Precision fermentation

Precision fermentation is the use of engineered microbes — yeasts, fungi, or bacteria — as programmable cell factories that secrete specific proteins, fats, or small molecules. In a food context, the target molecules are the ones animals have historically been the only commercial source of: milk proteins (whey, casein), egg proteins (ovalbumin, ovomucoid), structural proteins (collagen), and pigments and cofactors (heme, lactoferrin, B12). The microbe is the host; the product is molecularly identical to the animal-derived version.

The platform is not new. Recombinant human insulin has been made in E. coli since 1982, and recombinant chymosin — the enzyme that curds milk into cheese — has been made in Aspergillus and Kluyveromyces since 1990 and now supplies the large majority of the world’s hard- cheese production. What is new is the extension of that forty-year industrial biology from pharmaceuticals and processing aids into bulk food ingredients (Teng et al., 2021).

How it differs from classical and biomass fermentation

Three distinct fermentation modes show up in the food literature, and conflating them produces confused commentary.

Classical fermentation uses whole, typically wild-type microbes to transform a substrate — beer, wine, yoghurt, sauerkraut, sourdough, tempeh, miso, kimchi. The microbe is not engineered; the value is in the metabolic transformation of the input food.

Biomass fermentation grows the microbe itself as the food. Quorn’s mycoprotein, made from Fusarium venenatum since 1985, is the canonical example; Nature’s Fynd, The Protein Brewery, and Enough (formerly 3F Bio) are more recent entrants. The cell mass is the product.

Precision fermentation engineers the microbe to secrete a specific target protein that is then separated from the biomass and purified. The microbe is a host, often consumed in trace quantities or removed entirely; the value is in the recombinant molecule. Rubio et al. (2020) in Nature Food sit these three modes side by side in a useful schematic.

The canonical product families

Milk proteins. Bovine beta-lactoglobulin — the dominant whey protein — and the four bovine caseins (alpha-s1, alpha-s2, beta, kappa) are the main targets. Perfect Day (US, founded 2014) was first to market with recombinant beta-lactoglobulin, now sold into ice cream, cream cheese, and protein powders under partner brands. Remilk (Israel, 2019) and Imagindairy (Israel, 2020) pursue parallel whey routes. Formo (Germany, 2019) focuses on recombinant caseins for cheese analogues and launched ricotta- and cream-cheese-style products in Europe in 2024. Change Foods (US/Australia, 2019) targets casein for pizza cheese.

Egg proteins. The EVERY Company (formerly Clara Foods, 2014) produces recombinant ovomucoid and ovalbumin. Onego Bio (Finland, 2022), a VTT spinout, uses Trichoderma reesei to produce ovalbumin and in 2024 received FDA “no questions” response to its GRAS notice for its Bioalbumen product.

Structural and bioactive proteins. Motif FoodWorks produces recombinant myoglobin and other flavour and texture proteins. Impossible Foods makes soy leghemoglobin — the “heme” in its burger — by expressing the soybean gene in Pichia pastoris. Geltor produces recombinant collagens for cosmetics and food applications. Helaina and Turtle Tree Labs produce recombinant human-identical lactoferrin for infant and adult nutrition.

Process fundamentals

A typical precision-fermentation run looks, from thirty thousand feet, like pharmaceutical biomanufacturing scaled toward food economics. The gene encoding the target protein is inserted into the host genome under a strong inducible promoter. The strain is grown in a stirred- tank bioreactor on a defined medium of sugars, nitrogen, salts, and trace elements. Induction triggers high-level expression; the protein is secreted into the supernatant (for secretion hosts) or accumulates intracellularly. Downstream processing separates cell mass, concentrates the protein via filtration and chromatography, and formulates the finished ingredient.

The engineering targets that determine commercial viability are well understood: titre (grams of product per litre of broth), rate (grams per litre per hour), yield on substrate (grams of product per gram of sugar), and downstream recovery. Teng et al. (2021) summarise the state of the art across these axes; industrial targets for bulk food proteins converge on titres above 10 g/L with yields above 0.1 g product per g glucose to reach commodity-adjacent prices.

Regulatory path

In the United States, most precision-fermentation products enter the market via self-affirmed GRAS (Generally Recognized As Safe) with voluntary FDA notification. The FDA reviews the notifier’s safety dossier and issues either a “no questions” letter or a rejection. Perfect Day’s beta-lactoglobulin (GRN 863, 2020), The EVERY Company’s recombinant ovalbumin (GRN 1001, 2022), and Formo’s beta-lactoglobulin and koji-produced caseins (multiple GRNs, 2023–2024) have all cleared this pathway. Onego Bio’s Bioalbumen cleared in 2024.

The European Union treats these products as novel foods under Regulation (EU) 2015/2283, and the application pathway is slower and more document-intensive. Formo’s products entered the EU market via a reformulation strategy that blends recombinant caseins with traditional ingredients; full novel-food dossiers for single-ingredient recombinant proteins remain in review. Singapore’s Novel Food framework has approved several precision-fermentation products since 2021.

Life-cycle assessment

Sinke and Swartz (2023), in a CE Delft LCA commissioned by the Good Food Institute and covering yeast-based recombinant proteins produced at commercial scale, modelled greenhouse-gas emissions, land use, and water use across assumed scenarios with renewable-powered facilities and conventional sugar feedstocks. The analysis reported roughly 72 percent lower greenhouse-gas emissions, over 90 percent lower land use, and substantial water reductions versus the average of the dairy and egg proteins it substitutes, depending on scenario assumptions. Like all prospective LCAs it is sensitive to feedstock assumptions (cane versus beet versus corn sugar), electricity mix, and allocation between primary product and biomass side streams.

The Humbird (2021) techno-economic analysis of cultivated meat is often cited as an analog ceiling for fermentation economics. The two processes share bioreactor capital, sterility requirements, and downstream unit operations, but diverge sharply on media cost — precision fermentation runs on defined microbial media costing dollars per kilogram of product, not the hundreds to thousands of dollars per kilogram that growth-factor-laden cultivated-meat media imply. This is the main reason precision fermentation is already selling into food channels while cultivated meat remains at tasting volumes.

Cost-curve projections

RethinkX’s Rethinking Food and Agriculture (Tubb and Seba, 2019) argued that precision fermentation would fall along a Wright’s-Law cost curve — roughly halving in unit cost with each doubling of cumulative production — and would drive bulk recombinant protein below 10 dollars per kilogram by 2025 and below 1 dollar per kilogram by the early 2030s, displacing the majority of commodity dairy and egg production within fifteen years. The report’s timeline has been widely criticised as aggressive; its directional thesis that fermentation proteins will cross commodity parity within a generation is closer to industry consensus. Good Food Institute’s annual State of the Industry fermentation reports track the actual trajectory — investment, capacity, and published cost points — which as of 2024 sits behind the RethinkX curve but meaningfully above the Humbird- anchored bear case.

Where this sits in the vegan argument

Precision-fermentation products are not animal-derived in any meaningful sense — no animal is used, kept, or killed in their production — and the protein itself is bioidentical to the animal version. For most vegans the ethical question reduces to allergen profile (whey is still whey and will still trigger milk allergies) and consumer labelling. Unlike cultivated meat, which uses real animal cells and draws genuine vegan disagreement, precision fermentation sits cleanly on the vegan side of the ledger.

The systems-level point is larger. If the cost curve continues even at a fraction of the RethinkX pace, recombinant whey, casein, and ovalbumin will compete directly with their animal analogues on price for bulk ingredient markets — protein powders, processed cheeses, baked-goods binders, infant formula — long before they reach the retail dairy aisle. That is where the displacement of animal protein actually begins, in industrial supply chains rather than at the grocery shelf.