dsm-firmenich is preparing to run its Protopia yeast protein at 120,000 litres this year and told AgFunderNews on 17 August that the ingredient “can already be used in the US and EU without requiring a GRAS notice or EU novel food authorization.”

That is correct. It is also the single most valuable asset in the product, and it has nothing to do with the strain, the process, or the safety package. It rests on a date: 15 May 1997.

The comparison that makes this legible is not a plant protein or a cultivated product. It is Yarrowia lipolytica yeast biomass, which sits on the same EFSA Qualified Presumption of Safety list, in the same “Yeasts” block, carrying the identical qualification — and which still required a full novel food authorisation taking two years and one month. The variable separating the two is not safety, not the organism class, and not QPS status. It is whether the species was eaten in the European Union before a Thursday in May 1997.

Article 3(2)(a) of Regulation (EU) 2015/2283 defines a novel food as

“any food that was not used for human consumption to a significant degree within the Union before 15 May 1997, irrespective of the dates of accession of Member States to the Union, and that falls under at least one of the following categories: … (ii) food consisting of, isolated from or produced from microorganisms, fungi or algae”

The two limbs are cumulative — they are joined by “and”. This is the part that is routinely described loosely in trade coverage, and getting it right changes the analysis.

A torula yeast biomass is unambiguously inside category (ii). Protopia is not outside the novel food categories. It escapes the Regulation because it fails the first limb only: the species was consumed in the Union to a significant degree before the cut-off. Recital 7 explains why the date is fixed there — continuity with the predecessor Regulation (EC) No 258/97, whose entry into force it marks.

So the correct statement is not “Protopia is not a novel food category”. It is: Protopia is in a novel food category and is exempted by a historical consumption test.

How the exemption was actually established

The EU Novel Food Catalogue carries two relevant entries, both with status NOT_NOVEL_IN_FOOD:

Catalogue reference Entry name Part Created Description
POL-NF-00749 “Torula yeast (Cyberlindnera jadinii) biomass” Whole 11 March 2022 “The request was submitted to the Irish authorities (FSAI) under Article 4 of Regulation (EU) 2015/2283. It was concluded that, Torula yeast (Cyberlindnera jadinii) is not novel when used as a food ingredient.”
POL-NF-15119 “Cyberlindnera jadinii” (synonym: Candida utilis) Whole 17 December 2025 “The entry concerns the microbial culture of Cyberlindnera jadinii”

This is the mechanism, and it is more specific than “history of use”. Non-novel status was established by an Article 4 consultation to a single member state authority — the Food Safety Authority of Ireland — not by an EFSA risk assessment. Article 4 determines novel food status. It produces an answer in months, publishes no dossier, and requires none of the toxicology, allergenicity and exposure work that an Article 10 authorisation demands.

Practical consequence: for a company selecting a production organism, the Article 4 route is a real strategic option and it should be checked before strain selection, not after. The catalogue is searchable, free, and dispositive. If the organism is already listed NOT_NOVEL_IN_FOOD, the EU regulatory programme for a biomass ingredient collapses from a multi-year dossier to a status confirmation.

The comparison table

Product Organism Route Filed Authorised / determined Elapsed
Protopia (dsm-firmenich) Cyberlindnera jadinii biomass Article 4 status determination via FSAI (Ireland) Not published Catalogue entry created 11 March 2022 Not published
Yarrowia biomass (Skotan S.A.) Yarrowia lipolytica biomass Full authorisation 10 April 2017 Implementing Regulation (EU) 2019/760, 13 May 2019 ~2 years 1 month
Fermotein (The Protein Brewery) Rhizomucor pusillus mycelium Full authorisation, Article 10(1) 16 May 2020 Implementing Regulation (EU) 2026/1507, 3 July 2026 ~6 years 2 months

The Fermotein timeline comes from the face of the authorising Regulation: application 16 May 2020, EFSA mandated 7 January 2021, EFSA opinion adopted 29 September 2025, Regulation of 3 July 2026 published in the Official Journal on 6 July 2026.

Set that against the statutory design. Article 11(1) gives EFSA nine months from receipt of a valid application to adopt its opinion. Article 12(1) gives the Commission seven months from publication of the opinion to submit a draft implementing act. The statute contemplates roughly sixteen months. EFSA took four years and nearly nine months from mandate to opinion; Article 11(4) permits that extension where additional information is requested.

The Protein Brewery’s CEO Thijs Bosch attributes the duration to being first: “Fermotein is the first novel mycelium ingredient to successfully go through EFSA, which meant assessment questions were being defined as we went, including two rounds of additional, costly studies requested two and four years after initial submission.”

Note what the applicant’s own dossier summary says about the category: “Fermotein™ falls under the novel food category ‘Foods consisting of, isolated from or produced from microorganisms, fungi or algae’ as per Article 3(2)(a)(ii)”. That is the same category a torula yeast biomass falls under. The category is not the difference. The date is.

We covered the Fermotein outcome and its stalled US counterpart in our piece on The Protein Brewery’s two jurisdictions.

QPS is not the reason, and conflating them is an expensive mistake

Cyberlindnera jadinii is on EFSA’s QPS list. The entry in EFSA’s own published dataset (version 25, 6 July 2026) reads:

“QPS applies for ‘production purposes only’ (the qualification ‘for production purpose only’ implies the absence of viable cells of the production organism in the final product and can also be applied for food and feed products based on microbial biomass).”

That parenthetical expressly reaches “food and feed products based on microbial biomass” — which is exactly what Protopia is. It is tempting to read that as the regulatory basis.

It is not, and EFSA says so on its own topic page: “The QPS assessment is conducted separately from and independently of the safety assessment of a regulated product proposed for market authorisation. Therefore, it does not necessarily lead to its market authorisation.”

The proof is Yarrowia lipolytica, which appears on the same list, in the same yeasts block, with the identical “production purposes only” qualification — and which took a two-year authorisation from an Article 4 filing to Poland in April 2017 through an EFSA opinion in January 2019 to Implementing Regulation (EU) 2019/760 in May 2019.

Practical consequence: QPS status is worth having and is not a market access route. A regulatory plan that treats a QPS listing as a shortcut to EU sale is mispriced. The question to ask about an organism is the 1997 consumption question, and then, separately, the QPS question.

The US position is also widely misstated

Torula yeast is not affirmed GRAS in the United States. It is a listed food additive with a compositional limit. 21 CFR 172.896 — in Part 172, “Food Additives Permitted for Direct Addition to Food for Human Consumption”, Subpart I, “Multipurpose Additives” — reads in full:

“Dried yeast (Saccharomyces cerevisiae and Saccharomyces fragilis) and dried torula yeast (Candida utilis) may be safely used in food provided the total folic acid content of the yeast does not exceed 0.04 milligram per gram of yeast (approximately 0.008 milligram of pteroyglutamic acid per gram of yeast).”

(The spelling “pteroyglutamic” is as printed in the regulation.) The eCFR timeline records no changes to this content after 3 January 2017.

GRAS affirmations live in 21 CFR Part 184. There is no Part 184 affirmation for torula yeast. A full-text search of Title 21 for “torula” returns four sections: 172.896 and three macaroni and noodle standards of identity.

Searches of FDA’s GRAS Notice Inventory return zero records for “torula”, zero for “Candida utilis” and zero for “Cyberlindnera”. Those zeroes are real rather than a broken query: the same interface returns 34 records for “yeast” and 28 for “Saccharomyces”.

The absence of a GRAS notice is therefore consistent with dsm-firmenich’s account — none is needed, because § 172.896 does the work instead.

Practical consequence: the folic acid ceiling in § 172.896 is a live, quotable specification limit, not a formality. Any torula-derived protein sold into US food must meet 0.04 mg total folic acid per gram of yeast. That is a specification question for a high-protein, high-biomass ingredient, and buyers should ask for the certificate of analysis against it. We note that dsm-firmenich has not published a folic acid figure for Protopia.

What dsm-firmenich itself publishes, and what it does not

Having checked the company’s own Protopia page, three absences are worth recording, because the regulatory argument above is assembled from public records rather than from the company’s disclosure.

The page states: “It delivers 70% digestible protein, complete with all essential amino acids. Free from GMOs and allergens, it features a neutral taste and color, along with versatile functional properties such as gelation and emulsification, all supported by regulatory approval.”

  • The species is never named. Not torula, not Cyberlindnera jadinii, not Candida utilis — on the product page or on the two related dsm-firmenich articles we checked.
  • The feedstock is never named. The AgFunderNews interview describes a “feedstock agnostic” strain running on “ethanol, methanol, acetate, sugar, molasses or other substrates”. None of that appears on the product page.
  • “All supported by regulatory approval” is the only regulatory language, with no instrument and no jurisdiction cited. That phrasing sits awkwardly beside the position given to AgFunderNews, which is that no authorisation is required. “Supported by regulatory approval” and “requires no authorisation” are different legal claims, and the page does not reconcile them.

There is also a protein-content wording gap: the page says “70% digestible protein”; the interview says “a minimum of 70% protein”. Those are not the same statement, and on a beat where stated basis is the whole story, the difference matters.

The counter-argument

Two points cut against reading this as pure regulatory arbitrage.

First, the exemption is not a loophole — it is the Regulation working as designed. The pre-1997 test exists because foods with a long European consumption history have a de facto safety record that a dossier is meant to substitute for. Torula yeast has been eaten in Europe for decades. Applying the same four-year EFSA process to it would be regulatory make-work, and the Article 4 route exists precisely to dispose of that question cheaply.

Second, the asymmetry is time-limited in one direction. Once Rhizomucor pusillus mycelium is on the Union list, it is on it for everyone in scope of the authorisation’s conditions, and the next mycelium applicant benefits from the questions EFSA has already settled. Bosch’s own framing — “the duration was largely a function of being first” — is an argument that the six-year figure is a first-mover cost rather than a standing tariff. That may prove right. It is also, at present, untested.

What survives both objections is the planning point. A founder choosing between a novel organism with better titres and a pre-1997 organism with worse ones is choosing between a possible six-year regulatory programme and a possible six-month one, and the EU framework prices that choice at the point of strain selection.

What we could not establish

  • Whether dsm-firmenich relies on either catalogue entry. The entries exist and the company’s claim is consistent with them. We have no statement from dsm-firmenich naming its legal basis, and it did not name one on its product page.
  • Whether the Protopia strain is wild-type or an engineered derivative. The product page says “Free from GMOs”; the interview describes “a highly optimized yeast strain” and “The strain is the magic”. The optimisation method is not stated anywhere we could find. This matters legally: the Irish Article 4 determination covers “Torula yeast (Cyberlindnera jadinii)” generally, not a specific commercial strain, and a materially altered organism could reopen the status question. We put this as an open question rather than a concern.
  • A citable deep link to the individual catalogue entries. The Commission’s catalogue is a single-page application; the entries above are reproduced from the portal’s own backing data and are reachable in the interface by searching “Torula” or “Cyberlindnera”.
  • The accepted species name. EFSA and the Commission both use Cyberlindnera jadinii. NCBI Taxonomy, updated 29 May 2026, gives the accepted name as Williopsis jadinii under a 2026 recombination, carrying Cyberlindnera jadinii and Candida utilis as synonyms. The regulatory documents have not followed the taxonomy. We have used the regulatory name throughout because the regulatory instruments are what determine market access.
  • Any Protopia regulatory filing outside the EU and US. AgFunderNews reports that dsm-firmenich is “separately assessing regulatory requirements in markets including India and Brazil”. Assessing requirements is not filing, and we found no filing.
  • A folic acid figure for Protopia against the § 172.896 ceiling.
  • The Fermotein EFSA opinion publication date. The Regulation gives the adoption date, 29 September 2025. Trade coverage refers to a December 2025 publication. We did not obtain the opinion itself and have used the Regulation’s date.

What to watch

  1. Whether any competitor challenges the strain question. If a Protopia competitor argues that an optimised strain falls outside the general Cyberlindnera jadinii determination, the answer will define how far Article 4 determinations travel.
  2. Whether the second mycelium applicant moves faster than six years. This is the direct test of Bosch’s first-mover claim, and the EU Biotech Act’s sandbox provisions — which, as we have set out, do not currently extend to the food itself — are the obvious variable.
  3. Whether dsm-firmenich publishes a species name and a regulatory basis. A company marketing an ingredient into human food on the strength of a historical consumption exemption has a straightforward interest in naming the organism that carries the exemption.
  4. Whether the Article 4 route sees more traffic. If it does, expect the competitive frontier in single-cell protein to shift toward organisms with pre-1997 European consumption — which is a constraint on strain choice imposed entirely by a 1997 date.