Section
Capital equipment, plant design and the throughput constraints that decide whether a product ever ships.
FDA's letters on AGRN 26 and AGRN 33 describe Methylobacterium extorquens biomass as a fishmeal replacement for finfish at 10% and crustaceans at 6%. In 2026 Knip moved into 165,000-litre contract tanks and says JUV is 'not a commodity protein replacement'. The regulatory basis did not move.
In July 2024 the Belgian mycoprotein start-up said it would 'rapidly' convert Metabolic Explorer's Carling biorefinery to 10,000 tonnes a year. Its August 2026 financing funds a 12 m³ demonstration unit on the same site, a plant built for 6,000 tonnes of chemicals and never expected to be profitable at that scale.
Nourish Ingredients says a 10-ton Creamilux run at SD Guthrie has removed manufacturing as a barrier to growth. The same interview puts that partner's capacity at up to 20 tons a day. The stated next steps — 100 tons, then 500 tons — are five and 25 days of one line.
Adisseo's own Shanghai Stock Exchange filing puts Calysseo's 2025 operating revenue at USD 0.71m against a net loss of USD 15.52m. Production is suspended because the plant has not yet met its design standards. The loan matured on 30 June and was not renewed.
On 14 July ADM named Clinton, Iowa as the US base for EVERY's OvoPro, citing its "industry-leading precision fermentation capabilities". On 21 August the state awarded $2.23m against a $55.5m project to convert 463,503 sq ft of underutilised fermentation space at the same site. Both documents are correct.
The same 6,500–7,650 tonnes of non-binding commitments were reported against a 12,800-tonne Factory 02 in July 2025 and a 6,400-tonne Factory 02 in December. No customer was added. The denominator halved. The binding order book is EUR 0.2 million.
UPSIDE Foods terminated its $50m stalking-horse bid on 14 August. The receiver's 17 August filing records no qualifying bids by the deadline, no auction and no sale hearing. The bid that vanished was already below the two senior secured claims on the file.
Angel Yeast's Yichang line runs 11,000 tonnes a year at over 80% protein. The Protein Brewery plans 600 tonnes for 2027 at about 50%. These are the two producers who publish the number that makes their tonnes convertible — and converting them widens the gap instead of closing it.
Europe's most-used precision fermentation contract site lists one industrial culture size on its own website: 160 m³. A silk protein company for textiles and automotive now has a dedicated line of exactly that size. Neither party says whether any is left.
The first purpose-built precision fermentation plant in the US was to produce commercially by end-2024. Three restatements later the date is 'in 2026', and no start has been announced. The 600,000-litre nameplate anchoring its customer commitments is a vessel volume, not an output.
The Bioreactor RF launched on 'up to 12% lower operational costs'. Five footnotes disclose the model: a €2.87m baseline, €219,000 from cutting eight operators to five, €121,000 from a lower contamination rate. Almost no vendor in this sector shows that much, and showing it makes the arithmetic checkable.
Three European mycelium producers announced capacity in 2026: a wet whole-cut, a dried powder at 50% protein, and a finished oat-fungi food at 14%. No announcement says which basis its tonnes use. Rank them by capital per announced tonne and Millow looks cheap; rank them by tonne of protein and it is the dearest.
ADIO's release says the partnership 'will explore the establishment' of a 4-million-litre facility. Ten months on, no capex, site, timeline, equipment order or FID has been disclosed — and 4 million litres is also Liberation Labs' published expansion target for Richmond.
The June 2026 Huvepharma release publishes a 4x capacity multiple and a 550% order multiple with no base for either. A base does exist — 63,000 litres in 2025, doubled at the start of 2026 — but it was given to a trade interviewer, not the wire. Here is what is on the record.
The Gujarat joint venture with Zydus was announced for Q1 2026, then moved to H2 2026. Four years after FSSAI cleared the product for India, there is still no disclosed capacity, no disclosed capex, and no Indian market — the output is for export.
GFI APAC's July report says a $100m US facility costs RMB100m in China. The source is a rule of thumb from the companies selling the capacity, backed by one non-public speech. Its flagship commercial-scale run was six tonnes — and GFI's own 2023 analysis said pharma capacity is functionally inaccessible to food.
Enifer, ENOUGH and Planetary have each built beside a sugar refinery, a starch plant or an ethanol mill, and none of them used a contract manufacturer. Enifer's €33m for 3,000 tonnes a year is the only disclosed capital cost in the set: €11,000 per annual tonne, a useful number to hold against the sub-$1/kg claims.
Meatly has disclosed two hard unit costs almost nobody else in cultivated meat will publish. Neither of them is the number that decides whether the business works — and the only figure it has given for that is the word 'double figures'.
Everyone in alternative protein agrees fermentation capacity is the bottleneck. This is what closing it actually looks like: Liberation Labs' 600,000-litre Richmond plant, plus an assembled map of who has tanks, who is renting and who is stuck — in four incompatible units.
A $154 million plant, full FDA and USDA clearance, and more than $390 million raised — then insolvency within six weeks of full approval. Its assets are now up for sale to rival UPSIDE Foods for $50 million, with a court hearing set for 20 August 2026.
GFI screened 190 publications and found four techno-economic models for precision-fermented food protein. Their estimates span under $20/kg to $15,000/kg — with nothing in the $20–200 band where the industry actually operates.