On 7 September 2026 Nourish Ingredients said it had completed a 10-ton industrial production run of its animal-free specialty lipid Creamilux at SD Guthrie International Speciality Ingredients in the Netherlands, and that the run had “effectively removed manufacturing as a barrier to growth.”
In the same interview, chief executive James Petrie put the partner’s capability at “up to 20 tons a day.”
Those two statements are both true and, read together, they say something the announcement does not: the run that proved the manufacturing question was half a day of one facility’s stated capacity. That is not a criticism of the achievement. It is the measurement that explains why the asset-light route worked here and why it will not work everywhere.
The arithmetic the announcement implies
Every figure below comes from the company’s own statements in the 7 September interview.
| Stated quantity | Source statement | As a share of “up to 20 tons a day” |
|---|---|---|
| First industrial run | 10 tons, “completed in a matter of days” | 0.5 days |
| Next stated step | “100 tons” | 5 days |
| Step after that | “500 tons and beyond” | 25 days |
Chief technology officer Anna El Tahchy gave the demand-side conversion: “It’s a potent lipid, so a little goes a long way, when we are looking at around 0.2% inclusion in a finished product.” At that inclusion rate, she said, “the 10-ton run translates into roughly 5,000 tons of finished food product.”
That conversion checks out: 10 ÷ 0.002 = 5,000. It is the reason the whole model holds. An ingredient used at 0.2% needs 1/500th of the tonnage of the food it goes into, so a partner line that would be trivial for a bulk protein is oversized for this product.
Practical consequence: before assuming a co-manufacturing partnership is a capacity solution, divide the partner’s daily rate into your five-year volume plan. If the answer is measured in days, capacity was never your constraint and the partnership has solved a problem you did not have. If it is measured in years, you are in a queue.
This is the case that settles an argument
The industry has spent 2026 disagreeing in public about whether contract manufacturing works. The disagreement has been treated as a difference of opinion. It is better read as a difference of product.
Against. Planetary’s cofounder and chief executive David Brandes, speaking to AgFunderNews in April 2026 after a CHF 16 million equity round supplemented by CHF 6 million in credit, was categorical: “The production infrastructure needs to be owned or co-owned or at least exclusively accessible. In the food space, unit economics are everything and for bulk fermented commodities, contract manufacturing is not viable.” He added that retrofitting existing upstream equipment “rarely works”, does not save substantial capex, and usually drives up cost of goods.
For. Arjen van der Wijk, cofounder of scale-up consultancy Cibus Nexum, told Protein Production Technology International on 4 September 2026 that the sector has moved through three positions and is entering a fourth: “We saw solutions move from CPG products to B2B ingredient solutions. Then companies were saying, ‘I want to have my own plant’. Then they said, ‘No, I don’t want to have my own factory. I want contract manufacturing solutions’.” He expects the next step to be licensed manufacturing — “a combination of both manufacturing and sales and distribution”. On why: “Investors generally aren’t too eager to put their money directly into stainless steel – into factories and heavy infrastructure.”
The reconciliation. Brandes is talking about mycoprotein sold as a bulk B2B ingredient, where the target is a commodity price and every point of conversion margin surrendered to a tolling partner comes out of a thin spread. Nourish is talking about a lipid dosed at 0.2%. Petrie made the distinction himself: “Many of our peers are working on recombinant proteins that face many scale-up challenges whereas we are working with commodity-scale and well understood existing ingredient substrates.” He was explicit about the limit: “It would be very difficult if we needed to produce a commodity-scale product at commodity prices.”
So the two positions are not in conflict. Inclusion rate decides which model works. A product used at 0.2% can hand production to a partner and still clear its economics, because the value per kilogram is high enough to share. A product that has to land near a commodity benchmark cannot, because there is nothing to share.
Practical consequence: the make-or-buy question is answerable before the pitch deck is written. Take the finished-product volume you intend to address, multiply by inclusion rate to get ingredient tonnage, and divide your target ingredient price into it. If the gross margin per kilogram will not carry a tolling fee, an asset-light plan is not a strategy — it is a deferral.
Where this sits against the capacity we have already mapped
This is the fourth distinct route to industrial volume we have documented this year, and the first where the constraint was explicitly not capacity.
| Route | Example | What the constraint turned out to be |
|---|---|---|
| Build your own plant | Believer Meats’ Wilson site | Cost of operation; the asset failed to attract a qualifying bid |
| Take space in an existing fermentation estate | ADM’s Clinton site | Access to a third party’s underutilised capacity |
| Take exclusive space in a pharma estate | AMSilk at Ajinomoto Nesle | Exclusivity — 160 m³ taken, and taken off the market |
| Co-locate with a feedstock owner | mycoprotein next to sugar mills | Feedstock cost and logistics |
| Toll at a specialty lipids maker | Nourish at SD Guthrie | Conversion of trials into orders |
Nourish is also not new to this model. It announced a biomanufacturing partnership with CABIO Biotech in November 2024 for Tastilux, and opened a commercial hub in Leiden in September 2025. SD Guthrie is therefore a second named manufacturing partner rather than a first, and will act as “lead European manufacturing and scale-up partner” while also helping develop customers and distribution — which is close to the licensed-manufacturing structure van der Wijk describes, rather than plain tolling.
Fooditive, profiled by the same publication on 4 September, is pursuing US manufacturing partnerships for casein and 5-KDF on an explicitly asset-light basis. Casein is a bulk protein. Whether that works is the test of the boundary this article draws.
The counter-argument
The strongest case against this reading is that a 10-ton run proves very little. It is a first campaign at one site, and El Tahchy’s own framing was about de-risking rather than volume: “The biggest validation was that the process holds up at industrial scale without the performance drop-off or contamination risks you often see when you move to pilot and commercial scale.” Nothing here demonstrates that the partnership survives contact with a demanding customer, a quality excursion, or a second product competing for the same line.
There is also a real risk in the model Petrie describes. “We won’t scale ahead of demand,” he said; each volume step will be “backed by real orders rather than speculative capacity.” That is prudent while a partner has spare capacity. It becomes a liability if the partner sells that capacity to someone else, because the discipline that avoids stranded capex also forgoes any contractual claim on the tank. Our reporting on the AMSilk arrangement at Ajinomoto Nesle showed what happens when a competitor takes the exclusive: 160 m³ that a site advertised publicly stopped being available to anyone else. Nourish holds a partnership, not, on the public record, a reservation.
What we could not establish
- Whether “ton” means metric tonnes or US short tons. The interview uses “10-ton”, “20 tons a day” and “5,000 tons” without stating a basis. The facility is in the Netherlands and both companies are non-US, which makes metric tonnes likely, and one trade summary describes it as a “10-tonne” run. The primary source we read does not say. Every ratio in this article is unaffected, because the same unit appears on both sides; any absolute tonnage is not.
- Whether the 20 tons a day figure is nameplate, demonstrated, or Creamilux-specific. Petrie says “SD Guthrie’s facility can support up to 20 tons a day.” Whether that is the site’s total specialty lipids throughput or the rate available for this product is not stated. If it is the site total, the share available to Nourish is smaller and the “days of capacity” figures above are correspondingly understated.
- Whether this run involved fermentation at all. Nourish is described as a precision fermentation company, but SD Guthrie’s general manager describes Creamilux as made through “a controlled modification of lecithin at molecular level”, and Petrie contrasts his inputs with the “recombinant proteins” his peers work on. We could not establish from the published material which steps of the Creamilux process are fermentative and which are lipid modification. This matters, because a lecithin-modification tolling deal is weaker evidence about fermentation capacity than the coverage implies.
- Commercial terms, price, or any offtake volume. No contract value, tolling fee, minimum volume or customer name is disclosed. “Active trials with a range of global companies” is the extent of the demand evidence.
- Corroboration from either company’s own channels. As of 8 September 2026, Nourish Ingredients’ newsroom lists nothing about the SD Guthrie run; its most recent entry is dated 17 June 2026. We could not locate a press release from either party and have relied on the trade interview, which quotes three named executives directly.
What to watch
- Whether a 100-ton run is announced within twelve months. Petrie set that as the next step and tied it to offtake conversion. Twelve months without it would indicate the constraint is demand, exactly as the company says.
- Whether “products on shelves within the next six months” — so, by early March 2027 — arrive, and whether any name a customer. Trials that do not convert are the failure mode this model is exposed to.
- Whether SD Guthrie takes on a second alt-lipid client. A specialty lipids line with spare capacity is a contested asset. If it fills, Nourish’s optionality narrows without any change to its own plans.
- Whether Fooditive’s asset-light casein route reaches industrial volume. Casein is a bulk protein sold near a commodity benchmark. If it works there, the inclusion-rate boundary set out above is wrong, and we will say so.