dsm-firmenich is taking Protopia, a torula yeast single-cell protein of minimum 70% protein, from 15,000-litre batches to a 120,000-litre run this year and a 2,000 to 10,000 tonne facility in 2027. Its director of single cell protein, Karim Kurmaly, gave AgFunderNews two price anchors for the same powder.
On aquafeed: “Today 65% high-quality fish meal is around €3,000/ton and we can compete at those levels. Providing it doesn’t drop below €2,000/ton, we’re okay.”
On human food: whey protein has “skyrocketed to $25 to $28 per kilo, and that’s if you can get it.”
Those are the two markets Protopia is aimed at. Converted to a common basis, the reference prices are six to eight times apart. That is not a detail. It is the difference between two entirely different businesses running out of the same tank.
Methodology
All Protopia figures are from the AgFunderNews interview of 17 August 2026 and the dsm-firmenich product page, both fetched and read in full.
The fishmeal benchmark is the World Bank series, specified as Peru fish meal/pellets 65% protein, CIF, US dollars per metric ton. That is the same 65% protein specification Kurmaly cited, which is why we use it rather than a broader feed index. The most recent published month in that series is March 2026, at USD 1,836.55 per tonne.
Currency conversions use the European Central Bank reference rate for 2 September 2026, EUR 1 = USD 1.1578.
Protein-basis conversions divide the tonne price by the stated protein content: 65% for the fishmeal benchmark, 70% for Protopia’s stated minimum, 80% for whey protein concentrate. This is the correction we applied when we found that the same fungal biomass is 50.8% and 40.2% protein depending on the convention used, and it is the correction most price comparisons in this sector skip.
The two benchmarks, on one basis
| Reference | As quoted | Per tonne, USD | Protein content | Per tonne of protein, USD |
|---|---|---|---|---|
| Fishmeal, dsm’s stated current level | EUR 3,000/t | 3,473 | 65% | 5,344 |
| Fishmeal, World Bank benchmark Mar 2026 | USD 1,836.55/t | 1,837 | 65% | 2,825 |
| Fishmeal, dsm’s stated floor | EUR 2,000/t | 2,316 | 65% | 3,562 |
| Whey protein concentrate 80%, low end | USD 25.35/kg | 25,350 | 80% | 31,688 |
| Whey protein concentrate 80%, high end | USD 28.33/kg | 28,330 | 80% | 35,413 |
Take Protopia priced to compete at the fishmeal level dsm named — EUR 3,000 a tonne, or USD 3,473 — and put it on a protein basis at 70%: USD 4,962 per tonne of protein.
Against whey at USD 31,688 to USD 35,413 per tonne of protein, that is a ratio of 6.4 to 7.1. On a straight tonne-for-tonne basis, ignoring protein content, the ratio is 7.3 to 8.2.
Practical consequence: if you are evaluating a single-cell protein supplier, ask which benchmark its price list is built against before comparing it to anything. A quote built for aquafeed and a quote built for sports nutrition are not the same product decision, and a supplier that cites both in one conversation has told you it has not yet chosen.
Why the same tonnage is two different companies
dsm-firmenich says the 2027 facility will make 2,000 to 10,000 tonnes a year, with 400 to 500 tonnes from the middle of 2027 to seed the market. Run those tonnes at each benchmark.
| Annual output | Revenue at fishmeal parity (EUR 3,000/t) | Revenue at whey parity (USD 25,000/t) |
|---|---|---|
| 2,000 t | EUR 6.0m (USD 6.9m) | USD 50.0m |
| 10,000 t | EUR 30.0m (USD 34.7m) | USD 250.0m |
Same plant, same organism, same downstream. A seven-fold difference in what the asset is worth, determined entirely by which market absorbs the volume.
This is why the licensing model matters more than it first appears. Kurmaly’s plan is that dsm-firmenich supplies what he calls the “software” — strain, process technology and application expertise — against a royalty, while licensees supply the steel. “The strain is the magic, not the facility, not the steel and the concrete.” A royalty on 10,000 tonnes of aquafeed protein and a royalty on 10,000 tonnes of food-grade protein are different propositions, and the licensee is the party carrying the capital either way.
Practical consequence: for a prospective licensee, the royalty base is the question to settle first. A per-tonne royalty set with reference to the food market makes an aquafeed plant uneconomic; one set with reference to aquafeed leaves value on the table in food. Ask which market the term sheet’s assumptions were built on.
The floor is not a distant risk
Kurmaly named a specific break-even: fishmeal must not drop below EUR 2,000 a tonne. Companies almost never disclose this, and it is the most useful sentence in the interview.
At the ECB rate, EUR 2,000 is USD 2,316 a tonne.
The World Bank’s benchmark for the same 65% protein specification, at its most recent published print in March 2026, was USD 1,836.55 — about 21% below the floor dsm-firmenich named.
Fishmeal has since risen. Undercurrent News reported Peruvian fishmeal at record highs in China through June and July 2026, with inventories at multi-year lows, following an El Niño disruption to Peru’s anchovy season — the same disruption Kurmaly cited. His EUR 3,000 is consistent with that spike; the World Bank series simply has not published a month since March.
So both numbers are defensible. The point is what the pair implies. The price level at which Protopia stops working is not a tail scenario requiring a collapse. It is roughly where the published benchmark sat six months ago, before a weather event.
| Month | World Bank fishmeal, USD/t | Versus dsm’s stated EUR 2,000 floor (USD 2,316) |
|---|---|---|
| Jan 2025 | 1,580.81 | 32% below |
| Dec 2025 | 1,825.24 | 21% below |
| Mar 2026 | 1,836.55 | 21% below |
Anchovy biomass recovers. When Peru’s season restarts normally, the benchmark returns toward its multi-year range, and that range has spent the last two years between roughly USD 1,580 and USD 1,880 — entirely underneath the floor.
Practical consequence: a fermentation product whose competitiveness depends on a commodity staying above a level it has been below for two straight years is not a cost-advantaged product. It is a product with a hedge on somebody else’s fishery. Anyone underwriting a Protopia licence should model the aquafeed case at the World Bank range, not at the 2026 spike.
The whey number checks out, which is the point
We spent an article establishing what whey actually costs because the sector kept comparing fermentation costs to a price series that does not exist. The conclusion there: USDA’s late-August 2026 narrative range for WPC 80% regular, USD 11.50 to 12.85 per pound, converts to USD 25.35 to 28.33 per kilogram.
Kurmaly’s “$25 to $28 per kilo” lands inside that range almost exactly. This is worth saying plainly: the whey number is right. An executive at a major ingredients house quoted, in kilograms, the figure USDA prints in pounds.
That removes the easy objection — that the whey anchor is inflated to flatter the comparison — and leaves the harder one. Both anchors are accurate. They are accurate about different markets, and the gap between them is the finding.
It also corrects a comparison we ourselves have had to make twice. When we wrote that $2.18 whey is the wrong benchmark, the error was people comparing fermentation protein to dry whey commodity prices rather than to the concentrate formulators buy. Here the error would be the reverse: assuming a product that competes at the fishmeal end can be valued at the whey end.
The regulatory position, which is the real asset
One thing genuinely separates Protopia from most of the field. Because torula yeast has a long history of food use in multiple geographies, dsm-firmenich says Protopia can be used in the US without a GRAS notice and in the EU without novel food authorisation. It is separately assessing India and Brazil.
We set out why in the 15 May 1997 date, not safety, decides which fermentation proteins need EU approval. The Union’s novel food regime turns on whether a food was consumed to a significant degree in the EU before that date, not on a fresh safety judgement. An organism with pre-1997 food history is outside the regime; a functionally identical organism without it faces the six-year path that The Protein Brewery took to authorisation.
That asymmetry is worth more than a price advantage, and it is not reflected in either benchmark above. A competitor with a superior organism and no history of use starts several years and several million euros behind, whatever its cost curve looks like on paper.
Practical consequence: when comparing single-cell protein suppliers, put regulatory status in the same table as price. A product available today at a worse price frequently beats a product available in 2030 at a better one.
The counter-argument
The strongest case against this analysis is that serving two price tiers from one platform is normal, deliberate, and how ingredient businesses actually work.
dsm-firmenich is not a startup discovering its market. It sells carotenoids, enzymes, vitamins and eubiotics into feed and food simultaneously, at wildly different price points, from shared manufacturing. Segmenting a single molecule across tiers by grade, specification and packaging is standard practice. Kurmaly’s claim is explicitly that the strain and process were reverse-engineered from a target cost — “We started in reverse order: if this is the COGS, how do we get there?” — which is the discipline the sector usually lacks.
He also claims 2 to 3 times the productivity of most liquid biomass fermentation systems, a feedstock-agnostic strain that runs on ethanol, methanol, acetate, sugar or molasses, and semi-continuous rather than continuous operation to limit contamination losses. If the cost base is genuinely at the aquafeed end, then every tonne diverted to food is margin, and the fishmeal floor is a downside case for the least valuable slice of output rather than a constraint on the business.
That is a coherent strategy. It is also unverifiable from outside: dsm-firmenich has published no cost per kilogram, and a claimed COGS target is not a disclosed COGS. What can be checked is that the company chose to name a floor. A business confident of clearing the aquafeed benchmark by a wide margin does not usually specify the point at which it stops working.
What we could not establish
- Protopia’s actual price or cost. Neither is published. Every figure in this piece is a reference price for a competing ingredient, quoted by dsm-firmenich or by the World Bank, not a Protopia quotation.
- What “high-quality” fishmeal means in Kurmaly’s EUR 3,000. It may denote a premium grade — steam-dried, low-TVN, super-prime — trading above the World Bank standard 65% benchmark. If so, part of the gap between EUR 3,000 and the March print is specification rather than timing. dsm-firmenich did not specify a grade.
- Current fishmeal prices. The Undercurrent report of record highs is behind a subscription and we could not read it. We cite it only for the fact of the report, which is in its published headline and standfirst, and we take no price figure from it. The World Bank series remains the last figure we have actually read.
- Whether Protopia is priced per tonne or per unit of protein. We converted both ways and showed both. Which convention dsm-firmenich uses commercially is not public.
- The 400 to 500 tonne figure’s exact year. The August 2026 interview says “from the middle of next year onward,” which we read as mid-2027, consistent with the stated 2027 facility. It is our reading of “next year,” not a stated date.
- Indian volumes. The claims about India’s yeast protein imports being limited to 10,000 tonnes a year from a neighbouring country, with 11,000 more coming, and about 20 billion litres of ethanol capacity built for USD 5 billion, come from a single interviewee, Raman Chandrasekar of Sustein Ltd. We found no independent source for them and have not used them in any calculation above.
What to watch
- The 120,000-litre run. dsm-firmenich says it happens this year. It is the step where most single-cell protein programmes discover their bench yields do not survive scale, and it is four months from now.
- Which market the 400 to 500 seeding tonnes go to. If they go to salmon feed and pet food, the aquafeed benchmark is the operative one and the floor matters. If they go to sports nutrition and beverages, the whey benchmark is, and the floor is a footnote. This single observation resolves most of the ambiguity in this article.
- The World Bank fishmeal print for the months since March 2026. If the series resumes above USD 2,316, dsm’s floor has headroom. If it prints back in the 1,600 to 1,900 range once Peru’s season normalises, it does not.
- Whether the Eastern European site is named. Kurmaly said “most likely” Eastern Europe, with access to bioethanol and beet sugar. A named site with a stated capacity is the point at which the 2,000 to 10,000 tonne range becomes a number.
The testable claim we are making: at the World Bank benchmark’s two-year range, Protopia cannot compete in aquafeed on price alone, and its aquafeed case rests on fishmeal staying elevated. We will check the series when it publishes.