence underneath.

The collar is the foundation. Solar-powered, weather-sealed, and built to sit on an animal year-round, it carries two speakers that play directional audio and a vibration that confirms when the animal is moving the right way. That is how a farmer holds cattle inside a virtual boundary and shifts them on command, and most learn to respond within three interactions with a virtual fence.

The connection is where Halter recently pulled ahead. Collars used to rely on towers on the ranch, fine on a compact dairy farm but expensive or impossible across remote rangeland. In 2026, Halter launched direct-to-satellite collars, the first virtual fencing that works anywhere you can see the sky. The collars form a mesh network, and every few minutes one takes a turn as leader, connecting to satellite to pull down the latest fence lines and push up the herd’s location and health data before sharing it back across the rest.

The app is the control panel: a live digital twin of the ranch that shows where every animal is and how each paddock is grazing, and lets the farmer draw a fence or move a mob with a tap. Set a shift in the app, and minutes later, the cattle move in the real world.

The fourth part is the one you cannot see. Every collar streams over 6,000 data points a minute to Halter’s cloud, feeding what the company says is now more than 7 billion hours of animal behaviour data. Machine learning turns that into the cues, health alerts, and grazing insights that come back to the farmer. At the centre sits the Cowgorithm, Halter’s trademarked model that teaches cattle to respond to the audio cues in the first place. It is the main reason the virtual fence works at all.

How coming in with no farming background was an advantage

Toby joined Halter from an unlikely place: law. He had been a corporate litigator at one of Auckland’s big firms, and he treats that outsider’s start as one of his biggest advantages.

He loved litigation for the hard problems and the grind, but could not see the path: ten years to make partner, then a job that is mostly client relationships rather than the work itself. So he started looking. His path crossed with Craig, who was looking to hire Halter’s first product manager as the company moved from “build the collar, does it work?” to taking a product to market. He started as an associate product manager and grew into the role he holds now.

His first six months were spent as a sponge, talking to farmers and working out what they did and why. Then his legal training took over: find the most important problems, work out why they are the most important, and find creative ways to solve them.

The advantage was having no attachment to how things were supposed to be done. "I had never used any competitor products. I wasn't familiar with what normal looked like," he told me. That let him ask a sharper question than most insiders would: what is actually worth reinventing? Some parts of farming rest on decades of science and do not need tearing down. Others do. Coming in fresh made it easier to tell the two apart.

Most of the product travels across regions since the ingredients don’t change

New Zealand dairy and US beef ranching look nothing alike on the surface. Different herd sizes, different terrain, different economics. How much of the product needs to be rebuilt each time it enters a new market? Toby’s answer was to look past the surface, to what every one of these operations has in common.

Every Halter customer is working with the same ones: land, grass, and cattle. The land grows grass, the cows eat it, and the farmer’s job is to turn that grass into a product. A New Zealand dairy farm, a Texas ranch, and a Canadian or Brazilian operation are all solving that single problem.

Halter built its foundations around it. The collar is identical in every market, gives the same cues, and pulls the same data, and the machine learning models are built to transfer across regions rather than be rebuilt for each one.

What localises is the layer on top: the tooling, and how the mobile and web apps present things to a given customer. “For the most part, it’s really the translation layer that changes,” Toby said, not the hardware or the core systems underneath. Craig has described the same split, saying that the team reworked the software to fit how US ranchers work and pushes updates to the collars remotely, so the system keeps improving without anyone buying new equipment.

A foundational system may need to change for a new market in the future, but so far that has not happened. Every market has run on the same foundations. For a company entering several countries at once, that is a real head start, because each new market stands on something already proven rather than a blank page.

Subscribe now

When software is cheap, owning the data matters, and owning the execution layer matters more

The collar does a lot of the work. The other half of what makes Halter hard to compete with is the software and the data beneath it, and Toby is clear-eyed about why. Building software has become cheap, and so has AI. Pay a hundred dollars for a Claude subscription, he points out, and you can do powerful things without being a developer. When the tools are that commoditised, the edge moves to the one thing that is hard to copy: access to the data that feeds them.

Halter pulls that data from everywhere, on and off the collar. For pasture, it layers weather data, satellite imagery, and now drone data on top of what the collars gather. As Craig told TechCrunch, the health and fertility detection have sharpened as Halter has built up what is likely the world’s largest dataset of cattle behaviour.

The part Toby pushed back on was the shallow version of the data story. People picture the collar as a Fitbit for cows, a wearable that just collects data, and forget it is also a fencing and shifting device. "If you build the most unbelievably accurate optimised plan for how you should run your farm," he said, "if you can't actually go out and execute that with precision every day, then it doesn't really materialise into an outcome."

The collar collects the data and acts on it. An AirTag tells you where the cow is. A Halter collar moves her. That execution layer is what a competitor running pure analytics cannot easily replicate, and as one investor told AgFunder, the collar is only the entry point. Beef Pro is the newest thing built on top of it. Halter launched it in the US and Australia just after we spoke, and as Toby put it, it “adds the measurement and planning layer on top of the execution layer that is virtual fencing.”

‘You earn the right to simplify’ by shipping fast, then stripping back

Halter ships constantly. Craig has said the product a rancher uses today is radically different from what they bought a year ago, with new things going out every week. So how does the team keep adding capability without bloating the product away from the job a farmer bought it for? "We think of everything from a first principles perspective," Toby told me.

The north star is the short list of jobs a pasture operation has to get right to turn grass into product, and that rules most things out. Halter does not chase nice-to-haves that save a little time but do nothing for whether the business is more profitable or more sustainable.

Toby thinks about simplicity differently from most product people. The instinct is to chase a clean, elegant product from day one, and he thinks that is a trap, because you never learn what was too much or what did not need to be there. “You very much earn the right to simplify,” he said.

The approach is to nail an important problem, then ship, test, and validate as fast as possible. In the short term that can mean several things running at once, which can look like bloat, but what it is really doing is generating signal about what works. Only once you can see the full picture do you strip back what does not matter. More iterations, simpler product.

The testing is risk-tiered. Low-stakes changes, like showing data as a graph instead of a table, get shipped and tested liberally. Core systems, like measuring grass or a cow's health, run through a disciplined process on Halter's own research farm before they reach customers. Toby counts himself lucky to have beta and closed-tester groups across countries who are happy to try things and say what works, which builds confidence before any wider release.

Hardware and software get you in the door, trust is what makes the product stick

Halter is not a fully self-serve product, and may never be. Toby is at peace with that.

Every product team wants customers who buy and onboard themselves with no support, because that is what margins and scale reward. Halter is asking for something harder. Physical devices go onto the land. The humans learn to run their operation through a phone. The cows have to learn entirely new cues. That is a rethink of how a farm runs, and it does not happen unattended.

According to Craig, this is the deal: Halter is not a set-and-forget product. “You need to be willing to use the product and change what you were doing to some degree,” he told AgFunder. “If you really want the returns, you have to be willing to do some level of system change in how you allocate your land.”

So Halter goes hands-on at the start, with a dedicated team sitting alongside customers through onboarding. Toby’s read is that confident, happy farmers and happy cows are the single biggest predictor of long-term success, so that early stage is where the effort goes.

The support will shift as the company scales, but he expects the trust-building part to matter indefinitely. The hardware and software get Halter in the door. Trust is what keeps it there.

Modelling the whole farm would be the “quantum leap”

Given everything Halter has already built, I asked what he was most excited to build and what the hardest unsolved problem was. He gave one answer to both.

It comes back to the same equation: land, grass, cattle, product. The difficulty is that you are dealing with several systems interacting at once: the land and grass, the weather, the cows, and the human running it all.

The end state is a model of that entire system. Change one variable today, and Halter tells you what happens in a month, six months, a year, with economics and market pricing overlaid. “You’re not going to build that all at once,” he said, “but that is kind of the end state that I’m most excited about us building.”

It is also the least solved problem he has. If you could predict the weather six months out, know to the calorie how much energy a cow needs, and know to the calorie how much energy is in the grass, this would all be easy. But you cannot. That uncertainty is the hard part.

Halter’s bet is that more data and more investment in its systems close the gap over time. Everything it has built so far has been a big leap forward. What comes next, Toby said, is a “quantum leap” forward.

If you found this useful, consider sharing it with a founder or operator building at the messy seam where hardware, software, and biology meet.

Share

Or, if someone forwarded this to you, consider subscribing.

Subscribe now

Disclaimer: The views and opinions expressed in this newsletter are my own and do not reflect those of my employer, affiliates, or any organisations I am associated with.

This week:

• Stenon raised €18M Series B to scale its real-time nitrogen and soil data platform

• Parima produced tonne-scale cultivated duck on Vow’s 22,000-litre line at a 99% lower cost

• ProFuse raised $1.5M to turn its muscle-tissue platform into a discovery engine for the GLP-1 era

• The Protein Brewery added €18M to its Series B to scale Fermotein and launch in the EU

• Omni raised £11M to grow its vegan dog food and launch an Ozempic-style weight-loss supplement for dogs

• Tropic acquired Rahan Meristem to become a fully integrated banana genetics company

• And more…

Also, I joined Rhishi Pethe on his Software is Feeding the World newsletter to talk about what it takes to scale the bioeconomy, why Asia may decide its future, and how my read on agrifood tech has moved from a technology problem to an ecosystem problem over time. Check it out!

PS: If you are an investor or working at a corporate in food, agriculture, or biomanufacturing, do reach out. I would love to exchange notes on what I am seeing across the sectors. Reply to this email or drop me a message.

Below are the 10 most interesting ag and food tech stories I came across last week, paired with my thoughts on what they mean for where the industry is heading.

🦆 Parima proves tonne-scale cultivated duck at viable cost on Vow’s 22,000-litre line in Australia

• France’s Parima, the parent of Gourmey and Vital Meat, produced its cell-cultured duck at tonne-scale volume in a single run on Australian partner Vow’s 22,000-litre bioreactor, the largest food-grade cell culture vessel globally. The two companies began collaborating in 2025, and this first run took place this year.

• Parima says it hit the milestone at a 99% lower cost than earlier runs. The startup credits successive gains in cell-culture yield, the removal of the most expensive inputs such as growth factors, albumin, and insulin, and the volume efficiencies of operating at scale.

• The process ran at 22,000 litres on the first attempt with no performance loss versus smaller scales, according to Parima. The company frames the result as clearing the biological performance, process efficiency, and media-cost hurdles that have drawn scepticism. It says viable unit economics for cultivated meat are now proven rather than extrapolated from pilot runs.

Source: Green Queen

Thoughts 🤔

This is a big milestone for the sector. The line that I found most interesting is the ‘first attempt’ bit. Parima transferred its process onto Vow’s 22,000-litre line and hit tonne-scale with no performance loss versus bench, in a single run.

Tech transfer is one of the key areas where cultivated meat has historically struggled, because what works at 2,000L often falls apart an order of magnitude up. Clearing that gap on the first go is a real de-risking marker.

What also makes it interesting structurally is the split between biology and steel. Parima keeps control of cell lines, media, and process, and rents capacity from Vow, which has pivoted into a co-manufacturer with very large, food-grade reactors. That looks a lot like the division of labour that emerged in fermentation and pharma: specialised innovators on the biology side plugging into shared manufacturing platforms instead of owning the whole stack.

In a funding environment where few cultivated meat companies can finance end-to-end infrastructure, this model is a plausible path for the category to industrialise without over-building balance sheets.

🌱 Stenon raised €18M Series B to scale real-time nitrogen and soil data management platform as farmers face fertiliser price swings

• The German company makes FarmLab, a portable soil-analysis device that pairs optical and electrical sensors with AI. It maps thousands of data points in seconds, including plant-available nitrogen, soil organic matter, temperature, and moisture. The software turns that into fertiliser guidance growers can act on immediately in the field, rather than waiting on lab results.

• Stenon positions plant-level nitrogen measurement in the field as its core edge, allowing more frequent readings that account for soil variability within a single field. It claims 2-8% yield increases across corn, beans, cotton, sugarcane, coffee, and grains, plus a 20-40% average ROI on nitrogen fertiliser.

• Stenon says the platform already runs across several million acres in Europe, Central Asia, Brazil, and the US. The round funds deeper expansion in existing markets, with Brazil a priority through a longstanding partnership with input retailer Lavoro Ag. Part of the capital goes to a new product (no specific details provided), developed over several years and set for release later this year.

Investors: Pymwymic, DeepTech & Climate Fonds (DTCF), Atlantic, Oyster Bay, Founders Fund, TIME Ventures, and Bernd Hoffmann (former VP at AGCO and Claas KGaA)

Source: AgFunder

Thoughts 🤔

I see FarmLab’s defensibility sitting downstream of the device, in the models that convert its optical and impedance readings into usable soil chemistry signals. Those models are trained against lab data across diverse soil types, and both Stenon and independent studies flag calibration as a key driver of accuracy and a current limitation when it is not well tuned.

Spectral and impedance-based soil sensing have a solid track record in controlled settings but often struggle as you move into new field conditions, especially when models are tuned for one soil class and then applied to very different ones. That is why the “several million acres” is important: if Stenon can systematically fold that breadth of data back into calibration, each new region increases the robustness of its models.

Each new soil environment is both a barrier and a potential asset. It takes real work to recalibrate for Brazilian oxisols, US prairie soils, or Central Asian conditions, but once Stenon has done that work, any competitor faces a similar calibration cost before they can match field performance. The hardware is replicable, but the loop from local data to proven accuracy is where defensibility stacks up.

💪 ProFuse raised $1.5M to advance its muscle health discovery platform for the GLP-1 era

• The startup, known for media supplements that optimise muscle tissue formation in cultivated meat, is repositioning around a high-throughput platform screening supplements and therapeutics for muscle atrophy. Its lab-grown human skeletal muscle tissue screens 10,000+ molecules a week. Screening began in early 2026 and has already surfaced supplement combinations and early drug candidates.

• It grows mature human muscle fibres in 3D between pillars, which it claims are more biologically relevant than the mouse cell lines and 2D systems drug companies typically use. Engineered reporters track muscle maturation, stress, wasting, and recovery in real time. Electronic pulses measure contractile force, testing whether growth translates into function, not just size.

• The model runs on two tracks. On supplements, ProFuse is screening hundreds of GRAS molecules for food and nutrition partners that build muscle-preserving products for GLP-1 users. On pharma, it plans to develop candidates to the lead stage before handing off via R&D payments, milestones, and royalties.

Investors: Green Circle Foodtech Ventures, Siddhi Capital, Tempo Beverages

Source: AgFunderNews

Thoughts 🤔

The capability I find most interesting is that it measures muscle function at the discovery stage, not just growth. It grows human fibres in 3D, stimulates them with electric pulses, and reads how much contractile force they produce.

That lands on the open question hanging over the GLP‑1 muscle field. Many of the heavyweight muscle programmes, including work from Lilly, Regeneron and Scholar Rock, focus on the myostatin and activin/TGF‑β pathways, and a persistent doubt is whether the mass they add is truly functional muscle or mostly volume.

In a lot of these programmes, functional strength and quality‑of‑life endpoints are secondary and not always powered for clear readouts, and the most convincing gains tend to appear in patients starting with substantial deficits. As projects move from early screening into animal models, measuring muscle force with enough resolution becomes more complex and resource‑intensive than just tracking mass.

🍄 The Protein Brewery raised €18M Series B extension to scale its Fermotein mycoprotein and launch in the EU

• The Dutch fermentation company makes Fermotein, a minimally processed mycelium ingredient that packs protein, fibre, and bioactives into a single building block. The Series B extension lands days after Fermotein became the first whole-food mycelium protein to secure EU novel food approval and follows the €30M Series B closed in September 2025.

• The startup has sold out its entire 2026 capacity to US clients and expects its Breda facility to produce 600 tonnes in 2027 against confirmed commitments across the US, Singapore, and Europe. The new capital funds extra fermentation capacity, downstream de-bottlenecking, and hiring, targeting output beyond 2,000 tonnes by 2029.

• The EU push centres on active nutrition, with first customers expected to launch in Q4 2026 using Fermotein in nutrient-dense, GLP-1-compliant all-in-one shakes. Funds will also pursue regulatory clearance in Canada, Australia, New Zealand, and India, and advance clinical research into the ingredient’s longevity benefits.

Investors: ABN AMRO Sustainable Impact Fund, Invest-NL, Novo Holdings, Madeli, and the Brabant Development Agency

Source: Green Queen

🐶 Omni raised £11M Series A to expand its vegan dog food and launch an Ozempic-style weight-loss supplement for dogs

• The UK startup makes vet-formulated plant-based dog food and supplements, spanning dry foods, training treats, dental sticks, multivitamins, and pre- and probiotics, catered to pets with allergies, anxiety, and gut issues. The company says it has grown revenue 10-fold over 12 months, from £1M to close to £13M in annualised sales, and helped more than 300,000 dogs.

• Part of the raise will fund LeanPaws, a GLP-1-style weight-loss supplement teased for a 2026 launch. The company says it uses fibres and resistant starches that mirror Ozempic-like effects, plus prebiotics and probiotics aimed at fat metabolism. An estimated 1-8% of British dogs suffer from some form of allergy, a gap Omni is also targeting with novel allergen-free proteins.

• The funds will push Omni into major UK supermarkets, test the US market, expand the team, and fund clinical research to validate its products. The raise lands amid a wider boom in low-carbon pet food across Europe, as owners increasingly weigh the climate footprint of conventional meat-based diets.

Investors: IW Capital, Redrice Ventures, RootBridge Capital, Digitalis Ventures, Lever VC, Deborah Meaden

Source: Green Queen

Thoughts 🤔

Omni grew revenue 10-fold by selling vet-formulated health outcomes: gut health, allergen management, weight, skin. The recipes happen to be plant-based, but the pitch is what the food does for the dog, and that framing is what gets a product onto a supermarket shelf and reaches owners beyond the plant-based base.

LeanPaws takes that further. A weight-management supplement that mirrors GLP-1 effects, with a placebo-controlled trial behind it, attaches the most talked-about human health trend of the moment to a daily dog product. Canine obesity is large and underserved, and the Ozempic wave has done the consumer education for free, so owners already grasp what a satiety product is meant to do.

It’s a good example of how plant-based pet food goes mainstream: lead with the dog’s health, let the sustainability sit underneath. The fit between a real health need and a cultural moment that needs no explaining is hard to argue with.

🍌 Tropic acquired global plant propagation company Rahan Meristem to become a fully integrated banana genetics company

• Tropic develops improved banana varieties through gene-editing and breeding, keeping the Cavendish’s agronomic and consumer qualities while adding disease resistance, reduced browning, longer shelf life, and higher yields. Rahan Meristem, with operations across Latin America, Asia, and the Middle East, gives it large-scale production and distribution across the major banana-exporting regions.

• The deal makes Tropic the first company to span variety development, propagation, and grower supply in one platform. Rahan’s existing grower relationships hand it direct routes to market, pairing its gene-editing pipeline with proven propagation capacity at commercial scale.

• Bananas are one of the world’s most exported crops at over 20M tonnes annually, yet production relies almost entirely on the Cavendish, leaving growers exposed to Panama disease TR4. Following its recent $105M Series C and the first commercially viable new varieties in over 75 years, Tropic is now scaling global production and grower adoption.

Source: Tropic

Thoughts 🤔

Tropic’s first two banana varieties launched in 2025, and the company says demand already exceeds supply. That points to where a real constraint sits: scaling clean planting material. Commercial Cavendish are seedless and propagated vegetatively, so a new variety starts from zero plant stock and must be multiplied through tissue culture before it can occupy meaningful acreage.

So the only way to scale at commercial speed is laboratory micropropagation. The edit can be finished and agronomically proven, but the variety scales no faster than tissue-culture labs can produce disease-free plantlets and channel them to growers. Even with a $105M Series C, building that footprint across Latin America, Asia, and the Middle East from scratch would take years.

So buying Rahan removes the dominant bottleneck on Tropic’s early revenue, handing over ready-made propagation capacity, mother plantations, agronomy know-how, and grower access in the regions that matter for export bananas.

🦠 Myota raised $4.5M Series A to scale its prebiotic fibre blends and build a B2B ingredients business

• The London-based startup makes patented multi-fibre blends designed to improve gut and metabolic health. Its UK D2C business tripled last year to over 60,000 customers. The round funds a dedicated B2B sales operation across Europe and the US, a faster D2C scale-up, and further clinical research.

• The startup says most fibre supplements lean on 1 or 2 commodity fibres, and the same fibre behaves differently in each person because individual microbiomes ferment it into varying amounts of short-chain fatty acids. Myota claims its blends work across diverse microbiomes to maximise short-chain fatty acid output while cutting the GI discomfort common to high-fibre products.

• The startup positions the platform as a rare combination: clinical-trial efficacy plus usability at meaningful doses in everyday food and drink. Its gut blend cut bloating and diarrhoea in 80% of users over 4 weeks, and its metabolic blend improved insulin sensitivity in 61%. It already supplies Joe & The Juice and a coffee alternative brand, with dairy and bakery products in development.

Investors: PeakBridge

Source: Green Queen

Thoughts 🤔

Myota sells commodity fibre. The guar gum, oat and wheat fibre, inulin, FOS and GOS in its blends are inputs anyone can buy, none of them patentable. What it has built is a randomised-controlled-trial dataset showing the blends move insulin sensitivity, blood sugar, inflammation and mood, plus the formulation logic to make that work across varied microbiomes.

That distinction matters more now than it would have two years ago. The GLP-1 wave has pulled every fibre supplier toward the same pitch, and “GLP-1-friendly” claims have gone up. When the input is undifferentiated and the tailwind is shared, the one thing a B2B buyer cannot source more cheaply elsewhere is a defensible, claimable health outcome. PeakBridge seems to be betting on that layer, the same evidence-first logic it used backing clinically validated functional mushrooms.

My read is the durable value sits in the trial data and the claims it unlocks, not the blend. If the RCTs hold across cohorts and the claims survive regulatory scrutiny in each market, evidence is the moat.

🫚 Herbalife and IIT Madras open a plant cell fermentation centre to scale biomanufacturing of herbal biomass and phytochemicals

• Billed as India’s first facility dedicated to translational research in this category, the Chennai centre takes lab discoveries to pilot-scale production using customised plant cell bioreactors, aimed at startups, entrepreneurs, and industry researchers.

• The Chennai hub uses plant cell fermentation to produce herbal biomass, enriched extracts, and high-value phytochemicals at scale, pairing upstream bioreactor cultivation with downstream processing and metabolomics platforms.

• The partners frame the centre as a move to cut import dependence through domestic, sustainable production, generate IP, and build a skilled biomanufacturing workforce, reinforcing India’s wider push to become a global biomanufacturing hub.

Source: Green Queen

🧬 ChemT Biotechnology raised $4M seed round to build the intelligence layer for biomanufacturing

• The Singapore-based startup builds AI directly into the biomanufacturing process rather than into drug discovery or lab automation. Its goal is to fix the long timelines, unpredictable yields, and high costs that slow biologics production, all of which trace back to the lack of precise, tunable ways to control cells.

• At the core is CelMo, a Virtual Cell platform trained on billions to trillions of proprietary sequencing reads and lab-validated. It simulates how cells respond to manufacturing conditions, genetic changes, and stress, then maps which biological processes are switched on, suppressed, or rewired.

• In CHO cells, ChemT says the platform delivered a 50% increase in antibody output and a 40% cut in production timelines. Flagship small-molecule Chemplify applies the same method to T-cell manufacturing for cancer therapies, with the company claiming 50% faster development, 3x scalability, 60% lower costs, and 10x higher cell expansion yield.

Investors: Wavemaker Ventures, SEEDS (SG Growth Capital), Draper University Ventures, Temasek Life Sciences Accelerator

Source: PR Newswire

Thoughts 🤔

Most of the money going into AI virtual cells right now points at drug discovery. They are all trying to predict how a cell responds to a perturbation so you can find a molecule that works in a patient. ChemT trains the same class of model on a narrower target: how cells behave inside a bioreactor, and what raises titre or shortens a cell run.

That manufacturing‑first framing may prove more defensible because of the data it can lean on. ChemT is training CelMo on process datasets from real or representative production runs, not just on omics readouts tied to preclinical or clinical response. For a buyer, the metric is straightforward (yield per batch, run time, and cost per dose) and the feedback loop is a fermentation or cell‑therapy manufacturing run rather than a multi‑year clinical trial.

By contrast, discovery‑facing virtual cells are fighting for advantage against heavily funded discovery platforms and pharma R&D groups on a long‑cycle question: whether a candidate works safely in humans. Even with good modeling, proving that edge can take years, and the link from virtual cell predictions to clinical benefit is harder to make concrete for buyers.

🧫 Jellatech partners with South Korea’s Hugel to develop and commercialise cell-based human collagen

• The US-based startup’s deal with Hugel gives the Korean medical aesthetics firm exclusive global rights to its bioidentical human collagen for select aesthetic uses, marking Jellatech’s first major commercial milestone. The collaboration centres on next-generation collagen injectables and cosmetics, starting with skin-boosting applications.

• The platform produces full-length, triple-helical human collagen from cells cultured in bioreactors, bypassing animal sources, peptides, and non-native recombinant methods. Its portfolio spans a Type I atelocollagen (telopeptides removed for lower immunogenicity) and a Type I telocollagen (telopeptides intact for stronger fibrillogenesis), giving formulators complementary aesthetic and regenerative options.

• Conventional collagen comes mostly from cattle, pigs, and fish, with cattle farming tied to Amazon deforestation, and animal-derived versions struggle to replicate native human collagen’s structure, a gap Jellatech’s platform is built to close.

Source: Green Queen

CAPITAL & CONVICTION

The sharpest thinking in agrifood tech often happens off the record. My interview series brings it on the record:

• Ag and Climate Tech Don’t Have a Capital Problem, They Have a Capital Stack Problem - Renaissance Philanthropy’s Joshua Elliott

• Why 2026 Is the Great Shakeout Year for Food and Ag Tech - EcoTech Capital’s Adam Bergman

• Agtech’s Real Bottleneck Is the Translational Layer, Not the Technology - Beanstalk AgTech’s Justin Ahmed

• What It Takes to Build a Fundable Food Company After “Peak Stupid” - Siddhi Capital’s Steven Finn

• Engineering the Exit: How to Get Acquired in Deep Tech - SOSV’s Cyril Ebersweiler

Browse the full archive. More conversations dropping soon. Stay tuned!

Thanks for reading!

If you found value in this newsletter, consider sharing it with a friend who might benefit! Or, if someone forwarded this to you, consider subscribing.

Share

Subscribe now

Disclaimer: The views and opinions expressed in this newsletter are my own and do not necessarily reflect those of my employer, affiliates, or any organisations I am associated with.

Last week, I sat down with Joshua Elliott, Chief Scientist at Renaissance Philanthropy and Director of its Advanced Research for Climate Emergencies (ARC) initiative.

This is the first time I have spoken to someone from the philanthropic side for the newsletter. Lately, I have been spending more time learning about the capital stack and the different funding models that support food, agriculture, and biomanufacturing.

As a VC, I think about where venture is the right fit for these sectors and where it is not. Venture is only one layer of a bigger stack. So I went into this conversation curious about the layers I understand least.

Renaissance Philanthropy is a nonprofit that designs, incubates, and runs time-bound, thesis-driven funds led by domain experts, with a stated goal of building the equivalent of venture capital for public good. The model is moving quickly. The organisation has mobilised $533m in the first two years, $268m of this directly for their own programs and funds, with $265m unlocked for other organisations. A meaningful part of that work is in agriculture and climate.

Joshua has watched the funding ecosystem from different angles. He took a PhD in theoretical particle physics, then spent a decade doing computational climate economics and crop modelling at the University of Chicago and Argonne National Lab, where he co-founded a centre for climate and energy policy and started the Global Gridded Crop Modeling Intercomparison Project.

Along the way, he ran his own ag-focused climate informatics startup, Praedictus Climate Solutions, as CEO. He then spent six years at DARPA as a programme manager, where he programmed close to $600 million in federal R&D, before a stint in climate philanthropy and his current role at Renaissance Philanthropy. He has been a startup founder, an academic grantee, a government funder, and now a philanthropic investor.

In our chat, we spoke about what that full-stack vantage point teaches him about how climate and ag tech get funded, when philanthropic capital is the right tool for a founder, how Renaissance Philanthropy turns philanthropic dollars into commercial companies, and what the capital stack could look like a decade from now.

Let’s jump in!

The sector has enough capital, but it just refuses to work together

Joshua’s organising idea is that climate and ag tech do not suffer from a shortage of money. They suffer from money that operates in silos. Venture, government, philanthropy, and project finance each optimise for their own logic, and nobody coordinates them around the path a single technology has to travel from lab to commercial scale.

A venture investor sprays and prays, a Series A fund only looks at Series A, and philanthropy, Joshua says, is sometimes the worst offender. “I’m a philanthropic investor, so all I think about is philanthropy. We silo all these forms of capital. They don’t work together.” The fix is to treat the whole stack as one coordinated toolkit instead of four silos optimised in isolation.

This goes wrong most often when the sector reaches for venture capital too early, for technologies that venture capital was never built to carry. Venture economics run on a low hit rate and a few massive winners, which is rational for a fund but brutal for a technology that just needs more time.

The casualties are not the bad technologies. “There’s a lot of potentially promising technologies that end up getting abandoned not because they don’t have promise,” he told me, “but because they couldn’t scale fast enough.” Joshua is the first to admit he does not have a tidy replacement model.

What he does propose is simpler: bring in patient capital earlier, and you give a good technology time to mature before the venture clock starts running against it.

Make the green thing the cheap thing, not the subsidised thing

Joshua calls himself a philanthro-capitalist, especially in climate, because for most climate tech, the road to real impact runs through scalable, profitable, market-rate products. He has watched the best version of this up close, working with family offices that run philanthropy, impact investing, and market-rate investing under one roof.

What he learned from them is the whole model in miniature: the same backers walk a single technology across all three pockets, using philanthropy to push it to the point where the impact investing team will fund it, then accelerating it until the market-rate team will. Each handoff is the proof. Philanthropy’s job is to hand the next investor down the line something they can back.

His worked example is cement, because it shows philanthropy doing something venture cannot. At the factory gate, he argues, the problem is (at least in part) already solved. There are cement products that are cost-competitive, 50% lower carbon, stronger, and longer-lasting than the traditional kind.

But they are not everywhere because of the pile of frictions sitting downstream of the factory: missing building standards, transport departments with no training to specify the new product, higher insurance, and higher capital costs.

Each adds an implicit cost that the technology itself does not. So Renaissance Philanthropy built a program that addresses all of those frictions at once. “We’ve created a systemic approach to look at all of those little downstream things that need to be solved, and then solving them all together and in tandem,” he said. That is field-building, the full-stack systems work Joshua and his co-authors lay out in their Philanthropy 2.0 essay, where one strategy spans science, policy, market-shaping, and adoption instead of stopping at the lab.

The goal underneath all of it is to make the clean option the cheap option, because that is the only version that lasts. Lean on a green premium and a subsidy, and the industry crumbles the day the policy changes.

He points to solar as proof that the durable path is cost, not subsidy. It became the world’s cheapest and fastest-growing source of energy, he points out, not because anyone was paying a green premium, but because four decades of work finally drove the price below fossil fuels.

The same logic should apply everywhere else. The way to win is to make the next cement plant, steel plant, or geothermal build cheaper as a low-carbon project than as a traditional one, so the economics make the choice for you.

The problem is time. “We cannot afford for all these new technologies to take the same four decades that solar took to move down that cost curve,” he said. That impatience is the whole reason for using philanthropic capital to push a technology down the cost curve faster, rather than waiting for the market to get there on its own.

Subscribe now

For startups, philanthropic capital fits two specific moments

Philanthropic capital is rarely the first source of funding a founder thinks of, so when does it fit?

The first is at pre-commercial proof. If you have a real result but are three to five years from commercialisation, the Focused Research Organization is a strong option: a time-bound nonprofit startup funded in the tens of millions to prove one big thing, rather than the few million an academic lab scrapes from NSF (US National Science Foundation) grants.

He pushed back on the fear that founders have that taking philanthropic money means surrendering their moat. People assume open IP will ruin them, “whereas what they find later is that a large fraction of their moat is based on the tacit knowledge they’ve acquired along the way, not the specific admixtures or blueprints.” The know-how in the team’s heads is usually the more defensible part.

He also pointed founders to a growing set of government capital sources that, especially in the US, are far friendlier to commercialising IP than philanthropy is and are worth weighing alongside it.

He also flagged a related trap founders underrate: scaling is not just an engineering risk. “There’s actually a massive amount of new science risk that emerges in the process of trying to scale,” he said, “particularly in biology.” That is the risk patient, pre-commercial capital exists to absorb.

The second moment when philanthropic capital comes in is later, at the first demonstration plant, where the hurdle is the cost of money. Ask for capital to build the 200th traditional cement plant in a region, and lenders barely blink. Ask for capital to build the 3rd low-carbon plant, and you pay a premium, simply because the technology is less familiar.

Build in Africa or India, and the premium gets worse. “You can’t pay 16% interest rates on the capital you need for your demonstration plant,” he said, and most of the time, the math just collapses.

This is where a small amount of philanthropic capital can move a large number. By guaranteeing part of a loan, philanthropy absorbs some of the lender’s risk, which lets the lender drop the interest rate to something the project can carry. The capital is not spent, it is pledged as a backstop, and most of the time, it is never called.

The smart version of this taps money that is already sitting still. Joshua points to the roughly $250 to $300 billion parked in US donor-advised funds, charitable money earning 5 to 7% while it waits to be given away, a pool he expects to grow toward a trillion dollars as new AI wealth lands.

Renaissance Philanthropy runs a program, Tertiary Impact Capital, that leaves those funds untouched and simply uses them as the guarantee behind clean-tech loans. The donor’s money keeps earning its return, and the same dollars lower the cost of capital for a demonstration plant. “If you package things right,” he said, “you can get a pretty attractive set of investments and make a massive impact with a relatively modest amount of capital expenditure.”

When should a breakthrough become a for-profit company versus a public good?

Joshua is using philanthropic capital to spin out for-profit companies, which sounds like it should contradict the public-good rule. It does not, because the structure flexes to the situation. He has at least two live deals, each built differently on purpose.

The first is the partnership with Deep Science Ventures, the UK deeptech venture creator, with whom Renaissance Philanthropy built the Climate Emergencies Resilience Lab. In March, the two unveiled a venture creation project aimed at priming resilience in crops, a category meant to move past slow, static genetic modification toward resilience that can be tuned to climate risk.

Here, Renaissance Philanthropy takes equity at a fair market return. “Someday in the future, that equity will hopefully scale and be worth money, and then it can be returned to its philanthropic accounts, where it can be used to do even more philanthropic good,” he said. The company keeps its IP, because the philanthropy owns a fair share of the company.

The second, a pre-seed methane abatement startup, is structured as a concessional loan, and the asymmetry is the whole design. “If they succeed and scale, they will pay the capital back. If they don’t, then we made a philanthropic grant to something that didn’t produce a scalable for-profit company, so it’s fine.”

The downside is a grant, and the upside recycles into the next bet. The decision about whether something becomes a company follows the mission and the structure that gets the technology built.

The hard part is convincing people to deploy philanthropic capital

Two years in, the thing that has turned out to be harder than Joshua expected reframes the whole sector’s narrative away from a money shortage.

The hard part, he said, is convincing people to deploy philanthropic capital at all. “You look at the amount of capital that’s out there, and you think, oh, this can’t be that hard to do fundraising,” he said. “But it is, in fact, a lot harder than you think.”

Foundations pledge to spend $10 billion over a decade and are only three billion in seven years later. Philanthropy remains a network-driven, trust-based process, and for good reason: there is no single metric of quality the way there is in for-profit investing. That missing metric is, for Joshua, the root structural problem, and it is the argument he and Renaissance Philanthropy CEO Tom Kalil lay out.

For-profit investing specialised dramatically after the 1950s, from generalist banks into venture, then hundreds of fund types. But philanthropy did not do that. “We still use basically the model that Rockefeller developed in the 1930s,” he said, noting that the Gates Foundation and the NSF have each landed at roughly 2,000 staff giving away roughly $8 billion a year, the same basic architecture under different logos.

Kalil frames the gap as a shortage of institutions, not capital. Joshua’s own prescription follows from it: a GP/LP model, where philanthropists act as limited partners who build a portfolio of expert-led funds rather than vetting every grant themselves. “If philanthropists would think of themselves as more portfolio allocators rather than the ones selecting every single investment,” he said, “we would be much better able to scale philanthropic giving at the level that’s needed.”

Capital that follows the technology all the way to scale

If the Renaissance Philanthropy model works, what does the ag and climate capital stack look like in a decade? Joshua’s view is that philanthropic capital will be working across the whole stack rather than only to the left of the factory.

It will treat the cost of capital as a problem philanthropy is built to solve, especially the over-conservatism of capital in emerging markets and the Global South. It will write the small-ticket cheques that venture, even impact venture, cannot justify, the deals where the opportunity is real in places like Africa, but each one is too expensive to evaluate relative to its size.

Above all, it will fund the part everyone skips. Joshua’s frustration from his academic years was watching project after project build a clever solution, a new drought-tolerant seed or distributed fertiliser system, prove it in one town, declare victory, and walk away. “Without thinking about the incredible amount of work and capital it takes to scale that solution across tens or hundreds of thousands of towns across a region or a continent.”

That is where his critique lands closest to home. The valley of death in food and ag is not only a venture problem. It is a coordination problem across every pool of capital that could help cross it. Closing that gap is not about inventing the next drought-tolerant seed. It is about building capital that sticks around long enough to carry the good ones all the way to scale.

Want to connect with Joshua?

If this approach to philanthropic capital resonates with you, and you are a philanthropist or funder looking to deploy it, Joshua and the team would love to hear from you at info@renphil.org

If you found value in this newsletter, consider sharing it with a friend who might benefit from it!

Share

Or, if someone forwarded this to you, consider subscribing.

Subscribe now

Disclaimer: The views and opinions expressed in this newsletter are my own and do not reflect those of my employer, affiliates, or any organisations I am associated with.

This week, Vivici pulled €12.5M from the EIC to scale its cow-free whey, while Helaina pushed its precision-fermented breast-milk protein into kids’ multivitamins. Rainbow Crops raised €9.7M seed round to scale AI-guided multiplex gene editing, and The Every Company quadrupled its egg-protein capacity on a pharma partner’s idle fermenters.

On the institutional side, A*STAR and NUS launched a synbio joint lab in Singapore, Döhler acquired Nukoko’s fava-bean chocolate, and Anterra closed $100M to rewire the food system rather than reinvent it. I also dig into what agtech can learn from biotech about VC returns.

Below are the 10 most interesting ag and food tech stories I came across last week, paired with my thoughts on what they mean for where the industry is heading.

🥚 The Every Company quadruples precision-fermented egg protein capacity with Huvepharma after 550% hike in orders

• The animal-free egg protein maker has expanded its partnership with pharma manufacturer Huvepharma to scale OvoPro production fourfold, tapping subsidiary Biovet’s 9M-litre fermentation facility in Bulgaria. The precision-fermented egg white protein already features in products at Walmart, Target, and Amazon.

• The startup says it secured annual orders worth 55% of its total 2025 order volume in just the first four months of this year, driven by B2B retail, e-commerce, and foodservice adoption. It had already doubled capacity at the start of 2026 with plans to double again next year, but the surge accelerated that schedule.

• OvoPro is positioned as a hedge against a volatile egg supply chain hit by bird flu and price spikes. The company says it is shelf-stable for up to 24 months, needs no cold chain, and is priced competitively with or cheaper than battery-caged commodity eggs on a cost-in-use basis.

Source: Green Queen

Thoughts 🤔

One of the biggest bottlenecks for scaling biotech-derived food ingredients is the lack of fermenters at a scale that justifies the cost. The space badly needs more food-grade capacity to support scale-ups.

A greenfield food-grade plant runs hundreds of millions of dollars with multi-year lead times, roughly the cost structure that sank the first full-stack precision fermentation (PF) wave. Pharma fermentation already carries EU and US FDA GMP approvals, sits at industrial scale, and is sometimes underused. Renting into that base, then converting a site as volumes justify it, sidesteps both the CAPEX and the time. It works only because food protein can clear pharma-grade process bars, not the reverse.

Expect more food PF to scale on someone else’s depreciated steel rather than its own balance sheet. The catch is that this only works where pharma capacity is both idle and willing to convert to food use. That pool is smaller than the 9M-litre headline suggests, and once a site goes fully dedicated to one customer, both sides inherit the single-tenant risk that makes fermentation capacity so hard to underwrite in the first place.

🧬 Rainbow Crops raised €9.7M seed round to scale AI-guided multiplex gene editing for complex crop traits

• The Belgium-based startup, spun out of research institute VIB in 2025, uses AI to predict which gene combinations drive complex traits like drought tolerance and higher yield, then applies multiplex editing and breeding to create populations of edited plants for high-throughput screening. The traits it targets are polygenic, controlled by networks of interacting genes rather than one.

• Rainbow Crops claims an edge in multiplexing, editing 50-100 genes at a time rather than a handful, and is moving beyond simple knockouts toward tuning regulatory elements to dial gene expression up or down. It says its approach generates many edit combinations from a single transformation event, easing the burden in crops that are hard to transform or slow to regenerate.

• The model is to license platform output to seed and breeding companies with strong germplasm but limited editing capability. Rainbow Crops earns royalties on commercialised varieties plus upfront R&D fees and milestone payments, and expects to announce a corn collaboration soon.

Investors: LIFTT EuroInvest, AIF (Agri Investment Fund), PINC, VIB, Maia Ventures, Corteva Catalyst

Source: AgFunderNews

Thoughts 🤔

What I find interesting about this round is how much VIB de-risked the company before venture money came in. By the time investors wrote the cheque, the science was already well advanced and validated in greenhouse and early field trials. That is uncommon at seed stage and worth paying attention to.

Rainbow did not spin out in 2025 as an idea and then go raise. The yield-gene science was matured for years inside VIB, an entrepreneur-in-residence was brought in specifically to commercialise it, the company gets access to VIB’s transformation and automated phenotyping facilities that screen up to 16,000 plants at once, and a Gates Foundation grant landed three months before this close. Those are resources a startup at this age rarely has, and the result looks unusually mature for its stage.

Investors are pulling back from early-stage capital. A studio that absorbs the science risk upstream is one way to make a young company look fundable in a market that no longer wants to underwrite raw technical risk.

🥛 Vivici secured €12.5M from the European Innovation Council to scale its cow-free whey proteins

• Vivici makes animal-free dairy proteins via precision fermentation, sold under its Vivitein brand. The Dutch startup secured €12.5M in blended grant-and-equity finance through the EIC’s 2026 accelerator, one of 38 businesses picked from 87 proposals. The capital expands supply and market access for its whey ingredients.

• Its flagship Vivitein BLG is a bioidentical beta-lactoglobulin prized for gelling, foaming and binding. It carries all essential amino acids, runs clear and neutral in flavour, and works at up to 50% by weight in meal replacements, functional drinks, nutrition bars and dairy analogues.

• Earlier this year, Vivici added Vivitein LF, an animal-free lactoferrin at 95% purity targeting gut health, recovery and women’s wellbeing verticals. The company has tested its process in a 75,000-litre fermenter and works with co-manufacturers in Europe and the US for commercial-scale output.

Source: Green Queen

Thoughts 🤔

Great to see Europe putting public money behind fermentation scale-ups. Vivici added a blended EIC grant-and-equity package a year after its Series A, and in the same week Solar Foods drew €77.8M from Business Finland for its second factory.

The instruments are not identical, Solar Foods is roughly half repayable IPCEI-linked debt while Vivici’s is grant plus patient EIC Fund equity, but the direction is shared: European public capital is stepping into the part of the curve where private venture has pulled back.

That part is scale-up, and it is capital heavy. Commercial-scale fermenter capacity costs a lot, and the climb from a validated process to positive unit margins can run years. Grants and concessional equity are built for exactly this, longer timelines and softer return expectations.

🏛️ EU adopts product-based gene-editing rules, clearing a faster path to market for NGT crops

• The European Parliament has adopted new rules for plants made with new genomic techniques (NGTs), finalising a process that began with a provisional Parliament-Council deal in December 2025. The reform shifts the EU from process-based to product-based oversight, assessing plants on their genetic traits rather than the technique used to make them.

• A two-tier system splits the field. NGT-1 covers edits comparable to conventional breeding, which, once verified, are treated like conventional crops, listed in a public EU database, and labelled as NGT-1. Herbicide-tolerant and insecticidal traits are excluded. NGT-2 covers more complex modifications and stays under existing GMO rules, including risk assessment, authorisation, and full traceability.

• The framework aligns the EU with the US, Brazil, and parts of Asia, where gene-edited crops already reach markets, and applies equally to domestic and imported products. It validates the regulatory philosophy long pushed by Bayer, Syngenta, and Corteva, preserves the option to patent NGT innovations with safeguards on market concentration.

Source: AgNavigator

Thoughts 🤔

One way I’m reading the EU’s NGT decision is through the question Europe asks before letting a plant in.

The old GMO regime asked how a plant was made, so using a new genomic technique triggered strict authorisation no matter the result. The new rules ask what is in the plant instead. NGT-1 edits, the kind achievable through conventional breeding, get verified on their genetic characteristics and then treated like conventional crops.

The EU was one of the last major agricultural market still regulating by process while the US, Brazil, and parts of Asia had moved to product-based pathways. Traits cleared abroad still hit a separate, slower hurdle in Europe.

The catch is that judging a trait on its characteristics does not automatically make it quick to place. The whole fast-track promise rests on how NGT-1 verification actually runs once the secondary legislation defines it.

If verification is quick and predictable, the product-based question becomes a fast lane. If it turns into a slow case-by-case assessment, the shift is real on paper but the timeline advantage is smaller than the headline suggests.

💊 Helaina’s precision-fermented human lactoferrin moves into kids’ multivitamins via Vital Nutrients

• US supplements brand Vital Nutrients has launched Mighty Multivitamin Without Iron, a gummy for children aged 2 and up built around effera, Helaina’s recombinant human lactoferrin. The ingredient is paired with six fruit and vegetable concentrates, minerals, and vitamins A, B, C, D and K1.

• Lactoferrin is a whey protein found in human breast milk and bovine colostrum, with far higher concentrations in the former. Helaina uses precision fermentation to make a bioidentical version of the human protein, sidestepping the supply limits of dairy-derived lactoferrin.

• The launch pushes precision-fermented breast milk proteins into children’s nutrition, a new application alongside Helaina’s existing targets across functional foods, longevity, women’s health, and infant formula. The product is free of the top 9 allergens, gelatin and pectin, and Vital Nutrients says adults can take the meltable tablets too.

Source: Green Queen

Thoughts 🤔

Clearing the steepest trust bar in supplements, children’s nutrition, with a recombinant breast-milk protein is a strong vote of confidence. It also sits one step below the highest-stakes application of all, infant formula, which is exactly where Helaina is heading after signing a Nestlé formula deal with effera this month.

A kids’ multivitamin seems like a controlled, low-dose way to normalise a human-identical milk protein in a child-facing product before the formula conversation gets real. If the kids’ products gain shelf trust without safety or labelling friction, it strengthens the case for the formula play.

🍫 Döhler acquired Nukoko to scale cocoa-free chocolate made from fava beans

• The German ingredients giant has bought the UK startup behind a cocoa-free chocolate built on fava beans, two years after first investing in the company. The deal folds Nukoko’s approach into Döhler’s wider plant-based ingredients portfolio, pairing the startup’s biotech platform with Döhler’s flavour and formulation expertise.

• Nukoko puts fava beans through a biotransformation and controlled fermentation process before drying, roasting, and grinding them into a cocoa-style powder. It claims the product generates 90% fewer emissions and carries 40% lower sugar with no hit to taste, alongside higher protein and fibre.

• Cocoa prices have swung violently, and Döhler is positioning fava-bean chocolate as a hedge against that volatility for confectionery makers. The acquisition lets Döhler offer customers a ready alternative across formats from confectionery to baked goods and ice cream, moving the technology from early-stage concept to market-ready ingredient at scale.

Source: Green Queen

Thoughts 🤔

Döhler invested in Nukoko two years ago at pilot stage, watched the process de-risk through its transition toward 10,000-litre batches, then bought the whole thing once the technical risk was largely retired. That is corporate venture used as a staged option on process IP.

It works because Döhler already had everything except the process. It is deeply vertically integrated across pulses, seeds, and nuts, with the distribution and formulation muscle to take an ingredient to market. What integration in raw materials does not give you is a fermentation-and-roasting route that turns fava vicilin into chocolate notes.

Döhler clearly bought more than that, a brand, a patent-pending portfolio, a team, and a B2B pipeline, but the piece I find hardest to build in-house, and the one that looks most like the prize, is that biotransformation. On that read, the acquisition takes a route off the table that rivals would now have to build themselves.

This is more a template than a chocolate story. Expect more ingredient majors to run CVC stakes explicitly as acquisition options on de-risked process IP, leaving founders a clean exit and the major a capability it can keep in-house.

🤔 What can agtech learn from biotech about VC returns?

• Michael Lee, managing director at Syngenta Group Ventures, argues agtech should borrow lessons from biotech, a structurally similar but far larger VC arena. Healthcare is ~17% of US GDP vs agriculture’s ~0.9%, a 20:1 gap that shows up across exit metrics. Biotech has 1,000+ buyers, while agtech relies on a handful of agchem and tractor makers, collectively closing just 1-2 deals over $20M annually.

• Plotting deal frequency against deal value, Lee finds agtech M&A clusters tightly in the $30M-$100M band before dropping off sharply, leaving almost no mid-market and no path to large standalone exits. The top 20 non-SPAC agtech exits over 20 years averaged $400M. Back-solving from that with 3x step-ups implies tight round valuations: ~$8M pre at seed, ~$25M at Series A, ~$100M at Series B. Inflated intermediate valuations, he says, are toxic to both returns and company survival.

• Lee flags vertical farming and row-crop genetics as poor fits for VC, citing airline-like economics of high CAPEX, long timelines, and weather and commodity volatility. His prescription: laser focus on entry valuations, avoid companies that have raised over $50M, look for pre-acquisition collaboration signals, and actively widen the buyer universe into food, climate, fintech, and AI-enabled players.

Source: Ag Tech Navigator

Thoughts 🤔

Biotech and agtech are, in my view, very different underneath the similarities Michael starts with. Long timelines, regulatory gauntlets, efficacy trials: the surface case for treating them alike is real. But the one variable that sets valuations runs the opposite way in each. Health is inelastic demand. Food is elastic and commodity-priced.

Michael gives the mechanism in a single line that deserves more weight. A 10% yield gain is priced at the value of the extra yield minus any change in input cost, a hard arithmetic ceiling. A 10% improvement in an oncology outcome is priced at whatever someone will pay to live.

One number is bounded by a commodity price the grower does not control, the other by willingness to pay with almost no ceiling. That is why biotech sustains a premium agtech has not been able to command at scale. Everything downstream follows from that gap.

You might think the ‘food is medicine’ shift rescues this from above, with consumers paying more for better food. But that is a different buyer at a different layer. The grower buying a seed trait is not the shopper buying the product, and where a wellness premium does exist, it has mostly stayed downstream with the brand rather than flowing back to the trait or the input that created it.

🤝 A*STAR and NUS launch a synthetic biology joint lab to move Singapore’s bio-based research closer to market

• The new joint lab pairs A*STAR’s Singapore Institute of Food and Biotechnology Innovation, which brings bioprocess development and scale-up know-how, with the National University of Singapore’s synthetic biology group, which contributes fundamental science and talent development. Its initial focus is nutrition and consumer care, with broader applications across advanced materials and health.

• The lab targets three areas to speed up translation: AI-guided enzyme and pathway engineering to shorten development timelines, industrially deployable microbial platforms to make complex molecules at scale, and access to novel bio-based compounds for ingredients and functional applications. It will work with companies to co-develop, test, and validate sustainable alternatives to conventional chemical manufacturing.

• Beyond R&D, the lab doubles as a training ground for synthetic biology, metabolic engineering, and industrial biomanufacturing talent through joint supervision, internships, and fellowships. Demand for these skills is rising as more companies build in-house biotech capabilities.

Source: NUS News

💰 Anterra Capital raised a $100M first close on Fund III to “rewire” the food and agriculture system

• Anterra is an agrifood VC that backs companies fixing the existing food system rather than new categories. The maths drives the thesis: the meat industry is worth close to $2T, the alt-meat market $10-20B. Founding partner Maarten Goossens argues there is more value in rewiring the bigger system than in reinventing the smaller one.

• The fund targets $200M total and has already made 2 investments. One is veterinary biologics company Animerra, which Anterra founded and funded in-house. The firm builds startups from within its own walls, a model that produced 3 companies in Fund I and 7 in Fund II, with a target of 7-10 for Fund III.

• Anterra is also leaning into vertical AI, backing Anchr, an agentic platform tackling inefficiencies across the seafood chain from vessel to restaurant. Goossens says the firm benchmarks against general VC rather than niche agrifood peers, and claims Funds I and II are delivering top-tier returns.

Source: AgFunderNews

🐄 Deco Labs upcycles canola oil waste into a plant-based albumin replacement that cuts cultivated meat’s most expensive media cost

• The Tufts spinout’s cAlbumin is a protein isolate extracted from rapeseed meal, a canola oil byproduct normally destined for livestock feed. It drops in for both bovine serum albumin and recombinant versions, the single most expensive ingredient in serum-free culture media.

• Conventional albumin and transferrin together add nearly $100/kg to cultivated meat production. Deco Labs says cAlbumin contributes ~$0.02 per litre of media at scale, works at lower concentrations, and has outperformed both recombinant and bovine albumin across more than a dozen cell types.

• GFI estimates the industry could need up to 10,000 tonnes of albumin by 2030, volumes that don’t yet exist. Deco is validating every batch for 10+ cell doublings before shipping and is already sampling cultivated meat, seafood, leather, and collagen firms, plus cell therapy and vaccine manufacturers. A pipeline of amino acid (cAminos) and FGF-2 (pFactor1) replacements is in development.

Source: Green Queen

Thoughts 🤔

Some companies are trying to engineer albumin out of the process entirely, adapting cells to grow without it. Deco is making the opposite bet. Its thesis rests on the claim that albumin-free cells underperform at scale, so the smarter move is supplying a cheap drop-in replacement rather than removing the protein company by company.

That is what makes the supplier position attractive: a rapeseed isolate at ~$0.02/L sells into most cell lines, where removal has to be re-solved by each producer.

cAlbumin is being sampled by cell therapy, regenerative medicine, and vaccine manufacturers, not just meat and seafood startups. With cultivated meat funding down and companies shutting down, tying the company to that market alone would be fragile. Biopharma media is growing the other way. The biopharma-adjacent demand likely carries Deco near-term, with cultivated meat as the upside.

CAPITAL & CONVICTION

The sharpest thinking in agrifood tech often happens off the record. My interview series brings it on the record:

• Agtech’s Real Bottleneck Is the Translational Layer, Not the Technology - Beanstalk AgTech’s Justin Ahmed

• Why 2026 Is the Great Shakeout Year for Food and Ag Tech - EcoTech Capital’s Adam Bergman

• What It Takes to Build a Fundable Food Company After “Peak Stupid” - Siddhi Capital’s Steven Finn

• Engineering the Exit: How to Get Acquired in Deep Tech - SOSV’s Cyril Ebersweiler

• How Agrifood Tech Investing Is Shifting and Where Value Will Be Created - PeakBridge’s Yoni Glickman

Browse the full archive. More conversations dropping soon. Stay tuned!

Thanks for reading!

Let me know your thoughts in the comments or by replying to this email. If you want to connect with me on LinkedIn, you can find me here.

If you found value in this newsletter, consider sharing it with a friend who might benefit! Or, if someone forwarded this to you, consider subscribing.

Share

Subscribe now

Disclaimer: The views and opinions expressed in this newsletter are my own and do not necessarily reflect those of my employer, affiliates, or any organisations I am associated with.

This week covers SWARM and Leaf, both betting that the operational and data layer beats the model, Aphea.Bio’s bioinsecticide deal with Bayer, Tate & Lyle widening its plant-cell sweetener work just before its Ingredion takeover, Neion Bio turning eggs into biologics factories, and the case for biomanufacturing as strategic infrastructure.

Below are the 10 most interesting ag and food tech stories I came across last week, paired with my thoughts on what they mean for where the industry is heading.

🥚 Neion Bio raised an oversubscribed $23M Series A to scale its egg-based biologics manufacturing platform

• Neion Bio uses precision genetic engineering to turn chicken eggs into factories for complex, glycosylated proteins. Its Raptor™ platform restricts recombinant protein production to the egg itself, leaning on a vessel evolution and selective breeding have already optimised to produce and stabilise large quantities of protein in a sterile environment.

• The company says the approach can produce biologics at meaningful scale within days, with advantages in scalability, resilience, capital efficiency, and cost over traditional bioreactors. It also claims the platform can express difficult-to-make proteins that have resisted conventional manufacturing, opening new product categories.

• Proceeds fund pipeline expansion across biosimilars, innovative medicines, critical reagents, and animal health. The raise follows Neion’s March 2026 emergence from stealth and a first commercial co-development and supply deal with a major pharma company, a multi-product biosimilars collaboration for global markets.

Investors: Caffeinated Capital, Digitalis Ventures, Ensemble VC, Trust Ventures, Haystack, Basis Set Ventures

Source: BioSpace

Thoughts 🤔

The framing of the round is interesting. Both the CEO and Caffeinated Capital anchored the pitch on domestic supply chains and access to critical medicines, and that framing carries more weight now the BIOSECURE Act became law in December and biomanufacturing is treated as a national security domain. An oversubscribed round, in a market that has lost more than 200 biotechs to cash problems in four years, seems hard to explain on biosimilar economics alone.

A flock is domestic, distributed, and does not need a $500M facility, which maps onto the reshoring thesis neatly. I suspect the capital here is following the policy category as much as the cost curve, and if so, the framing that cleared this round also raises the bar later.

Worth noting the biology is not new. Alexion’s Kanuma was approved out of transgenic hen egg whites back in 2015, and egg-based production stayed a one-drug curiosity for the decade since while everyone kept building CHO capacity. That proven biology is reassuring for regulatory risk. This could become a platform if the differentiator is repeatability across proteins rather than the egg itself.

🌾 Leaf Agriculture raised $13M Series B to standardise farm data and power AI tools across agriculture

• The US startup cleans and structures fragmented farm data so partners can build AI tools on top of it. Field devices store information like farm boundaries, product names, and agronomic practices in incompatible formats and calibrations. Leaf resolves this so the data is usable across pre-planning, planting, growing, harvesting, and selling.

• Leaf positions itself as the infrastructure layer for agriculture, comparing its role to Stripe in payments or Plaid in banking: invisible to farmers but mission-critical for the companies building on farm data. The company says it processes data from over 20% of global acres annually, working with crop insurers, ag retailers, input companies, food processors, and commodity traders.

• The funding lands in a tough farm economy where buyers are pushing to optimise operations and adopt AI. Leaf says the payoff shows up downstream. Crop insurance partners pay out claims in days rather than months, retail partners run field-specific seed and chemical models, and compliance partners open new revenue streams while saving farmers hundreds of hours of manual data entry.

Investors: Leaps by Bayer

Source: Leaf Agriculture

Thoughts 🤔

The round is a useful marker for where the binding constraint on agricultural AI sits. It is not the model.

A frontier system cannot just ingest farm data and produce useful agronomy, because that data arrives in incompatible OEM formats, proprietary calibrations, and inconsistent conventions. Reconciling it means knowing what a planting prescription, a harvest pass, or a soil result actually means. That is ag-specific knowledge encoded into the data layer, and the part horizontal AI does not bring with it.

A $13M Series B also seems modest for a company touching 20% of global acres, though that is not a knock and I am not encouraging companies to raise money they do not need! A business at that footprint pitching itself as critical AI infrastructure could plausibly command a larger growth round in a friendlier market. A small, strategically anchored cheque suggests the capital is buying ecosystem positioning as much as funding scale-up.

🤖 SWARM Engineering raised $10M to bring real-time AI decision-making to agri-food operations

• The San Francisco-based startup’s platform takes in structured and unstructured data from ERP systems, spreadsheets, and real-time feeds, then runs scenario analyses across thousands of variables in minutes. Customers reach full operational use within 8-10 weeks, versus the 18-plus-month timelines of traditional ERP rollouts.

• The platform is built on what the company calls an “operational ontology” of agri-food and manufacturing, embedding industry-specific decision logic, constraints, and variable relationships into its AI models. SWARM says this lets it work with fragmented, imperfect data rather than requiring deep ERP integration upfront.

• Early deployments anchor the pitch. At Ardent Mills, North America’s largest flour miller, teams now simulate hundreds of logistics scenarios in one session vs a multi-day manual rebuild. Springs Window Fashions cut planning cycles 40% and freed working capital. Funds go toward new use cases, deeper ERP and supply chain integrations, and go-to-market expansion.

Investors: S2G Investments, AgRogue Growth Partners, Radicle Growth, Grit Road Partners, Middleland Capital, Open Prairie, Serra Ventures, and Trailhead Capital

Source: AgNavigator

Thoughts 🤔

SWARM is selling deployment speed as the headline differentiator: 8-10 weeks to production vs the 18-month ERP transformation that kills most enterprise planning deals. It works with fragmented, badly labelled data instead of demanding customers clean it first. As the model layer commoditises, that sequencing is the competitive axis.

The more durable asset is buried in how the CEO describes deployment. The most valuable part, he says, is mapping how decisions get made, including the informal constraints most operators cannot articulate. That is SWARM encoding tacit operational knowledge that “can take decades to accumulate” into a reusable structure. A single deployment improves one customer’s planning. A library of agri-food decision logic compounds across customers, because protein production, grain processing, and food manufacturing share constraint structures.

The asset that compounds is the decision data more than the AI. My guess is the advantage goes to whoever maps the most operating reality across the most sites, and the models increasingly just run on top of that layer.

🍬 Tate & Lyle expanded plant cell culture partnership with BioHarvest Sciences to develop multiple plant-based sweetener molecules

• The British ingredient giant has broadened a collaboration it first struck with Canadian plant cell culture company BioHarvest Sciences in late 2024. The original deal aimed to create plant-based molecules that cut sugar in food and drink while addressing demand for affordable, sustainable, and nutritious alternatives.

• The expanded scope now covers multiple sweetener molecules, giving manufacturers a flexible toolkit for different formulation needs. Tate & Lyle is betting that no single sweetener can deliver everything customers want (sugar-like taste, nature-anchored solutions, reduced calories), so it is developing complementary molecules usable independently or in tandem.

• The work draws on BioHarvest’s Botanical Synthesis platform, a non-GMO technology that produces high-value plant compounds without growing the plants. It combines plant cell biology, elicitation, AI-driven development, and industrial bioreactors to make “precision botanics” it says offer enhanced potency, purity, consistency, and scalability.

Source: Green Queen

Thoughts 🤔

The timing of this partnership is interesting, as Tate & Lyle broadened a multi-molecule sweetener commitment just weeks before agreeing to a £2.7B all-cash takeover by Ingredion, with completion not expected until the second half of 2027.

Ingredion is buying Tate & Lyle largely for its sweetening and sugar-reduction book, the part carrying 16-18% operating margins against Ingredion’s ~9%. That makes the BioHarvest work less a standalone bet and more a feature of the asset being sold. Late-stage R&D expansions before an acquisition tend to do one of two things: build real pipeline depth, or signal to the acquirer that the franchise is worth paying up for. No molecule is named and no timeline disclosed, so it is hard to tell which from the outside.

Interesting to see whether Ingredion keeps funding the BioHarvest relationship after close, or quietly lets it lapse once the deal is locked.

🐛 Aphea.Bio partners with Bayer to co-develop bioinsecticides targeting sap-sucking pests

• The partnership combines Aphea.Bio’s pipeline of bioactive microbial metabolites with Bayer’s global development, regulatory and commercialisation reach. First targets are fruit crops including pome, stone fruit, citrus and grapes, with scope to expand into vegetables and row crops like cotton and soybean.

• The startup positions its metabolites as combining the environmental profile of biologicals with the stability and usability of conventional products, a barrier the industry has long struggled to clear. Sap-sucking insects are a growing problem as resistance and tighter regulation erode legacy chemistry.

• Candidates advance through joint field validation and early regulatory work, with milestone gates tied to efficacy, safety and manufacturability. The deal adds to a wider push by agrochemical majors into biologicals.

Source: AgNavigator

🥛 Lacto Japan made a strategic investment in Leaft Foods to scale Rubisco protein for the Japanese market

• The New Zealand startup extracts Rubisco, the most abundant protein on Earth, from alfalfa grown in Canterbury. It is a complete protein with a PDCAAS score similar to beef, egg whites, and dairy, and is rich in vitamins, minerals, and antioxidants. Functionally, it foams, gels, and emulsifies, setting like egg whites in baked goods.

• Most methods for extracting Rubisco from green leaves destroy its delicate structure. Leaft Foods developed a gentle, food-safe process that preserves protein integrity. It positions the ingredient as an egg alternative for a supply chain hit by climate change and avian flu, and as a way to fortify dairy and non-dairy products.

• The undisclosed investment deepens an existing collaboration to supply Japan’s largest food manufacturers, targeting revenues worth tens of millions of dollars within 5 years. Lacto Japan says the goal is to build a production and sales framework for Rubisco protein and strengthen the partnership.

Investors: Lacto Japan

Source: Green Queen

Thoughts 🤔

Leaft Foods has raised $15M so far and runs a demo plant making a tonne a week. Plantible already has US FDA clearance and a Texas facility producing hundreds of tonnes of Rubisco from lemna. Fudi is extracting from alfalfa on the same logic Leaft uses. On capital and capacity, Leaft is not the leader in this field.

What it has that the others do not yet is a locked route into Japan. Lacto Japan is a dairy ingredients trader, and its value to Leaft is decades of relationships with the country’s largest food manufacturers. Reading its own framing, the investment sits inside its whey protein book, where isolate prices are elevated on GLP-1 demand. A functionally comparable protein is a supply hedge it can sell through rails it already owns.

The differentiation in Rubisco is no longer who can extract but who can place it. If Leaft converts the Lacto Japan channel into signed manufacturer offtake before competitors establish their own Japanese routes, the early-mover lock on distribution likely matters more than its initial capacity disadvantage.

❄️ CryoVera taps an extremophile-inspired molecule to stop ice crystals from wrecking frozen food texture

• The Paris startup is developing antifreeze ingredients that bind to ice crystals and block new water from building on top, preventing the recrystallisation that turns ice cream and frozen products gritty or crunchy. The molecule is inspired by an extremophile organism that survives extreme cold, though CryoVera is withholding specifics until its patent filings clear.

• The real problem is not full thaw-refreeze cycles but the small, repeated temperature fluctuations frozen products face along the cold chain, down to shoppers opening freezer doors. CryoVera’s pitch is that ice recrystallisation inhibition assays show its molecules work at very low inclusion levels, keeping cost-in-use down for manufacturers.

• The team extracts the molecules from abundant natural sources using microbe-secreted enzymes, a faster path to market while keeping precision fermentation on the long-term roadmap. CryoVera is sampling frozen bakery, ice cream, and other firms, with interest also from plant-based meat brands.

Source: AgFunderNews

Thoughts 🤔

Unilever’s ice structuring protein cleared EFSA, sold in the hundreds of millions of units, and still never reached the wider market. The GM yeast it was made with, and the “GMO ice cream” labelling fight that followed, were likely part of why.

CryoVera is going after the same problem and steering around that production choice, extracting its molecule from a natural source already in the US, EU and Canadian food supply, with precision fermentation pushed to the back of the roadmap. That sequencing sounds like a deliberate move to keep a clean-label story intact, and it only pays if that story survives regulatory classification.

The team is candid that the ingredient could land as a food or as an additive with an E-number in Europe. A food classification keeps the clean label but an E-number puts the kind of additive marker on the pack that reformulation teams are trying to strip off. The extraction is hard in its own right, by the team’s own account, but the labelling outcome is what decides whether solving it is worth much.

🌾 China’s 15th Five-Year Plan puts farmer modernisation at the centre of rural policy with a strong push on smart agriculture

• China’s plan for 2026-2030 aims to lift comprehensive grain production capacity to ~725M tonnes by 2030, building on 2025 output of ~714.9M tonnes, the second straight year above 700M tonnes. The framing extends beyond yield to transforming farmers from traditional producers into participants in technology-driven agriculture.

• The plan places strong emphasis on smart agriculture, advanced machinery, improved seed systems, and digital tools, including AI, to raise productivity and ease labour constraints. It also pushes farmers beyond primary production toward rural industry, employment, and entrepreneurship to diversify income.

• The Ministry of Agriculture and Rural Affairs called agricultural and rural modernisation its “most arduous and challenging task” and described it as “lagging behind,” warning that “without agricultural and rural modernisation, there can be no modernisation of the entire country.”

Source: AgNavigator

🏭 Biomanufacturing as industrial infrastructure: Why operating biology, not just designing it, is now the strategic prize

• The global bioeconomy is worth $4-5T, with potential to reach $30T by 2050, and 50+ countries now have a national bioeconomy strategy. The resilience case is structural: chemicals and materials supply chains are concentrated in a few countries with oil, large agricultural bases, or specific processing capacity, which makes them fragile. Biomanufacturing offers a different geometry, with production sited regionally on local feedstocks.

• Most AI excitement in biology points at discovery: designing strains, predicting structures, accelerating R&D. That work addresses only half the problem. Shannon Hall argues the bigger unlock is operational. ML, digital twins, and predictive process control stabilise and transfer living processes in real time, encoding bioprocess knowledge into models that travel between sites.

• For policy-makers, Shannon says treating biology as strategic infrastructure means three shifts. Fund the manufacturing layer (pilot facilities, shared infrastructure, transferable data systems), not just discovery. Structure public-private investment for long horizons. Invest in the workforce and tools that let a distributed network function as one.

Source: World Economic Forum

Thoughts 🤔

Most attention on AI in biology has gone to discovery: designing strains, predicting protein structures, accelerating R&D. This piece is useful because it names the half that gets ignored.

Worth flagging that Shannon’s company, Pow.Bio, sells this kind of operational tech, so the framing is partly promotional, but the underlying point stands. The harder problem is operational: how you stabilise, transfer, and run living processes economically at scale. That is where most fermentation companies die, and the application of AI I am most interested in right now.

The policy asks land in the same underfunded place. Most bioeconomy strategies still skew toward research, so funding the manufacturing layer (pilot facilities and the shared data infrastructure that makes processes portable) targets the messy middle between lab and commercial scale that VC has retreated from.

The strongest of the three asks is funding shared, transferable-data infrastructure, because that is the part no single company can justify building alone and the part that determines whether a distributed network functions as a network. If private investors will not fund these facilities, governments have to step up.

💭 Is AgTech leaving the farm?

• Patrick Honcoop observes that AgTech companies, especially in automation, robotics, and sensing, are increasingly eyeing adjacent industries as their next market. Two recent deals make it visible. In April 2026, Caterpillar acquired the assets of Monarch Tractor, an autonomous electric tractor startup. In May 2026, Elbit Systems’ subsidiary FUSE acquired Bluewhite, an Israeli autonomy platform built for California permanent crops with 100,000+ hours of field operation. Neither buyer was an agriculture company.

• Patrick argues the pull is economics. A robotics company serving California almonds has a TAM in the hundreds of millions. The same autonomy stack pointed at construction, mining, and defence scales into the billions, and volume drives down per-unit cost on hardware that stays expensive because ag volumes are low. Defence has also turned investable again as NATO spending and geopolitical tension reframe dual-use tech.

• His framing splits the move two ways. Some founders are pulled outward by inbound interest after building something valuable in ag. Others are pushed out because the unit economics fail in a fragmented, seasonal, low-margin market. His warning: agriculture can become secondary by resource allocation rather than deliberate choice, leaving companies that serve everyone adequately and no one exceptionally well.

Source: Patrick Honcoop/LinkedIn

CAPITAL & CONVICTION

The sharpest thinking in agrifood tech often happens off the record. My interview series brings it on the record:

• Agtech’s Real Bottleneck Is the Translational Layer, Not the Technology - Beanstalk AgTech’s Justin Ahmed

• Why 2026 Is the Great Shakeout Year for Food and Ag Tech - EcoTech Capital’s Adam Bergman

• What It Takes to Build a Fundable Food Company After “Peak Stupid” - Siddhi Capital’s Steven Finn

• Engineering the Exit: How to Get Acquired in Deep Tech - SOSV’s Cyril Ebersweiler

• How Agrifood Tech Investing Is Shifting and Where Value Will Be Created - PeakBridge’s Yoni Glickman

Browse the full archive. More conversations dropping soon. Stay tuned!

Thanks for reading!

Let me know your thoughts in the comments or by replying to this email. If you want to connect with me on LinkedIn, you can find me here.

If you found value in this newsletter, consider sharing it with a friend who might benefit! Or, if someone forwarded this to you, consider subscribing.

Share

Subscribe now

Disclaimer: The views and opinions expressed in this newsletter are my own and do not necessarily reflect those of my employer, affiliates, or any organisations I am associated with.

Welcome to Issue #148 of Better Bioeconomy, insights on startups, capital, and ideas reshaping food and agriculture for better human and planetary health. Thanks for being here.

Below, I have curated the 10 most interesting ag and food tech stories I came across last week, paired with my thoughts on what they mean for where the industry is heading.

This week covers Nestlé’s bet on precision-fermented lactoferrin for baby formula, a plant immunity biofungicide for Asian rice, wave-form AI that measures how materials behave, illuminated fermentation as a plant cell culture platform, dairy’s inversion into high-value ingredients, €300M underwriting seeds as sovereign infrastructure, and where AI cuts food waste.

🍼 Nestlé teams up with Helaina to develop infant formula using precision-fermented lactoferrin

• Nestlé has begun a multi-year collaboration with Helaina, the New York-based startup behind effera, a precision-fermented human-identical lactoferrin. The two companies will explore new bioactive proteins in early-life nutrition and co-develop infant formula products using effera as a core ingredient.

• Effera is bioidentical to the lactoferrin found in breast milk. The partnership gives Nestlé access to a consistent, scalable supply of a protein that the company says can help bridge the gap between formula and breast milk, addressing one of the category’s longest-standing limitations.

• For Nestlé, this builds on more than 30 years of research in human milk bioactives. For Helaina, it is validation at the highest level of the infant nutrition category, accelerating the path from fermentation platform to mainstream product and clinical credibility.

Source: Green Queen

Thoughts 🤔

The parity story is the headline, but the timing points at supply security. A single ARA oil supplier failure in late 2025 cascaded across Nestlé, Danone and Lactalis at once, exposing how much of the industry’s risk sits in shared, concentrated inputs rather than any one company’s quality systems. Against that, a precision-fermented protein with a dedicated, auditable supply chain is a hedge as much as an upgrade.

Purified lactoferrin is already supply-constrained and expensive at hundreds to thousands of dollars per kg, because colostrum carries so little of it. Fermentation gives Nestlé batch reproducibility and contamination control on a single-source input, in a category where a recall costs far more in regulatory exposure and brand damage than the ingredient margin returns.

My read is that the durable value is in the process reliability as much as the molecule. If effera clears formula-grade QC consistently across runs and sites, the supply-security case carries the deal. If not, this reverts to a nutritional bet competitors with their own lactoferrin can match. The first co-development milestones are where that gets tested.

🧠 Apoha emerged from stealth with $36M to build AI models for materials discovery using a new class of liquid wave-form data

• The London- and San Francisco-based startup captures the wave patterns a material generates when suspended in liquid and subjected to controlled physical stress. A single run yields over 1,000 numerical descriptors in minutes rather than the days conventional lab tests require. The startup calls this readout VIBE (Variations in Inter-facial Behaviour Under Excitation).

• Apoha converts each reading into a “behavioural embedding,” a numerical fingerprint AI models can learn from. The company argues existing AI is trained on text and images, not on how matter tastes, smells, or dissolves, and that this data layer is what unlocks reliable materials design. One early customer used VIBE to find a plant-based vegan chicken substitute within two weeks after a supplier collapsed.

• The $36M includes a 2024 seed and an unlettered round closed this spring. Apoha’s Fortune 500 customers already span pharma, food and beverage, and materials. Series A funds will scale the custom hardware and platform to handle more sample types.

Investors: Singular, Draper Associates, Redalpine, Seedcamp, Wilbe, Nucleus; grant from Innovate UK

Source: Fortune

Thoughts 🤔

AI for molecules has matured along two axes: sequence and structure. Both sit on large existing datasets, which is why language models and AlphaFold-style prediction work as well as they do. Behaviour, how a substance dissolves, aggregates, holds flavour, or wears, has no equivalent dataset at scale, so models infer it indirectly.

Apoha is generating behaviour as a directly measured data type. The physics underneath is established. What interests me is industrialising it into a standard, machine-readable embedding.

This is a big deal because the binding constraint in molecular ML right now is data, not model architecture. The developability literature openly concedes models lag for lack of large physical training sets. When the bottleneck is the missing variable rather than the algorithm, owning a proprietary measurement of that variable is a different position from fine-tuning on public sequence data.

💰 Limagrain secured €300M EIB loan to accelerate plant genetics R&D for more resilient crops to support European agricultural independence

• The French agricultural cooperative is the world’s 4th-largest seed company. The financing channels through its listed subsidiary Vilmorin & Cie to fund a research programme developing innovative plant genetics aimed at major agricultural and food challenges. The loan ranks among the EIB’s largest agricultural financing operations in Europe.

• Research targets agronomic gains: higher yield, resistance to diseases and pests, drought tolerance, climate and regional adaptation, and optimised nutrient use, alongside nutritional and environmental quality. The aim is to put more resilient seed in farmers’ hands as climate pressures mount.

• The loan flows through the EIB’s TechEU programme under the European Commission’s InvestEU guarantee, positioning seeds as a strategic lever for European food sovereignty. This marks EIB’s 2nd loan to Limagrain, following a €170M facility in 2020 to adapt seeds to climate change.

Source: European Investment Bank

Thoughts 🤔

EIB routed this loan through TechEU, the same sovereignty platform it uses for semiconductors, broadband, and defence, and tagged it “agricultural independence.” The instrument is ordinary corporate R&D debt. What is new is the category: seed genetics is being underwritten as strategic infrastructure rather than agribusiness lending, sitting behind the €3B agriculture-and-bioeconomy package and the near-€15B EIB deployed across 2024-25.

The framing matters because it unlocks a funding rail that matches the asset. A breeding cycle runs 7-10 years and does not fit quarterly public markets, which is part of why Limagrain took Vilmorin private in 2023. Sovereign-guaranteed term debt at development-bank terms is structurally better matched to that horizon than the equity it walked away from. Once seeds sit in the same political bucket as chips, the test shifts from “is this bankable” to “can Europe afford not to fund it.”

🤝 Syngenta teams up with Ascribe Bioscience to bring new biofungicide to rice farmers in Southeast Asia

• Ascribe’s Phytalix is a biofungicide that mimics plants’ natural resistance processes. It enables crops like rice, soybeans, corn, and wheat to defend against a broad spectrum of fungal, bacterial, and viral pathogens throughout the growing season.

• Syngenta ran four years of trials before signing, with results including measurable yield gains and efficacy against Bacterial Leaf Blight, one of the most economically significant rice diseases in Asia. Phytalix offers high stability and low application rates, viable for farms of all sizes.

• Under the agreement, Syngenta gets exclusive commercial access to Phytalix for rice and other major crops in Southeast Asia, with potential expansion to other regions. The companies are now advancing registration, with first commercial launches in Asia planned for 2029.

Source: AgroPages

Thoughts 🤔

Ascribe now has its two biggest strategic investors both signing commercial deals on the same technology, carved cleanly by geography and use. Corteva Catalyst co-led the $12M Series A and is developing a Phytalix seed treatment. Syngenta Group Ventures backed that round and has now signed exclusive supply for rice across Southeast Asia. Two competing input majors, both on the cap table, each taking a non-overlapping slice.

A small-molecule biological would find it tough to fund global registration and distribution itself, and Phytalix needs separate filings across fragmented Asian regulators on top of the EPA. Splitting by crop, geography, and use lets Ascribe ride multiple incumbents’ regulatory and distribution rails without handing any one the whole asset. The equity-to-offtake path also gives each major a look inside before committing channel.

My thinking is the value accrues to Ascribe keeping these carve-outs non-exclusive rather than going deep with one. That holds only if the slices stay clean. If Corteva or Syngenta pushes for a broader lock-up once field data lands, the optionality compresses.

🥕 Ingredion acquired prebiotic fibre made from upcycled carrot pomace to expand its functional fibre portfolio

• Ingredion bought Benicaros, a patented prebiotic fibre made from upcycled carrot pomace, from Dutch innovator NutriLeads. The deal is an asset purchase covering all IP, trademarks, clinical trials, and manufacturing know-how.

• Benicaros is positioned as a “precision prebiotic” that selectively feeds beneficial gut microbes while interacting with the immune system. Ingredion says its rhamnogalacturonan-I (RG-I) structure differentiates it from inulin or FOS, working at a daily dose as low as 300mg with less gas.

• Launched in the US as an immune-health ingredient in 2021, Benicaros secured FDA GRAS no-objection in 2022 and now features in dozens of supplements worldwide. It holds approvals across Asia-Pacific and Brazil’s ANVISA, and awaits final novel food approval from EFSA.

Source: NutraIngredients

Thoughts 🤔

This Benicaros pickup landed the same week Ingredion agreed a $3.6B all-cash takeover of Tate & Lyle. Same buyer, two deals, three orders of magnitude apart. One buys platform scale by absorbing a rival. The other buys a single molecule the platform cannot make itself, through a carve-out that took the IP, trademarks, clinical file, and manufacturing know-how and left NutriLeads the company behind.

What interests me is what that smaller deal says about how a startup like NutriLeads realistically exits. It raised modestly, a €4.5M Series C in late 2023, and did the legible work: GRAS, EFSA, and ANVISA clearance, a published clinical file. What it could not fund with VC cheques was global distribution to manufacturers across 120-plus countries, which needs a sales force and balance sheet that take a decade to build.

That capital intensity sits downstream of where VC is comfortable, so the natural handoff is to a strategic that already owns the rails. The open question for founders is whether that makes a single-ingredient company a clean, fundable asset-sale story, or whether building only a transferable dossier caps your leverage when the one buyer who needs it comes calling.

🧫 Ayana Bio and Brevel received a $1.25M grant to combine plant cell culture with illuminated fermentation for bioactive ingredients

• Ayana Bio, a Boston-based Ginkgo Bioworks spinoff, uses plant cell culture to produce bioactives identical to (or more potent than) those in nature. The $1.25M grant from the Binational Industrial Research and Development (BIRD) Foundation is part of a larger $7.5M pool the foundation says has leveraged private funding into a combined $20M investment.

• Brevel’s illuminated fermentation platform runs microalgae bioreactors with both sugar and light. The joint development agreement will test whether Brevel’s process can advance Ayana Bio’s plant cell culture pipeline for use in food and wellness ingredients.

• The partnership targets a structural problem in botanical supply: climate change, shifting agricultural conditions, and contamination risks are degrading the quality and availability of plant-derived ingredients. A combined indoor platform would offer a predictable, scalable alternative for CPG, supplements, and functional food.

Source: Green Queen

Thoughts 🤔

Brevel built illuminated fermentation to make its own chlorella protein, and now appears to be expanding into a platform provider for the plant cell culture category. The product is no longer just biomass, it is the lighted tank itself, sold new-build or retrofit to companies that bring their own cell lines.

Cocoa and coffee wins are already in hand, more partners are signing, and a category full of startups with cell lines but no path to commercial-scale tanks is the gap Brevel appears to be selling into.

What makes this more than a capacity play is what light does to the product. For commodity microalgae protein, light improves yield per litre. For Ayana, the secondary metabolite is the entire product. If light reliably upregulates those pathways, it stops being a growth input and becomes a potency lever, and the value of a potency lever scales with the value of the molecule.

The bet I find more interesting is that Brevel may end up owning the one variable nobody else in plant cell culture has industrialised at scale. Everyone can buy stainless steel. Lighted tanks that work across unrelated cell lines are a narrower thing to control, and if the cocoa, coffee, and Ayana results converge, the leverage sits with the company selling the light.

🧬 Converge Bio received $2.5M Gates Foundation grant to extend its AI drug-discovery platform into crop genomics

• Converge Bio builds generative AI models for life sciences, until now focused on drug discovery. The 37-month grant funds its first move into agriculture, adapting the platform to identify causal genetic variants behind crop traits like yield and climate resilience.

• The company is developing a multimodal, long-context foundation model that treats the crop genome as one interconnected system, analysing millions of base pairs at once and ranking the genes or mutations most likely driving a target outcome.

• An interpretability layer surfaces which parts of the genome led to each prediction, helping scientists prioritise candidates for validation. Converge Bio says this can cut breeding cycles needed for climate-resilient varieties, shortening time to market.

Source: iGrow News

🌾 Pandawa Agri targets fertiliser crisis with microbial-friendly soil ameliorant for Indonesian smallholders

• The Indonesian ag tech company known for its pesticide reductants has built Balance Solution, a soil ameliorant built to complement or partly replace conventional fertiliser. It gives rice and coffee farmers more flexibility when subsidies arrive late or supply chains break down.

• Rather than introducing new microbial strains, the product creates a physical and biological “home” for the beneficial microbes already in the soil. Once those native microbes reactivate, plants take up nutrients previously locked away, sidestepping the common failure mode of inoculants that don’t establish in existing soil ecosystems.

• The company says coffee farmers cut NPK use by ~30% while raising yields 30-40%, and rice farmers cut fertiliser 10% for a ~10% productivity gain. Pandawa is scaling coffee trials from 20 to ~600 hectares in Lampung and South Sumatra, expanding rice to ~100 hectares, and extending palm oil and forestry tests beyond Indonesia.

Source: AgTechNavigator

Thoughts 🤔

The interesting detail is Pandawa’s distribution. Balance Solution is the second product line going through a farmer ecosystem the company already built around its pesticide reductants, complete with bundled agronomy support and downstream offtake to multinational F&B buyers.

That matters because smallholder input products almost always stall on distribution and trust. Reaching fragmented farmers is the expensive part, and most agri-input startups can’t get past it. Pandawa already paid that cost once, so the marginal cost of pushing a second input through the same channel is far lower than a standalone soil-amendment startup faces.

The durable asset is the ecosystem. If the product performs even moderately, the bundled channel could let it scale faster than a better product sold cold. The risk is the same channel concentrating reputational exposure across both lines if field results disappoint.

🥛 GERBER-RAUTH exits dairy commodities trading to focus on advanced dairy ingredients

• The Milan-based private investment company has sold Italian dairy commodities broker L’Interform to food sourcing firm Atlante, a deliberate retreat from commodity cheese trading. The deal reflects a view that synergies between traditional dairy trading and next-generation ingredient tech had largely disappeared by 2024.

• The firm has been building a precision-fermentation and advanced plant-based portfolio since before 2020, with stakes in Perfect Day, New Culture, and Change Foods. Its thesis now centres on dairy proteins, fats, and oleosomes as higher-value categories for specialty nutrition rather than commodity markets.

• According to Chairman Christian Pichler whey demand has grown so fast that “cheese is becoming the byproduct.” He argues that as food-grade fermentation capacity scales, the unit economics of biomanufactured whey will converge with, and potentially match, conventional whey protein plants.

Source: AgFunderNews

Thoughts 🤔

The part that stuck with me was the role reversal: animal dairy moving up into high-value applications, while biomanufactured and plant-based ingredients serve the commodity base. Today, fermentation is pitched premium-first, and rightly so given the cost curve. What Pichler describes is the long-term end state, the opposite of where the category starts.

A cow cannot turn supply on and off, and Europe’s milk pool is flat with farm succession failing. Biomanufacturing can match supply to demand, site production near consumption, and store it. That scale-and-storability edge is why the inversion can eventually be true.

You can already see the first step inside animal dairy itself: with US whey isolate around $11/lb on GLP-1-driven demand, cheese is becoming the byproduct of whey-making.

My read is this stays directionally right only if food-grade fermentation keeps coming online at lower-grade economics. If scale-up stalls, the argument falls back into the premium-niche story, and animal dairy just stays where it is.

📊 Report: How AI is reducing food waste across the US food system, and where the hype outpaces the evidence

• Nearly a third of the US food supply goes unsold or uneaten every year, and surplus food still hit ~70M tons in 2024 despite a 2.2% decline. ReFED and The Spoon drew on 40+ interviews plus academic research and pilot data to map where AI is moving the needle.

• The strongest results come from foodservice, retail fresh categories, and parts of food production, where measurement and forecasting tools have run for years. Companies reported waste cuts of up to 53% in commercial foodservice and manufacturing. AI demand forecasting reportedly prevented ~200M pounds of loss across 26 countries, and storage monitoring saved ~20M pounds of apples across 1,500+ storage rooms.

• AI’s most consistent contribution is visibility, shifting decisions upstream where prevention is still possible. One Afresh pilot with a US value grocer projected $2.7M in annual savings from lower shrinkage and less ordering time, and foodservice pilots showed waste cost reductions of up to 39% per meal.

• The impact is uneven because it depends on behaviour change and aligned incentives. AI recommendations are frequently overridden by store managers regardless of accuracy, since acting on them requires workflow and labour changes most organisations aren’t structured for. Generative AI and LLMs draw heavy interest but remain mostly at proof-of-concept.

• The headline message: AI is not a silver bullet, but it works where data is clean and connected, incentives reward reduction, and teams have authority to act. ReFED’s advice to operators is to fix the data before buying the AI, measure waste reduction rather than model accuracy, and align manager metrics so they stop over-ordering.

Source: ReFED and The Spoon

CAPITAL & CONVICTION

The sharpest thinking in agrifood tech often happens off the record. My interview series brings it on the record:

• Why 2026 Is the Great Shakeout Year for Food and Ag Tech - EcoTech Capital’s Adam Bergman

• What It Takes to Build a Fundable Food Company After “Peak Stupid” - Siddhi Capital’s Steven Finn

• Engineering the Exit: How to Get Acquired in Deep Tech - SOSV’s Cyril Ebersweiler

• Turning the Industrial Biomanufacturing Graveyard Into a Winning Playbook - First Bight Ventures’ Veronica Breckenridge

• How Agrifood Tech Investing Is Shifting and Where Value Will Be Created - PeakBridge’s Yoni Glickman

Browse the full archive. More conversations dropping soon. Stay tuned!

Thanks for reading!

Let me know your thoughts in the comments or by replying to this email. If you want to connect with me on LinkedIn, you can find me here.

If you found value in this newsletter, consider sharing it with a friend who might benefit! Or, if someone forwarded this to you, consider subscribing.

Share

Subscribe now

Disclaimer: The views and opinions expressed in this newsletter are my own and do not necessarily reflect those of my employer, affiliates, or any organisations I am associated with.

Welcome to Issue #147 of Better Bioeconomy, insights on startups, capital, and ideas reshaping food and agriculture for better human and planetary health. Thanks for being here.

Below, I have curated the 10 most interesting agrifood tech stories I came across last week, paired with my thoughts on what they mean for where the industry is heading.

This week covers natural food colours ahead of the Red No. 3 cliff, mycelium meat on brewery infrastructure, compliance-aware crop AI from China, post-bankruptcy precision fermentation playbooks, philanthropic capital filling the agtech valley of death, and virtual fencing’s US land grab.

🎨 Phytolon raised $23.6M Series B to commercialise its yeast-fermented beetroot red dye in the US

• The Israeli startup uses precision fermentation of baker’s yeast to produce natural pigments for food, replacing petroleum-based synthetics like Red Dye No. 3 and Red 40. Its platform is built on two core pigments, yellow and purple, which can be blended into a palette spanning yellow, orange, pink, purple, and red.

• Phytolon genetically engineers yeast to express betanin, the water-soluble pigment found in red beets, then strips the production organism from the fermentation broth. The resulting colourant is heat- and pH-stable, comes in liquid and powder formats, and works across bakery, savoury, frozen, dairy, and confectionery.

• The round closed in three stages, with the majority of capital landing in April. Funds will go toward sales, supply, and distribution partnerships in the US, where the FDA cleared beetroot red in February and simultaneously allowed “no artificial colors” labelling for products free of petroleum-based dyes. A prickly pear yellow is next in the pipeline.

Investors: Millennium Foodtech, NextGen Nutrition, Colorcon Ventures, Yossi Ackerman, and an undisclosed strategic investor

Source: Green Queen

Thoughts 🤔

Phytolon’s Series B sits on two structural conditions. The first is a deadline-driven buyer market. Red No. 3 exits the food supply by January 2027, with HHS/FDA pushing the other six petroleum dyes out by end of 2026.

For CPG procurement teams, that converts natural colour adoption from a clean-label preference into a fixed-date reformulation event. Phytolon’s February 2026 FDA clearance puts it 11 months ahead of the cliff while some startups are still in submission, and the next 18 months of contract pricing power look structurally stronger than the standalone unit economics would suggest.

The second condition is what the round reveals about food ingredient capital. A Series B closing in three tranches, with the majority landing in April from an undisclosed strategic, is staged commitment rather than a single-close conviction round, even with regulatory clearance and a deadline-driven buyer base behind it. My read is that this is what “good” precision fermentation rounds look like now: milestone-gated capital, strategic underwriters, and investors monitoring performance before deploying the full cheque. Founders in the category should plan around staged closes rather than single dates.

🍄 Infinite Roots acquired Bosque Foods to expand its mycelium meat portfolio

• The Hamburg-based biotech picked up the Berlin-founded startup for an undisclosed sum, nearly two years after Bosque liquidated its subsidiary following regulatory hurdles in the EU. The deal folds Bosque’s whole-cut chicken, pork filets, and bacon into Infinite Roots’ mycelium protein lineup.

• Infinite Roots runs submerged fermentation, growing edible mushroom strains in steel tanks on a nutrient-rich broth. Bosque uses solid-state fermentation, growing microbes on inoculated solid surfaces fed by low-value agrifood sidestreams. Combining the two methods gives Infinite Roots both scale and structured whole-cut formats from one platform.

• The acquisition adds Bosque’s IP, process data, and product knowledge to Infinite Roots’ industrial base, which the company says will speed scale-up and broaden its format range.

Source: Green Queen

Thoughts 🤔

This looks like a tidy mycelium consolidation story, and the tech synergies are real with few mycelium companies owning both. What interests me more is the sequencing. Infinite Roots has rescinded its FDA GRAS filing to update its safety dossier, has no near-term EU novel food path, and rerouted commercial activity through South Korea via Pulmuone. Bosque faced the same EU bottleneck and now operates only out of New York. The acquirer is buying capability before clearing its own regulatory bar.

Most consolidation waves see survivors roll up distressed assets after the gating constraint is solved. Here, the regulator is the gating constraint, and the lengthy EFSA approval window means the 2022-2023 cohort was always going to hit a capital wall before approval.

My thinking is that this works if the updated US dossier lands GRAS in the next 12-18 months and EFSA’s reforms open the EU window in time for whole-cut formats. If neither moves, the process synergy still buys optionality wherever approval lands first, but two regulatory-stalled assets end up sharing one runway.

🍗 Pacifico Biolabs raised €7M Series A to brew mycelium chicken in beer tanks

• The Germany-based startup uses fermentation to grow whole-cut mycelium meat, turning the microbes into a protein-rich biomass that mimics whole-muscle texture. It has pivoted from its original seafood focus to chicken, which it will sell to B2B clients under the Viando label and target supermarket shelves across Germany, Austria, Switzerland and the Nordics by late 2026.

• Pacifico’s process runs in standard beer brewery tanks rather than purpose-built bioreactors, and the startup claims this cuts CAPEX by >95% vs traditional biomass fermentation routes. The approach works because the EU’s beer production has declined for 5 consecutive years, leaving fermentation capacity sitting idle and available for retrofit.

• Funds will scale production to 200 tonnes/month in Saxony, expand the team, and lock in commercial partnerships ahead of the late-2026 launch. Pork and seafood formulations remain in the pipeline.

Investors: Stray Dog Capital, TGFS, Sprout & About Ventures, Simon Capital, FoodLabs, and a regional brewery partner

Source: Green Queen

Thoughts 🤔

Two German mycelium startups are now anchoring production on idle brewery infrastructure rather than purpose-built bioreactors. Infinite Roots has been doing the same, partnering with contract manufacturers and repurposing brewery sites on the same logic.

The timing fits the funding market. Purpose-built bioreactor plants priced in the tens of millions are uninvestable right now. Brewery retrofits compress the headline CAPEX number into something a Series A can handle.

That lowers the bar for the whole category to reach commercial scale without raising money the market won’t provide. The vessel is shared, but strain IP, fermentation process, downstream, and product quality still have room to differentiate.

🤝 BioPrime AgriSolutions and Mosaic team up to embed biologicals into mainstream fertilisers

• The Pune-based ag biologics startup will plug its SNIPR-platform biologicals into Mosaic’s potash and phosphate fertiliser lines, with the companies pitching the joint offering as a “Crop Performance System” that goes beyond basic plant nutrition to support crops across the growth cycle.

• The deal gives BioPrime a route into one of the world’s largest fertiliser distribution footprints. Mosaic supplies customers in more than 40 countries, so even modest attach rates on its volumes would put SNIPR-derived biologicals in front of growers at a scale most ag-biotech startups spend years trying to reach.

• The companies say the integrated product is designed to improve return on crop nutrition spend and lift output from existing farmland without a matching rise in environmental impact. The framing matches where the industry is heading as fertiliser prices swing on geopolitical disruption and regulators push for lower-footprint productivity.

Source: AgroPages

🐄 Monil raised $10M to enter the US virtual fencing market

• The Norway-based startup makes solar-powered cattle collars that create virtual boundaries for grazing herds. Monil has sold 50,000 collars across Norway, Sweden, and the UK since launching in 2022 and grew revenue 5x last year to $6.6M.

• Beyond fencing, the collars track activity to flag entrapments and illness, and the startup is now layering on reproductive monitoring. Heat detection launched in May using rumination data, with calving detection arriving in autumn based on other activity signals.

• The capital funds a US launch from a new Kansas City base, where Monil is recruiting sales and field staff and has completed regulatory prep. The company says it is ready to ship to the American market and views the US as a growing opportunity with rapid developments underway.

Investors: Firda and other Norwegian venture capital investors

Source: Beef Magazine

Thoughts 🤔

Halter raised $220M recently and now spans 200+ ranches in 22 states. Vence sits inside Merck Animal Health on 4M acres. Gallagher’s eShepherd has been live in North America since 2024 with existing ag retailer relationships.

Monil is entering at much smaller scale against entrenched competitors. I am curious to see how Monil carves out a defensible position. The Kansas City base hints at a cow-calf focus rather than Western rangeland, where the big three concentrate. Whether Nordic product economics hold for US operators is the other open question.

🌾 Nanjing Agricultural University unveiled China’s first open-source AI model for crop protection

• Called Green Shield, the model was built by NAU’s College of Plant Protection with the National Key Laboratory of Agricultural Biosafety and 30+ industry institutions. It was trained on 2.5B tokens from academic papers, patents, national standards, and field reports spanning rice, wheat, soybeans, vegetables, and fruit trees.

• Green Shield identifies crop type, growth stage, and disease symptoms to generate integrated pest management strategies. Before any recommendation, it cross-checks proposed chemicals against China’s national pesticide registration database, blocking non-compliant suggestions on banned substances, approved crop applications, or dosage limits.

• The release follows Sinong, NAU’s general agriculture LLM trained on 4B+ tokens and shipped in 8B and 32B versions earlier this year. Together they give Chinese agriculture a specialised open-source AI stack, from broad agronomic guidance to compliance-aware crop protection advice in a country where pesticide misuse and resistance remain persistent grassroots problems.

Source: iGrow News

Thoughts 🤔

A free, compliance-grounded crop protection advisor inside the world’s largest pesticide market could change how this layer gets monetised. In my view, tools like Climate FieldView and Cropwise have never really sold standalone advisory revenue per user. They are loyalty layers designed to pull through branded seed and chemistry. Green Shield’s threat is margin pressure on the chemistry book, since a neutral recommender could nudge farmers toward generic actives over branded portfolios.

China’s economics were already hostile to paid advisory products for smallholders, though mid-scale cooperatives and new-type operators do pay for advisory tooling. Open-sourcing a compliance-aware model removes whatever pricing power sat in the model layer.

If Green Shield holds up in deployment, value likely accrues to the players already touching farmer transactions, like input retailers and platform players, rather than the input majors defending branded chemistry through digital agronomy.

🎢 Amyris CEO on re-emerging from bankruptcy with a leaner model built on recurring revenue

• Two years after its 2023 Chapter 11 filing, the synthetic biology pioneer is operating with a monthly burn rate that the company says matches what it used to spend in a single month under prior management. CEO Kathy Fortmann says Amyris closed 2025 a year ahead of plan on both top and bottom lines, though the company is not yet profitable.

• Amyris has exited consumer brands, shed unfavourable contracts, and returned to its roots as a B2B biotech. The company is prioritising recurring revenue from licensing and royalties over large upfront partner payments, and applying back-of-envelope economics earlier to avoid burning cash on molecules that won’t pencil at scale.

• Fortmann has also broken the rule that every collaboration ends with Amyris manufacturing the ingredient. The company will now do strain engineering and process development for partners and license the output, reserving its precision fermentation plant for specialty molecules. Ag biologicals is a new focus, with Amyris targeting startups and incumbents that need help with strain optimisation, scale-up, or initial manufacturing.

Source: AgFunder

Thoughts 🤔

A useful read for founders building in biomanufacturing on which layer of the stack is worth owning. The first wave of precision fermentation companies bet that the durable moat came from owning everything: strain, process, manufacturing, often the consumer brand on top. That thesis assumed cheap capital and patient timelines, and both have gone.

Amyris is the visible example of what happens when full-stack ambition runs into the cash discipline needed to make any single layer profitable. Ginkgo is converging on a similar narrower repositioning after its own restructuring. The survivors are picking the layer they want to own, monetising it through recurring revenue, and treating adjacent layers as someone else’s problem. The strategic question for the next cohort is no longer “what molecule” but “which layer”.

💰 Renaissance Philanthropy bets on “venture capital for public good” to reshape agricultural innovation

• Renaissance Philanthropy has mobilised $533M across 22 programmes and five government partnerships, with $265M routed to third parties and $268M into internally managed initiatives. The two-year-old nonprofit pitches itself as a field builder, structuring thesis-driven programmes with stage-gated decision points more akin to venture studios.

• Much of its agricultural work sits inside the Advanced Research for Climate Emergencies (ARC) fund, targeting underexplored risks like methane and nitrous oxide emissions, ag-driven deforestation, and climate tipping points. Through the Climate Emergencies Resilience Lab, built with Deep Science Ventures, it identifies scientific opportunities and spins up agtech ventures where commercial pathways exist.

• The bet is that philanthropic capital can de-risk the agtech “valley of death” through 5-7 year programmes that produce decision-grade science and hand off to governments, corporates, or investors. Renaissance claims climate stabilisation receives <0.1% of global climate finance, framing agriculture as a node where risk-tolerant capital could unlock returns conventional funders can’t reach.

Source: AgNavigator

Thoughts 🤔

The “venture capital for public good” framing obscures the more interesting structural read. The operating model: 5-7 year programmes, milestone-gated, domain-expert-led, with handoff plans is closer to a Focused Research Organization than a venture studio. FROs have crowded into protein structure, brain connectomics, and climate measurement through folks like Convergent Research and Astera. But agriculture has been largely absent from that wave.

The gap they fill is the messy middle between USDA-style basic research, CGIAR-style smallholder programmes, and venture-backed agtech. None of those fund a 5-year team chasing a specific milestone in crop biology with a defined sunset and a public-good deliverable. FROs landing in agriculture now is a counter-cyclical move into a vacuum. The big ag players have pulled back internal R&D, USDA budgets are constrained, and CGIAR is in a funding crisis.

If Renaissance can scope milestones that produce shared infrastructure (phenotyping platforms, trait libraries, climate-response biological databases), the compounding leverage downstream could look closer to what AlphaFold gave structural biology than what venture has produced in ag tech.

🛡️Syngenta opens $10M Almería R&D hub to halve breeding timelines against fast-emerging crop pathogens

• The R&D Technology Centre in El Ejido, Almería, consolidates breeding, trait development, seed operations, fruit quality, applied data science, and digital tools under one roof. Biosafety infrastructure on site lets researchers study high-risk pathogens in quarantine.

• The location is strategic. Almería’s “Sea of Plastic” packs >30,000 hectares of greenhouses producing ~4M tonnes of vegetables annually, giving breeders constant, diverse disease pressure to stress-test varieties under commercial conditions. Syngenta says a new serious pathogen now emerges roughly every 2 years, with threats like ToBRFV and Downy Mildew driving the urgency.

• The company claims the facility can cut breeding timelines by up to 50%, taking varieties from 4 years to as little as 2, on the back of AI and ML in its breeding pipelines. Insights generated locally feed Syngenta’s global R&D network, with the field-to-lab model designed to shorten the loop between pathogen identification and resistant seed delivery.

Source: AgNavigator

💭 Ag tech hype cycle 2026: The sector is entering its industrialisation phase

• Hadar Sutovsky’s thesis is that value creation is shifting upstream into precision agriculture, irrigation intelligence, AI crop scouting, biologicals with proven ROI, and specialty-crop robotics. These categories are moving from experimentation into operational deployment, signalling AgTech’s transition from pilot purgatory to embedded infrastructure.

• Hadar argues physical AI is becoming real on the farm through task-specific automation rather than the fully autonomous farm narrative that dominated earlier cycles. Biology is entering its execution phase too, with climate-resilient genetics, next-gen biologicals, and regenerative infrastructure now judged on field performance and integration into existing farm workflows.

• The reset categories are indoor farming, generalised autonomy, and broad platform plays, all hitting the wall on unit economics, adoption friction, and infrastructure constraints. Winners will be the solutions that slot into systems farmers already use while delivering measurable ROI under real-world conditions.

Source: Hadar Sutovsky/LinkedIn

CAPITAL & CONVICTION

The sharpest thinking in agrifood tech often happens off the record. My interview series brings it on the record:

• Why 2026 Is the Great Shakeout Year for Food and Ag Tech - EcoTech Capital’s Adam Bergman

• What It Takes to Build a Fundable Food Company After “Peak Stupid” - Siddhi Capital’s Steven Finn

• Engineering the Exit: How to Get Acquired in Deep Tech - SOSV’s Cyril Ebersweiler

• Turning the Industrial Biomanufacturing Graveyard Into a Winning Playbook - First Bight Ventures’ Veronica Breckenridge

• How Agrifood Tech Investing Is Shifting and Where Value Will Be Created - PeakBridge’s Yoni Glickman

Browse the full archive. More conversations dropping soon. Stay tuned!

Thanks for reading!

Let me know your thoughts in the comments or by replying to this email. If you want to connect with me on LinkedIn, you can find me here.

If you found value in this newsletter, consider sharing it with a friend who might benefit! Or, if someone forwarded this to you, consider subscribing.

Share

Subscribe now

Disclaimer: The views and opinions expressed in this newsletter are my own and do not necessarily reflect those of my employer, affiliates, or any organisations I am associated with.

Welcome to the 146th edition of Better Bioeconomy, insights on startups, capital, and ideas reshaping food and agriculture for better human and planetary health.

Below, I have curated the 10 most interesting agrifood tech stories I came across last week, paired with my thoughts on what they mean for where the industry is heading.

This week’s edition covers livestock data infrastructure, methane-cutting feed tech, plant immunity platforms, precision fermentation, cocoa-free chocolate, and China’s industrial play on global food.

🧬 StrainX Bioworks emerged from stealth with $13M to scale precision fermentation ingredients for food and nutrition

• The India-based startup runs a 10,000-litre facility and 100+ team, with modular capacity that scales toward 100,000 litres over the next 12-18 months as commercialisation lines up. Specific ingredients remain undisclosed, though initial work is in alt proteins.

• StrainX takes a full-stack approach, building strain engineering, process development, fermentation scale-up, downstream processing, and manufacturing in-house rather than relying on co-manufacturers. The startup says India offers limited precision fermentation infrastructure outside pharma, so owning the stack lets it control process economics and iteration cycles.

• One molecule has self-affirmed GRAS status in the US with a no-questions letter pending, and multiple ingredients are awaiting Indian approval. StrainX claims price parity with conventional versions, helped by what it calls structural manufacturing advantages in India, and is positioning as a B2B partner to global food and ingredient companies.

Investors: Prime Venture Partners, Leo Capital, Good Startup, Sparrow Capital, Sun Icon Ventures, Dholakia Ventures, WindT Angels

Source: Green Queen

Thoughts 🤔

The full-stack framing in StrainX’s round is doing a lot of work. They call internal manufacturing a strategic choice for owning process knowledge and long-term defensibility, which is a real moat argument.

But India’s food-grade precision fermentation CMO infrastructure is still coming together outside pharma, so building strain engineering, downstreaming, plant operations, and commercialisation in-house was also the only path to 10,000-litre scale. Both can be true.

The harder question is what that combination costs to defend. StrainX is carrying the entire stack plus a 100-person team on a single Series A, while Western peers raising similar tickets stay asset-light through contract capacity and route burn into strain work and customer development. Whether full-stack compounds into defensibility or becomes a CAPEX overhang depends on how quickly the Indian CMO layer matures around them.

🌱 Resurrect Bio closed its Series A at $10.3M to restore crops’ natural disease defences through gene-editing targets

• The UK startup, spun out of The Sainsbury Laboratory in Norwich, identifies how pathogens suppress plant immunity and pinpoints the edits needed to switch defences back on. The round follows an $8.1M initial close in February.

• Plants rely on NLR immune receptors to fight off threats, but pathogens secrete effector proteins that bind to these receptors and shut down the response. Resurrect Bio’s FloraFold AI predicts those protein-protein interactions and screens them in-silico and in-planta, then hands seed company partners the exact gene-editing template to break the binding.

• The funds scale the team and develop resistance traits in major crops, building on a joint development agreement with Corteva in corn. The model sidesteps chemical inputs and works across most crops, pathogens, and pests, opening room for more deals with seed companies and breeders.

Investors: Corteva Catalyst, Calculus Capital, Pymwymic, Future Planet Capital (UKI2S), SynBioVen, AgFunder

Source: AgFunderNews

Thoughts 🤔

Corteva Catalyst has now made ~6 biology-first bets in twelve months, spanning microbes, peptides, AI proteins, and plant immunity. Resurrect Bio is the latest. The cadence and breadth point toward a distributed external R&D function rather than discrete corporate venture activity.

Synthetic crop protection registration costs have crossed $300M with 12-year timelines, regulatory windows are narrowing, and pathogen resistance is eroding existing chemistry. Catalyst lets Corteva access biological mode-of-action innovation through joint development agreements and pre-negotiated licensing options, without carrying internal R&D overhead.

This could be the template for how big ag runs R&D for the rest of the decade. If it does, value accrues to platform companies with optionality across incumbents, not to those who go exclusive early. The tell will be how many of these six investments convert to exclusivity over the next 18-24 months.

🐄 URUS acquired AgriWebb to bring integrated data infrastructure to the global beef industry

• The Australian livestock management platform collects field and cowshed data for farm mapping, grazing, and inventory tracking across 10,000+ farms, 23 million animals, and 150 million acres in 26 countries. It also acts as supply chain infrastructure for McDonald's, Nestle Purina, Wendy's, Ahold Delhaize, and Sainsbury's.

• URUS runs nine brands across dairy and beef genetics, reproductive services, and dairy software. AgriWebb pushes the platform to 25 million animals and extends URUS’s dairy flywheel into beef, linking breeding, animal health, sustainability, and traceability in one system.

• The deal reflects a wider structural shift: livestock technology is moving away from standalone tools toward integrated systems connecting genetics, animal health, environmental data, and supply chain traceability into something that functions more like critical infrastructure.

Source: AgFunderNews

Thoughts 🤔

URUS already runs a genetics-data flywheel in dairy through VAS and DairyComp, where individual-animal monitoring feeds back into which semen gets sold next. Beef has historically broken that loop because grazing systems are extensive and data-poor. Building telemetry across 10,000 farms in 26 countries is a 10-15 year capital problem, so URUS bought the instrumentation layer instead of building it.

The genetics-as-platform frame also reframes what AgriWebb is worth to this specific buyer. Standalone, it's livestock SaaS with thin comparables. Inside URUS, it's the input layer that lets dairy-style genetics decisions extend into beef, where AI penetration has plateaued and Scope 3 buyers are creating new demand for producer data.

🍫 California Cultured and UC Davis won $2.8M in US government grants to cut the cost of cell-cultured chocolate

• The California startup is working with UC Davis to scale its cultured cocoa platform, with commercial manufacturing slated for early 2027 to fulfil a first purchase order. The grants from BioMade and NSF are part of a 14-project DoD-backed push to strengthen domestic biomanufacturing.

• Current cellular agriculture economics for cocoa are punishing, with steel bioreactors, high-pressure steam sterilisation, and low batch productivity all driving up operating costs. The team will design custom high-density polyethylene (HDPE) bubble bioreactors, test alternative sanitisation strategies, and run semi-continuous operations with real-time in-line biomass monitoring to lift volumetric productivity.

• Internal tests show the HDPE reactors outperforming stainless steel in plant cell culture. The work also feeds into low-cost drying, food-safe media, and strain productivity, with techno-economic and life-cycle analyses to test transferability across plant, algal, microbial, and fungal systems.

Source: Green Queen

🧫 Melazyme raised $2M seed to advance precision-fermented melanin and sweet proteins platform

• Founded in 2025 by two Perfect Day alums, the US startup uses precision fermentation to make biomolecules, led by melanin. It engineers strains for high-yield melanin synthesis on renewable carbon feedstocks and uses solvent-free purification to yield a powder it calls Black Gold.

• Melazyme claims its melanin reaches >99% purity, well above the 70-80% from cuttlefish-extracted melanin and the 75% of chemically synthesised versions. The naturally occurring biopolymer combines broad-spectrum UV absorption, chemical stability, and strong affinity for metal ions, opening applications across cosmetics, functional coatings, filtration, and rare-earth element recovery.

• Near-term commercial focus is cosmetics, where Melazyme is engaging global manufacturers and has developed a melanin precursor that binds to hair keratin to restore colour without ammonia or oxidative dyes. It has also advanced brazzein, a heat-stable sweet protein from the West African Oubli fruit, 500-2,000x sweeter than sucrose, with commercial partners.

Investors: SeaX Ventures, Stellaris Venture Partners, Plug and Play Ventures

Source: Green Queen

Thoughts 🤔

The melanin-cosmetics framing makes Melazyme look like a category pioneer, but from my light research, it is the second venture-backed entrant on this molecule. IndieBio-backed Avisa Myko has been at it since 2019 on a cost-collapse pitch, claiming $0.20/gram melanin versus a $400/gram reference.

Melazyme is going the other way, leading with 99% purity versus 70-80% for cuttlefish-extracted and 75% for chemical synthesis. My read is that this is less a head-to-head and more a category bifurcation. Cost wins mass-market sunscreen and PPD-free hair dye. Purity wins premium cosmetics and any medical adjacency.

Separately, I would treat the brazzein side of the platform as narrative more than commercial intent. If they do get into the brazzein market, Melazyme is entering one of the most over-crowded subcategories in precision fermentation. If there is no GRAS pathway disclosed within 12-18 months, I’d treat brazzein as platform decoration and assume the real business is melanin.

🐄 Triple Bio raised €1.5M to advance lipid-based feed additive platform to cut methane and boost milk yields in dairy cattle

• The Netherlands-based startup runs two lipid formulations. RumeNRG-PL encapsulates existing methane inhibitors like bromoform, protecting actives from rumen degradation and enabling gradual release. RumeNRG-Mx1 is a standalone additive that redirects hydrogen flow in the rumen without any loaded compound.

• Rather than suppressing methanogens, the approach makes hydrogen more bioavailable to fermentative bacteria that generate most of the animal’s metabolic energy. In vitro rumen simulations showed a 28% boost in volatile fatty acid production, which Triple Bio says could mean 5-10% higher milk yield. The encapsulated formulation also matches methane reduction at 500x lower dose.

• Live-animal trials run later this year, with proof-of-concept data feeding a larger 2026 round. The longer-term ambition: one product line delivering productivity gains and up to 50% methane reduction without any loaded inhibitor.

Investors: Nucleus Capital, Positron Ventures, Climate Club

Source: AgFunderNews

Thoughts 🤔

Triple Bio’s near-term product is aimed at the companies already in enteric methane sitting on stabilisation, dosing, and palatability problems with their own actives. A 500-fold dose reduction on a known inhibitor is a margin gift to Rumin8, Symbrosia, and the bromoform/3-NOP cohort, not a competitor.

The split mirrors how pharma separated APIs from formulation once actives cleared regulatory and delivery became the next defensible layer. Rumin8 patented oil-based stabilisation for tribromomethane. DSM spent over a decade taking 3-NOP through EFSA and FDA. With the molecules now squeezed by dose-related cost, safety thresholds, and consumer backlash, encapsulation is where competitive economics are won, and the actives commoditise underneath.

🤝 Endless Food Co lands Dagrofa deal to push bean-free chocolate into Danish foodservice

• Denmark’s Endless Food Co has signed a strategic partnership with Dagrofa Foodservice, one of the country’s leading foodservice distributors, to get its cocoa-free chocolate alternative THIC (This Isn’t Chocolate) into professional kitchens nationwide.

• THIC is built from upcycled side streams, including brewer’s spent grain, cacao husks, and oat milk pulp, which make up to 40% of the product. The rest is wild-harvested shea butter and organic beet sugar, processed through traditional chocolate-making into a powder that drops into existing production lines.

• The deal opens a major new commercial channel for the startup as cocoa prices remain under sustained pressure from climate-driven supply shocks. It also follows the company’s earlier retail push with 7-Eleven Denmark, signalling a parallel build across foodservice and convenience retail.

Source: Green Queen

Thoughts 🤔

The cocoa-free chocolate story has been written mostly through retail and industrial confectionery rails. Barry Callebaut-Planet A Foods, Cargill-Voyage Foods, Mars and Nestlé picking up ChoViva. Endless Food Co’s Dagrofa deal opens a different rail and one I suspect is underrated. Dagrofa is Denmark’s market leader in canteens and hotels, runs ~35,000 SKUs across a nationwide fleet, and already supplies Compass Group’s roughly 100,000 daily meals in the country.

Institutional foodservice is a structurally different buyer to retail confectionery. Procurement turns on cost per kg, batch consistency, supplier reliability, and operator ease. The channel is slower to enter but stickier once in, and largely insulated from the retail cycle driving most cocoa-free coverage.

🥟 Intake unveiled Takein to push precision-fermented yeast protein and fibre into global B2B supply

• Takein is the new ingredient arm for the South Korean startup’s precision-fermented yeast protein and fibre, marking its shift into a ‘materials-focused’ model ahead of an IPO. The launch comes a year after Intake’s $9.2M Series C to scale its grape-derived wild yeast strain.

• The yeast protein isolate is positioned as a third-generation whey replacement, with a complete amino acid profile and what the company claims are higher digestion and absorption rates and lower allergy risk than dairy whey. The yeast-derived fibre is extracted from the cell walls left after protein fractionation and contains beta-glucan and mannose, which Intake says support immune function and gut health.

• Intake will trial commercial use through its own brands first, ProteeOne protein powders and Sugarlolo’s sugar-free range, before scaling supply to global manufacturers.

Source: Green Queen

🧫 Capra Biosciences cut fermentation monitoring costs with chickpea-sized wireless biosensors floating inside its bioreactors

• The Virginia-based startup, which makes retinol and other high-value ingredients via biomanufacturing, is partnering with Boston University on a wireless network of free-floating microbial-electronic sensors that pair engineered light-emitting cells with embedded electronics. Backed by NSF and BioMADE, the sensors will be miniaturised and tested at Capra's 10,000 sq ft pilot facility.

• Traditional pH, dissolved oxygen, and CO2 probes cost $5,000-$10,000 each, require insertion into the broth, risk contamination, and give a single-point view. Capra says the new biosensors could hit $10-$100 per unit at scale, measure multiple analytes through an onboard potentiostat, and float at different points for spatial visibility.

• The economics matter most for Capra’s modular setup, which runs multiple 1,000-L units rather than one large tank and could eventually involve hundreds or thousands of reactors per facility, making sensor cost a material driver of unit economics. The company has also just launched what it claims is the first fermentation-derived salicylic acid for personal care, and says it has run continuous production of that molecule for up to 100 days.

Source: AgFunderNews

Thoughts 🤔

The Capra biosensor story is being framed as a cost story. I think the more interesting frame is what free-floating sensors enable. Industrial fermenters are full of gradients. Oxygen, pH, substrate, and shear all vary across the vessel, and current control systems essentially ignore that because they can’t see it.

One probe near the wall is treated as a fair proxy for a well-mixed tank that may not really exist at scale. The mainstream workaround has been mechanistic CFD modelling and more aggressive mixing, both of which hit physical and energy limits as reactors grow.

A wireless network of free-floating sensors changes the data substrate. It gives spatially distributed, time-resolved readings of pH, dissolved oxygen, redox state, and bioluminescent stress signals from engineered Yarrowia lipolytica reporters across the broth. That is the missing input layer for the AI-driven fermentation control everyone has been pitching but few has had the data to actually train.

🇨🇳 Report: China is about to do to global agriculture what it did to solar and EVs

• Beijing is applying the same industrial playbook behind its dominance in solar, wind and EVs to food and agriculture. The 14th Five-Year Plan was the first to place food security alongside energy and finance as a pillar of economic security, and the 15th explicitly bets on synthetic biology, alt proteins and agricultural biotech. Commercialised GE maize and soya approvals in 2024 signal the policy machinery has switched on.

• By 2030, China’s soya bean imports are projected to fall by 23.5 million tonnes, a 25% drop worth roughly $12 billion. That is almost equal to everything the US shipped to China in 2024. Smaller contractions hit beef, poultry, dairy and egg imports, driven by feed reformulation, a 50% cut in avoidable food loss, and productivity gains on existing land.

• By 2040, the report projects China will flip to a net exporter across poultry, dairy, eggs and farmed aquatic products, with alt proteins already taking 14% of beef and 16% of seafood domestically. Producers in those categories will face Chinese exports in their home markets, on top of losing the buyer they were built to serve.

• The report models alt proteins meeting 35-55% of Chinese animal protein demand by 2050, with cultivated meat commercially viable in the 2040s. More importantly, it argues that China captures the upstream value chain, amino acids, fermentation infrastructure, and bioreactor supply, replicating its position in solar wafers and lithium cells.

• Exposure is concentrated. China absorbs 89% of Argentina’s soya exports, 71% of Brazil’s and 53% of the US’s, plus 75%, 54% and 41% of Argentine, Brazilian and New Zealand beef. Producers who move first on deforestation-free supply, traceability and market diversification will manage the transition. Those who don’t face simultaneous volume and price contractions.

Source: Systemiq

CAPITAL & CONVICTION

The sharpest thinking in agrifood and bioeconomy often happens off the record. My interview series brings it on the record:

• Why 2026 Is the Great Shakeout Year for Food and Ag Tech - EcoTech Capital’s Adam Bergman

• What It Takes to Build a Fundable Food Company After “Peak Stupid” - Siddhi Capital’s Steven Finn

• Engineering the Exit: How to Get Acquired in Deep Tech - SOSV’s Cyril Ebersweiler

• Turning the Industrial Biomanufacturing Graveyard Into a Winning Playbook - First Bight Ventures’ Veronica Breckenridge

• How Agrifood Tech Investing Is Shifting and Where Value Will Be Created - PeakBridge’s Yoni Glickman

Browse the full archive. More conversations dropping soon. Stay tuned!

Thanks for reading!

Let me know your thoughts in the comments or by replying to this email. If you want to connect with me on LinkedIn, you can find me here.

If you found value in this newsletter, consider sharing it with a friend who might benefit! Or, if someone forwarded this to you, consider subscribing.

Share

Subscribe now

Disclaimer: The views and opinions expressed in this newsletter are my own and do not necessarily reflect those of my employer, affiliates, or any organisations I am associated with.

For Issue #145 of Better Bioeconomy, I sat down with Adam Bergman, Founder and Managing Director of EcoTech Capital.

EcoTech Capital is a strategic advisory firm working with food and agriculture companies on capital raising, strategic partnerships, and M&A. Unlike a venture fund, Adam is not writing cheques. He sits alongside founders as they navigate one of the toughest fundraising markets the sector has seen, and he has done so across roughly 15 years of ag tech deals, starting with one of the first of them in 2012.

Adam has spent more than twenty-five years in CleanTech investment banking. Before starting EcoTech Capital, he built and led ag tech and food tech banking practices at Citi and Wells Fargo, and set up Wells Fargo’s ag tech cohort inside its St. Louis innovation incubator. Along the way, he has worked on financings and M&A across Citi, Deutsche Bank, Jefferies, JPMorgan, UBS, and Wells Fargo, covering ag tech, food tech, biomaterials, energy, water, and mobility.

The formative thing to know about Adam is what he saw before he ever looked at a farm. As he put it, his first ten years in finance were a crash course in downturns. The Asian financial crisis, the bursting of the internet bubble, and the great recession all landed within one decade. Three of the largest implosions of the last fifty years, all before he got to ag tech. That matters because a lot of the investors and founders active in food and ag today started their careers after 2010 and have never really seen a market reset up close.

In our chat, we spoke about what happens on the other side of the coming wave of food and ag bankruptcies, the trust gap that has opened up between founders and investors, why this sector will never scale like software, and where he is bullish despite all of it.

Let’s jump in!

2026 is the great shakeout year, and only one version of it ends well

Adam expects 2026 to bring the largest wave of bankruptcies food and ag tech has ever seen. A meaningful number of companies are going to shut down, and the sector will split into one of two scenarios on the other side.

The hopeful version looks like this. A handful of companies emerge with viable business models and a real path to profitability. The capital that is still in the sector consolidates around them as clear category leaders. A few reach EBITDA positive, some exit, and private equity money starts flowing in behind the specialists.

That re-opens the sector to new LP commitments and puts food and ag back on the steady growth path it was briefly on in 2018 and 2019, before the 2020 capital flood pushed valuations to levels no financial model could justify.

The unhopeful version is that no clear winners emerge. Exits stay thin. LPs look at the numbers and walk away. “It is going to be very difficult to justify to LPs why they should come into a sector that raised $40 billion between 2018 and 2024 and returned probably less than $5 billion,” Adam said. That is not a pitch many allocators are lining up to hear.

Which path the sector takes is not really about how many companies fail. It is about whether the ones that survive can prove that disciplined, profitable growth in food and ag is achievable. If a small group does, the capital comes back.

An advisor can tell founders what an investor cannot

Adam’s day job looks different from most of the people I speak to. He is compensated on capital raised, not on the valuation he raises at, and he is not sitting on a board defending a mark. That changes what he can say out loud to a founder.

Founders and investors, he argues, are wired to be optimistic. They have to be. But that optimism makes it hard for either side to sit with risk for long. An advisor has room to take a more holistic view, walk through the risks in detail, and push back on decisions that look good on a pitch deck, but cost the company later. He does not have skin in the game the way an investor does, and he is the first to say so, but the distance is part of what makes the advice useful.

Nowhere is that clearer than on valuation. “Maximising valuation is an exercise you should be doing on an M&A sale,” Adam said. “It is not always your best outcome when raising capital, because you might be raising too much of it at too high a valuation.” Raise at a price the business cannot grow into, and the only way the investor gets their money back is by pushing the founder to run at a pace the company was never built for. Many companies chase a short-term win on price and run out of road eighteen months later.

Investor FOMO is dead, and trust has to be rebuilt one milestone at a time

One of the most structural problems in food and ag right now, according to Adam, is trust. Over the last decade, too many companies told investors they would hit milestones they never hit, and the market is still paying for it. “FOMO is dead,” Adam said. Investors are happy to open diligence, sit in it for months, and watch founders prove out what they said they would do before committing a dollar.

His advice to founders raising today is to create monthly or bi-monthly milestones. Tell the investor exactly what you will deliver and by when. Then go and do it, come back, and do the next one. Raising in this market is not about selling the vision any more. It is about building a small, public track record of hitting commitments so that the big one becomes believable.

He also warned against the old pitch muscle memory. In 2020-2022, investment decks were built around upside cases and “investment highlights”. In the current market, what matters is risk factors and how you plan to mitigate them. A founder who can walk an investor through their top three risks and name a strategic partner mitigating each one is in a different conversation from one still pitching hockey stick charts.

Founders should also brace for tighter runway expectations. Eighteen to twenty-four months timelines for use of proceeds are no longer the default. A lot of rounds today are sized for twelve months, with the promise of more capital if the milestones land.

Founders keep building what farmers never asked for

There is a second trust gap that gets talked about more often, but is acted on far less. It is the gap between what founders think farmers want and what farmers actually want.

Adam’s favourite illustration comes from a farmer and co-op operator he used to work with, who told him the same thing every time they spoke. What he really wanted was an ag tech company that could pull rocks out of his field. That was his biggest operational headache, and he had told plenty of founders about it. No one ever built it, because no one found the problem sexy enough.

The point is not really about rocks. It is that a lot of products in this sector get built on the founder’s assumption of what a farmer needs, not on what a farmer has actually said they will pay for. Adam told me he used to ask early-stage founders a simple question whenever they pitched him. “Is it a big problem for them, or is it something that’s a nice to have, not a need to have?” In a lot of cases, the founder did not know, because they had not asked. They were confident the problem existed. They had not confirmed it was worth solving.

The second piece of the farmer story is risk. A farmer’s field is their livelihood. If a new technology kills a crop or knocks down yield, that is a direct hit to their income for the year, with no rerun until the next season. No other industry asks an operator to stake a full year of revenue on an unproven product from a startup. Proving anything out in a field takes time, and it takes more time than most investors and founders have been willing to wait.

Subscribe now

Calling it tech does not make it scale like tech

The word “tech” is doing a lot of work in how this industry talks about itself. In practice, what is really being built is industrial ag and industrial food. The buyer, whether a farmer or a food manufacturer, moves slowly. Farmers buy on an annual cycle tied to planting season. Miss it, and you wait twelve months for another shot. Enterprise software does not work that way, and valuing a company as if it does is how companies run into issues.

Adoption inside a single farm also takes years, not quarters. A typical winning product might land on 1 to 2% of the farm in year one, 5 to 10% in year two, 20% by year three, and full penetration by year seven or eight if everything goes right. “Its ability to scale is going to be predicated on getting a farmer to purchase it, or getting a consumer to buy a new product. That is not easy.” A SaaS rollout inside an enterprise is more binary. An ag rollout is a multi-year crawl through field trials, weather, and the farmer’s own risk tolerance.

The exit math matches the adoption math. There are no big IPOs coming in ag tech. The last real one, Adam pointed out, was probably Monsanto, about thirty years ago. Most exits happen through M&A, usually under $500 million and often under $250 million. Strategics, like the major input and equipment companies, are not lining up to pay premium prices for unproven technology. They will buy the tech when it is ready, bring the team along, and budget another $50 million of R&D on top to get it to scale.

The bullish case starts with what makes the farmer more money

For all the caution, Adam is bullish on a handful of areas, and the logic underneath all of them is the same. Each one has a clear path to providing farmers with a positive ROI in one season or less.

The first is automation & robotics and digitisation, with one important caveat. The winners in the short term will be additive automation, tools that make an existing picker or farmer more productive, not systems trying to replace farm labour entirely. Humanoid robots in the field are a long way off.

But the structural pressure from shrinking labour supply and rising labour costs is a real challenge in every developed country, and it is not going away. “Automation & robotics needs to be there to replace farm labour,” Adam said, and not just in the US. Picking crops in the heat is not going to win the recruiting war against a fast food chain paying the same wage with better working conditions.

The second is AI in the field. He has watched computer vision tools cut pre-harvest crop loss roughly in half at client operations, from 25 to 30% down to 10 to 15%, which on a thin-margin farm can move the business from breakeven to a 20% EBITDA margin. The only catch is psychological.

As Adam puts it, “a grower is going to see the data that shows them that they’re not doing a great job farming. That’s the only way they’re going to be able to improve performance.” Nobody loves being told they could be doing their job better, but the economics are hard to argue with.

The third bullish bet is advanced irrigation, from an unusual angle. Adam thinks the hyperscalers building new data centres could end up funding farmers to switch from flood to drip irrigation, because the water savings can be redirected to cooling data centers at scale. That is a capital source the sector has not really priced in, and it could move advanced irrigation faster than another round of climate funds ever did.

The fourth is alternative proteins, but not where most of the capital has gone so far. Cultivated meat, in his view, is at least a decade away from solving its scaling and cost problems. He is more interested in what he calls “future proteins” more broadly, where global demand for animal protein is outrunning supply, and the gap has to be filled by something. The technology is still early, and the taste is not there yet, but the midterm direction is real.

The bear case starts where the ROI story falls apart

On the other side of the ledger, Adam has concerns about parts of ag biotech, especially biologicals. The issue is not the science, but the challenge of showing a grower a short-term positive ROI. A great biological product in a bad weather year will look mediocre, and a mediocre product in a great year will look fine. When the signal from your own product is that noisy, farmers have no structural reason to adopt.

He is also watching the traditional CPG sector brace for the GLP-1 wave. People on GLP-1 eat roughly 80% less dessert, and the knock-on effects for the companies filling the middle of the grocery store are significant. The legacy CPG names either reinvent themselves around healthier, GLP-1 compatible products, or they watch a meaningful chunk of revenue disappear. Adam is bullish on the health outcomes overall. He is also honest that he does not yet know which specific businesses will emerge as winners.

Underneath both calls sits the deeper point Adam kept coming back to across our conversation. Food and ag are held to a higher standard than almost any other startup industry. The “move fast and break things” playbook does not apply, because the cost of breaking things is paid by a farmer’s harvest or a shopper’s plate. “If it is a break it, you’re probably sidelined for a decade,” Adam said. Companies in this sector need to launch when they are ready, not when the pitch deck says they are.

Want to connect with Adam?

Adam is open to hearing from founders and investors navigating this market. You can reach him at adam@ecotechcap.com

If you found value in this newsletter, consider sharing it with a friend who might benefit from it!

Share

Or, if someone forwarded this to you, consider subscribing.

Subscribe now

Disclaimer: The views and opinions expressed in this newsletter are my own and do not reflect those of my employer, affiliates, or any organisations I am associated with.

Thanks for being here. For Issue #144 of Better Bioeconomy, I sat down with Justin Ahmed, Director at Beanstalk AgTech.

Beanstalk is a Melbourne-headquartered innovation firm that operates across two arms: Beanstalk Advisory, which works with agribusinesses, governments, and development partners across the Indo-Pacific, and Beanstalk Ventures, which builds and invests in impact-focused agrifood companies. Founded in 2017, the firm now has a presence in Sydney, Perth, Brisbane, Hobart, and Singapore, and has delivered programs across Australia, Singapore, Vietnam, Indonesia, and India.

Justin leads the firm’s work translating ag tech innovation into commercial outcomes. Before Beanstalk, he was a project leader within McKinsey’s Global Agriculture practice, with prior roles at Enclude (now Palladium), the Syngenta Foundation, and CIMMYT. He has on-the-ground experience across North America, Africa, South Asia, and Australia, which gives him a sharp view of how ag tech moves between geographies.

In our conversation, we talked about why ag tech innovation rarely fails on the science, what separates Australian and Southeast Asian commercialisation pathways, where corporate-startup partnerships break down, how founders should think about non-dilutive capital, and what makes Justin both bullish and bearish on the sector right now.

Multiple valleys of death sit between ag tech science and the market

Most ag tech innovators frame their commercialisation challenge around a single valley of death, the gap between R&D and market. Justin pushes back on that framing. In his read, the gap is a sequence of separate chasms, each requiring a different kind of partner to cross.

“The biggest gaps in agrifood innovation rarely sit in one single place,” he said. “It’s a game of the weakest link. They sit between the science and the market, between pilots and procurement, between public sector ambition and private sector execution, between the capital models that already exist and what’s really fit for purpose.”

The work of bridging those gaps is the translational layer. It’s the connective tissue that turns a working technology into a working business: the partnerships, contracts, capital structures, regulatory pathways, distribution agreements, and customer relationships that move an innovation from “this works in a lab” to “this is being used and paid for at scale.” The technology, the customer, and the capital each sit on their own side. The translational layer is the infrastructure that connects them, and it’s the layer that most ag tech ecosystems leave unbuilt.

That framing is what shapes Beanstalk’s structure. The firm sits across advisory, venture building, and investment advisory because each discipline addresses a different valley. Advisory work helps governments, corporates, and donors define the right problems. Venture building converts those problems into commercially grounded companies. Investment advisory matches those companies to capital that fits their stage, risk profile, and growth pathway.

The stakes of getting this translational layer right are large, and Beanstalk has the data to show it. In the firm’s State of the Digital Agriculture Sector report covering low- and middle-income countries, the team mapped ~1,400 active digital agriculture solutions across Latin America, South Asia, Southeast Asia, and Sub-Saharan Africa, and modelled a roughly $450 billion per annum gap between a “thriving” and a “derailed” decade for the sector.

A more recent report, The Opportunity for AgriTech Investment in Southeast and South Asia, authored with Omnivore and Briter Bridges, narrows that lens to the SEASA region: agritech adoption could unlock around $90 billion per annum in Southeast Asia and around $160 billion in South Asia by 2033. The variance comes down to whether the systems around the technology, including capital, policy, partnerships, and trust, are functional enough for adoption to compound. Technology readiness is rarely the limiting factor.

What separates ag tech that scales from those that don’t

Across both Australia and Southeast Asia, Justin sees a single pattern that distinguishes the ag tech companies that reach commercial scale from the ones that stall. Founders who succeed stay extremely close to customer economics. They understand what moves the margin for their customer, even when those drivers vary widely by geography, crop, or value chain position. The practice of building a commercial model around where margin is being created looks the same everywhere, even when the inputs change.

Great technology, in his view, is the easier part. “Commercialisation is a separate skill from innovation,” Justin said. “A lot of the companies that we see that have struggled, we’d say waited too long to invest in that.” The ones that scale invest in commercial capability early, alongside the technology, rather than treating it as a problem to solve once the science works.

The market-specific differences sit on top of that common foundation, and they shape what “commercialisation” looks like in practice. Australia, in Justin’s framing, looks more like Canada or Eastern Europe than Asia. It is highly professionalised, tech-intensive, consolidated, and farmer-driven. Profit pools sit with large commercial producers. That structure means revenue realisation, customer acquisition, and after-sales support all scale on a different effort curve than they do across smallholder Asia.

Justin pointed to SwarmFarm Robotics and DataFarming as examples of Australian startups that have grown well in that structure. Both built early credibility with rural and regional ecosystems, leaned hard on Research and Development Corporations (RDCs), agronomist groups, and grower bodies, and earned distribution leverage through those channels first.

But there’s a catch. The real unlock for Australian ag tech is almost always getting out of Australia. The domestic market is too small to support a venture-scale outcome alone. So Australian innovators end up in California, Texas, or the Canadian Great Plains, wherever there are large-scale commercial farms that resemble what they built for at home.

Beanstalk’s original thesis was that Southeast Asia could be a strategic alternative for Australian innovators, leveraging Australia’s regional positioning. The reality has been more nuanced. A solution built for 5,000-hectare Australian commercial farms doesn’t pivot easily to 1-hectare smallholders in Java. The innovators who have had real success in Southeast Asia tended to flip the question, asking “Is my value proposition actually better in this context?”, and then committing to localisation rather than treating the region as a secondary market.

Southeast Asia rewards a different posture from the Australian playbook. “It’s all about listening, iterating, adapting,” Justin said. “Not just developing the solution that solves the right technical problem, but really understanding the complex reality in which it’ll operate.” Founders who do well over-invest in literacy and market education, build trust at the pace of seasons rather than quarters, and accept that scaling deeply and slowly beats scaling shallow and fast.

The data backs this up. In a recent baseline study Beanstalk conducted with the Indonesian government, surveying ~1,300 respondents on access, use, and benefit from digital ag tech, where ~80% of active users self-reported real benefits in productivity, cost savings, and reduced operational complexity. The technology delivers when it gets through. The harder problem is building trust fast enough for adoption to take hold.

Corporates should be looking for complementarity, not capabilities that mirror their own

Beyond the farm gate, ag tech adoption blockers sit at the interface between startups and agribusinesses, and Justin’s read is that corporates and startups frequently misalign on what a partnership is even for.

The mismatch shows up in two distinct archetypes of corporate engagement. The first is corporates running venture vehicles, where the question is what they’re investing in and why. The second, and where Beanstalk does most of its work, is corporates trying to integrate new technology into their core business. That second archetype is where the bigger gap lives.

What corporates say they want is fairly consistent: proven impact, real traction, a credible business behind the science. Smart corporates do look for that, but they look for it in the form of depth of impact for a small set of core customers, rather than the broad market validation many startups try to lead with. The trap is what comes next.

Corporates often evaluate startups against capabilities that mirror their own internal capabilities, when the bigger opportunity is to find genuine complementarity. What corporates uniquely bring to a partnership is a commercial ecosystem (manufacturing assets, distribution, customer access) that can shift a promising technology from trial phase to real scale. Treating a startup as a smaller version of an internal team forecloses that pathway.

The other failure mode sits inside the corporate itself. Companies often look for “plug and play” solutions, underestimating the readiness required within their own teams to use them. A new ag tech tool requires culture and capability inside the corporate to define, deploy, and build around the technology. The right tech partner understands that the corporate’s own staff need to be taken on a journey, with support on how they use the solution, before any pilot translates into adoption.

Justin’s takeaway for both sides: corporates need to be willing to be the test bed, and startups need to design with strategic partnerships and commercial pathways in mind from day one. That maps to a thesis Beanstalk laid out in its recent report, that the ag tech most likely to succeed in the region is the one that solves corporate challenges from the outset and builds strategic partnerships into the growth pathway from day one.

Subscribe now

Most ag tech founders pick the wrong capital and define the exit too late

Capital strategy is another area where some of the most consequential founder mistakes happen. There is no single right capital stack for ag tech, since the right answer depends on sales cycle, development cycle, capital intensity, customer acquisition rate, and how revenue is realised. But founders who get it wrong tend to get it wrong in the same direction.

The reactions to Beanstalk’s Omnivore-commissioned report told the story. Investors found it sobering. Founders found it deflating. Justin’s reading is that both reactions reflect a misunderstanding of what the report was scoped for, which was specifically the venture capital opportunity in ag tech. That’s a narrower pipeline than the full set of viable ag tech businesses.

As he put it, perhaps 5% of agrifood businesses have the kind of profile that warrants a venture capital outcome, meaning capital-light enough, scalable enough, with valuation upside in line with where venture multiples now sit. Most ag tech sits outside that profile, and the right capital path for those companies is different instruments entirely.

That recognition shapes how Beanstalk works with the ventures it builds. Rather than running a fixed-fund model that pushes everything toward venture-style returns, the team works with founders to identify the right capital pathway and link them to capital with the structure and return expectations that match.

The discipline founders apply to picking equity investors, including finding the right partner and evaluating fit, often disappears when it comes to grants. “People take a spray-and-pray approach to getting grants and public funding,” Justin said. “And that’s really dangerous.” Every minute a founder spends on the wrong grant is wasted effort. And it’s just as easy to lock yourself into a non-dilutive relationship that drags you off your growth pathway as it is on the equity side.

The other recurring mistake is treating exit strategy as something that gets defined at Series B. Particularly in Southeast Asia, where exit pathways are thinner, and acquirer universes are smaller, Justin argued that founders need to map their likely acquirers from day one, even when an IPO is the eventual goal.

That mapping informs everything downstream: which capital to take, which partnerships to build, which milestones to optimise against, and which corporates to position as future strategic investors before they become acquirers. Capital pathway and exit pathway are the same conversation, in his view, and treating them as separate ones is what locks founders into structures they later have to unwind.

Government-led ag tech adoption programs that work are the ones built to outlast political cycles

In addition to working with founders and corporates, a meaningful share of what the firm does sits with governments across the Indo-Pacific, helping them design and run ag tech adoption programs.

Government adoption programs are where Justin sees the widest variance between intent and outcome, and where political cycles make the right answer hardest to deliver. The programs that move the needle, in his view, share two things: a long enough timeline to match how agriculture works, and a shared definition of success that can survive across departments and government cycles.

The timeline piece is structural. Agriculture is a slow relationship-building environment by nature. Adoption takes years and is built on real trust. Seasonality matters. A technology has to demonstrate capability across multiple cycles before farmers commit. Programs running on a two-to-three-year horizon, or shorter, rarely deliver on that timeline.

Australia, Justin argued, is underrated on this front. The country runs a model called Cooperative Research Centres (CRCs): ten-year investments in collaborative research that pull innovation out of universities and into industry ecosystems. That timeline is what it takes to move past transactional engagement and build the trust required for ecosystem-level coordination.

The shared-definition piece is harder to see, and Beanstalk’s work with the Indonesian government on its digital ag tech policy and investment roadmap is what brought it into focus. When local agritech startups raised hundreds of millions of dollars a few years ago, the government wanted to know whether headline funding rounds were translating into real impact for smallholder farmers, or whether they were vanity metrics dressed up as success.

Answering that question turned out to be the work itself. Before any pilot could be designed, the government needed a common view across departments and agencies on what impact meant for the sector and how it would be measured. That work took the form of a dashboard that the relevant agencies could track and manage, built into the systems they already used.

The shift, from defining impact in a report to defining it in a tool the government uses every day, is what makes subsequent programs cumulative rather than parallel. Without it, departments run programs against different definitions of success, and even individually successful pilots fail to compound into ecosystem-level change.

Bear case is in the narrative, bull case is on the ground

Zooming out, Justin offered a clear split between what worries him and what gives him conviction.

The bear case is narrative. “Venture and capital is a land of narrative in many ways,” he said. He’s seen sectors outside ag tech find their bottom and start to rebuild after the global capital downturn. Ag tech, he’s not sure has reached that point yet. His worry lies in how generalist investors read the space. The headline narrative says ag tech has underperformed, and generalist allocators who haven’t been close to the sector tend to take that at face value rather than digging into where the real opportunity sits. That self-reinforcing cycle is what the Omnivore report was partly written to challenge.

The bull case is, paradoxically, on the supply side. Founders are adapting to the new reality faster than investors are. They’re more disciplined, more focused on learning from the mistakes of the first wave of ag tech, and increasingly clear-eyed about what kind of capital they need. Every second-tier and third-tier city Beanstalk visits in Indonesia turns up another 30 innovators with real energy.

What gets Justin most excited, though, is the role governments in the region are starting to play. Vietnam, Indonesia, Cambodia, Thailand, and Singapore all have governments that he sees as forward-thinking on ag tech, working out where they can be a catalyst in the ecosystem they’re trying to build.

That kind of government engagement, paired with disciplined founders, is what closes the translational layer he kept returning to throughout the conversation: the missing infrastructure between science and market, pilots and procurement, public ambition and private execution. The technology isn’t the bottleneck. The systems around it are, and they look more buildable now than they have in a long time.

Want to connect with Beanstalk?

For funders or investors looking to have more impact in the agrifood-tech ecosystem, reach out to Justin at justin@beanstalkagtech.com, follow their LinkedIn, or sign up for Beanstalk’s newsletter.

If you found value in this newsletter, consider sharing it with a friend who might benefit from it!

Share

Or, if someone forwarded this to you, consider subscribing.

Subscribe now

Disclaimer: The views and opinions expressed in this newsletter are my own and do not reflect those of my employer, affiliates, or any organisations I am associated with.

In Issue #143 of Better Bioeconomy, I want to explore a question that keeps surfacing as I read about AI-mediated purchasing: what happens when AI agents start filtering food choices before consumers see a shortlist?

Most commentary on AI and food retail focuses on the final act: the autonomous grocery order. That story is real, but it is not yet mainstream. The part that is already at scale, and underexamined, is the filtering step.

Before AI buys food for you, it is already deciding what gets considered. The shift underneath is structural: value is migrating from persuasion to proof, and legibility, the ability to produce machine-readable evidence of what you claim, is becoming the new shelf space.

The optimistic version of this story says quality finally gets rewarded because machines can see it. I think the direction is right. But the distributional question keeps pulling me back. The proof economy has a monetisation layer, a definitional politics, and a cost floor that together determine who participates and who gets excluded. That is what I want to sit with in this piece.

Let’s dig in!

AI filtering is already reshaping what gets considered

The share of US shoppers using AI assistants jumped from 12% to 35% in a single year (Adyen 2026 Retail Report, Censuswide survey, n=2,000). The definition is broad and captures everything from product research to checkout delegation, but even accounting for that breadth, the year-on-year move signals a behavioural shift that has already happened.

The pattern extends beyond the US. A Deloitte Asia Pacific report found nearly three-quarters of consumers in the region already use AI to discover and compare products, and 29% of APAC consumer businesses have adopted agentic AI, a figure projected to reach 76% within two years. At GrabX 2026, Grab launched an AI shopping agent that builds a grocery cart from a photo or voice note, with smart substitutions across merchants, no item-by-item browsing required.

Most of that activity is filtering, however, not directly buying. A 2025 Attest survey found 63.8% of consumers had used AI tools for food-related activities, with the majority concentrated in recipe recommendations and meal planning rather than checkout delegation. And when researchers at Yale, Columbia, and the University of Chicago tested what happens to products inside agent-mediated selection (featured in Kantar’s “What Is Your AI Agent Buying?” analysis), they found a 20-40% reduction in selection probability when a key product attribute was missing from the structured data.

Autonomous AI purchasing, meanwhile, remains structurally constrained. A Bluestone PIM survey from March 2026 found 73% of consumers uncomfortable with AI completing purchases autonomously, 68% assuming results are commercially influenced, and 65% refusing to store payment details. These are concerns about delegating spending authority and exposing household consumption patterns, the kind that do not go away just because the technology improves.

But filtering requires no delegation. The user asks a question, the agent answers, the user decides. The criteria are machine-readable or they are invisible. Packaging design, brand origin story, and shelf position, none of these register inside a filter that needs documented certifications, structured nutritional attributes, and composition data. That behaviour is already at scale, and it does not require autonomous purchasing to reshape which products get considered.

The Kantar figure is the one I keep returning to. A product losing 20-40% of its selection probability because its attributes are not machine-readable is facing a structural disadvantage. Which raises two questions. What determines whether a product is legible to the filter? And who controls the infrastructure that makes legibility possible?

When machines shortlist, value migrates from persuasion to proof

Here is a useful test for a food brand right now: if an agent filtered your product tomorrow, would it find machine-readable proof of what you claim, or would it find a story on your packaging?

If the answer is a story, the product is structurally disadvantaged in agent-mediated commerce. Better creative does not help a product score inside a filter that requires documented certifications, structured nutritional attributes, provenance chain records, and composition assays. The selection penalty that the Kantar study documents is what that disadvantage looks like in practice.

I find it useful to think about the infrastructure that processes these structured inputs in three layers.

At the top sit the agent platforms: OpenAI, Perplexity, Anthropic, Amazon Rufus, Walmart Sparky, and Instacart Cart Assistant. This is the interface layer, the surface where consumers interact, and shortlists get assembled. It is the layer most commentary focuses on because it is visible, named, and contested at high valuations.

At the bottom sits upstream measurement: soil sensing, nutrient assays, spectroscopy, and satellite imagery. This layer generates raw data about what is in the soil, what is in the crop, and what environmental and agronomic conditions shaped the harvest.

In the middle is what I am calling the claims-and-verification layer: the infrastructure that takes upstream measurements and translates them into machine-readable claims that agent platforms can query. This layer spans product content syndication, supply chain traceability, standards frameworks like GS1’s Digital Product Passport standard, and scoring systems that translate attribute data into structured outputs. These are different entities serving different functions, but they share a structural role: they manufacture the legibility signal that agents consume.

Antony Yousefian has argued the directional case well in The End of Brand Trust and The End of the Impulse Buy: verification costs collapsing, brand equity eroding, the farmer and soil layer reclaiming value. I agree with that direction. What I want to sit with here is where value concentrates inside the verification economy once it arrives, and who gets excluded from it.

My working view is that the translation step is where pricing power is most likely to sit, because it is the step neither the agent platforms nor the farms can fully own on their own. Platforms need verification but have incentives to commoditise it. Farms generate data, but most lack the aggregation infrastructure to translate it into agent-readable claims at scale, whether due to fragmentation, heterogeneous data formats, or the cost of standardisation across operations.

That is the clean version of the proof economy: quality that can prove itself gets rewarded, and the translation layer captures margin for making the proof legible. But three forces complicate the picture, and each one tilts the proof-equals-access equation in a different direction.

Three forces shift the proof-equals-access equation

1. Agent monetisation turns the shortlist into an advertising surface

The parallel I keep reaching for is SEO. In the early years of search, the dominant assumption was that good content, accurately indexed, would surface organically. Paid search gradually displaced organic results in most high-commercial-intent categories, and what emerged was a two-tier system where relevance determined eligibility and spend increasingly determined position within the eligible set.

Agent commerce is building a similar infrastructure. Stripe’s Shared Payment Tokens let agents pass scoped payment credentials to merchants. OpenAI introduced Instant Checkout in ChatGPT, then shifted toward directing shoppers to merchant storefronts, with Shopify building its own agentic storefront offering. Instacart’s Cart Assistant runs on the same commerce stack that underpins Instacart’s retail media business. Walmart Sparky now operates inside the Walmart app, ChatGPT, and Gemini, with checkout kept on Walmart’s own rails.

An agent monetised through retailer ad networks and platform revenue shares may not function as a neutral representative of the consumer’s dietary intent. Its shortlist would reflect some equilibrium between consumer preferences and platform economics, optimising for health within the constraints of whoever is paying for the infrastructure. That is a different dynamic from the rational health-optimiser that some of the more optimistic commentary assumes.

Machine-legibility gets a product into the eligible set. If agent commerce follows the search pattern, the position within that set will increasingly reflect spend.

Subscribe now

2. Whoever defines “healthy” controls the filter

When a user tells an agent “keep it healthy” or “low-inflammatory,” a translation has to happen. The natural-language instruction must map onto filter criteria applied to a product database. That mapping is unlikely to be neutral, and the entity that controls it shapes the outcome for the products downstream.

The FDA finalised an updated “healthy” nutrient content claim in December 2024, its first revision since 1994. It is a labelling standard. It is not a product-level scoring API that an agent can query in real time. When an agent operationalises “keep it healthy,” it queries whichever database has already done that translation work.

A handful of private systems have built that translation layer, and they do not all agree on what “healthy” means. Yuka scores products with nutritional quality weighted at 60%, additive content at 30%, and organic status at 10%. Open Food Facts uses Nutri-Score, scoring on a nutrient-only basis. HowGood is primarily a sustainability intelligence company, scoring on eight environmental and social metrics and integrating Nutri-Score separately for nutritional context. Personalised nutrition apps like Zoe score the same product differently for different users, calibrated to individual metabolic response.

Take a full-fat organic yoghurt with no additives. It scores well on Yuka (nutrition plus clean ingredients), variably on Zoe (depends on the user’s metabolic profile), and differently again on HowGood (depends on the sustainability footprint of its supply chain). An agent shortlist filtered for “healthy” produces different results depending on which system it queries, and the consumer has no visibility into which taxonomy shaped the shortlist.

The analogy that keeps pulling me is pharmacy formulary placement. In pharmaceutical distribution, a drug’s presence on a formulary often determines patient access more directly than its clinical profile. The formulary decision happens upstream of the prescription. In agent-mediated food commerce, the taxonomy endorsement happens upstream of the shortlist. A product’s presence in the scoring system of a given platform's queries may matter more to its agent-era visibility than its placement on any retailer’s digital shelf.

3. Platform fragmentation multiplies the legibility target

The optimistic version of this transition assumes interoperability: one proof layer, universal reach. That assumption is not holding up.

Last month, U.S. District Judge Maxine Chesney granted Amazon a preliminary injunction blocking Perplexity’s Comet browser agent from accessing password-protected areas of Amazon, including logged-in Prime experiences. A brand’s presence inside an authenticated Amazon shopping flow, the one with rich personalisation data, is not addressable by agents operating outside Amazon’s perimeter. The emerging protocol standards reinforce the pattern. OpenAI’s Agentic Commerce Protocol and Google’s Agent Payments Protocol (AP2) each require separate integration work, and neither is emerging as a universal standard.

In Southeast Asia, Google and Sea Limited announced a partnership to build an agentic shopping prototype for Shopee, which holds 52% of the region’s e-commerce market, and to pilot agentic payments through AP2 via Sea’s fintech arm Monee. That is yet another platform-specific integration a food brand would need to reach agent-mediated shoppers in a different region.

For a mid-sized food brand trying to be legible to agents, this means the target is not one data standard but several, each with its own integration cost and taxonomy. Machine-legibility is not a single investment. It is one investment per platform that maintains a walled experience layer.

Fragmentation also resolves a structural question I raised earlier: whether the claims-and-verification layer retains independent leverage or gets absorbed by platforms. If every platform maintained the same taxonomy, a single acquisition could internalise the middle layer’s margin. In a fragmented environment, no single acquisition gives a platform full coverage. The middle layer retains leverage because it is the translation infrastructure between incompatible systems. The more fragmented the protocol landscape stays, the more durable the middle layer’s independent pricing power.

Taken together, these three forces mean the legibility economy is not a clean meritocracy where proof equals access. Access depends on proof, position within the eligible set reflects spend, and the definition of proof depends on which taxonomy the agent queries.

The proof economy’s cost floor determines who participates

Even setting aside monetisation, definitional politics, and fragmentation, the basic economics of producing machine-verifiable evidence sort food producers by scale.

Certifications, provenance chain documentation, composition assays, soil health records, nutrient density measurements: these are the attributes that move a product from invisible to eligible in agent-mediated shortlists. Supplying verifiable evidence is the new table stakes for commercial relevance.

The unit economics do not distribute evenly. A soil assay at agronomic resolution has a cost. Spectroscopy-based nutrient analysis at the lot or batch level has a cost. Third-party certification audits also have a cost. Building the data systems that retailer traceability mandates and GFSI-level food safety programs increasingly require its own investment.

The EU’s Ecodesign for Sustainable Products Regulation (ESPR), which introduces Digital Product Passports for priority sectors like textiles and batteries first, signals where food and agricultural products are likely headed next. At scale, the per-unit cost of all this is manageable. At five acres, it may exceed the margin the producer is working with.

The consequence is a sorting mechanism embedded in the proof economy’s structure. Mid-sized producers with the operational capacity to implement sensing infrastructure and the volume to amortise it gain access to agent-mediated shortlists. Artisanal and small-scale producers, the kind of long-tail supply that specialty food brands and farm-to-table operators have built sourcing differentiation around, face a proof layer that is technically accessible but economically out of reach at current cost structures.

Where this thesis might be ahead of itself

If sensing costs follow the trajectory of other AI-adjacent hardware, falling significantly per cycle as compute and miniaturisation improve, the cost floor argument may be temporary. The distributional effects I describe could be a transitional feature of early infrastructure rather than a permanent structural condition.

If platforms consolidate around a single protocol, with GS1’s Digital Product Passport being the most plausible candidate, fragmentation collapses and the middle layer’s pricing power weakens. A unified standard would simplify legibility into a single investment rather than one per platform, which is better for producers but changes the value map I have outlined.

And if agent monetisation evolves toward subscription models rather than advertising, the shortlist may stay closer to consumer intent than the SEO parallel suggests. The search pattern is one possible outcome, and it may not be the dominant one.

The direction of the shift, from persuasion to proof, is clear to me. The speed, the severity of the distributional effects, and whether the cost floor proves durable or temporary are the open questions.

What this means for you

Every food brand needs a legibility audit before agents run one for them

The diagnostic question I posed earlier in this piece, whether an agent would find machine-readable proof or a story on your packaging, is worth turning into an exercise.

Work through it: what structured nutritional attributes does your product have in queryable databases? Which scoring taxonomies (Yuka, HowGood, Nutri-Score) can access those attributes? Which agent platforms can see your product data inside authenticated environments? This is not a five-year planning horizon. The filtering is already happening.

The most defensible infrastructure plays sit in multi-protocol translation

For investors evaluating data infrastructure in this space, the fragmentation argument implies a specific thesis. The most defensible companies are those whose value increases with fragmentation: multi-protocol translation capability, cross-platform data syndication, taxonomy interoperability.

These companies survive consolidation (they become the acquired integration layer) and thrive during fragmentation (they are the only path to multi-platform legibility). This is a different filter from “invest in the agent platforms” or “invest in the sensing layer,” both of which face commoditisation pressure from different directions.

Whether legibility becomes a public utility determines who participates

The cost floor and the definitional politics I have described will not self-correct through market dynamics alone. Three interventions would change the distributional math: cooperative data infrastructure that lets small producers pool sensing and certification costs across shared operations, public scoring standards that prevent private taxonomies from becoming the sole gatekeepers of “healthy” or “sustainable,” and protocol-level interoperability mandates that reduce the per-platform cost of legibility.

These are policy choices, and they need to be made before the architecture hardens. The sector needs to decide whether legibility is a public utility or a private good. That decision will shape whether the proof economy creates a more open food system or reproduces the concentration dynamics of the shelf-space era in a new form.

If you found value in this newsletter, consider sharing it with a friend who might benefit from it!

Share

Or, if someone forwarded this to you, consider subscribing.

Subscribe now

Disclaimer: The views and opinions expressed in this newsletter are my own and do not reflect those of my employer, affiliates, or any organisations I am associated with.