Calysseo’s plant in Chongqing was built to make 20,000 tonnes of single-cell protein a year. In 2025 it recorded operating revenue of USD 0.71 million and a net loss of USD 15.52 million. Production is now suspended. The plant, its 50% shareholder told the Shanghai Stock Exchange in June, still “requires technical transformation to meet its plant’s design standards.”
Those figures are not an estimate. They come from Bluestar Adisseo’s announcement No. 2026-031, filed on 18 June 2026, disclosing why it would not renew a shareholder loan to the joint venture it half-owns.
This is the number the sector has been missing. When we mapped gas fermentation’s food and feed capacity gap on 18 August, we could put Calysseo’s 20,000-tonne nameplate in a table but had nothing on what the plant actually shipped, because nobody had published it. Adisseo’s obligations as a Shanghai-listed company have now published it.
Methodology
Every financial figure below is taken from Adisseo announcement No. 2026-031, fetched and read in full. The announcement labels the 2025 figures unaudited, and we label them the same way.
Three of its numbers check against each other. Total assets of USD 78.70 million minus total liabilities of USD 63.95 million gives net assets of USD 14.75 million, which is the figure stated. Liabilities divided by assets gives 81.26%, which is the asset-liability ratio stated. Both reconcile exactly, which is the check we run on any figure lifted from a PDF before we print it.
Where we convert currencies we use the European Central Bank reference rate for 2 September 2026, EUR 1 = USD 1.1578, and say so at the point of use. Where we infer a quantity rather than read it, the inference and its assumption are marked.
What the filing says
| Item | Value | Basis |
|---|---|---|
| Calysseo Limited registered capital | USD 80.00m | Stated, 50/50 Adisseo and Calysta |
| Total assets, 31 Dec 2025 | USD 78.70m | Unaudited |
| Total liabilities, 31 Dec 2025 | USD 63.95m | Unaudited |
| Net assets, 31 Dec 2025 | USD 14.75m | Unaudited |
| Asset-liability ratio | 81.26% | Stated and recomputed |
| Operating revenue, 2025 | USD 0.71m | Unaudited |
| Net loss, 2025 | USD 15.52m | Unaudited |
| Adisseo shareholder loan facility | up to USD 16.25m | Approved 7 Feb 2025 |
| Outstanding principal at disclosure | USD 11.893m | Stated |
| Loan interest rate | SOFR + 1.3%, “ca 3%” | Stated |
| Initial loan term | to 30 June 2026 | Stated |
| Bad debt provision taken | RMB 82.8218m | Announcement 2026-016, 7 Mar 2026 |
| Exposure as share of Adisseo net assets | 0.41% | Stated |
The loss is 21.9 times the revenue. That ratio is the whole story of a first-of-a-kind plant that reached commercial approval and then could not reach its design point.
What 0.71 million dollars of revenue implies in tonnes
The filing gives revenue but not volume. Volume has to be inferred, and the inference depends entirely on the price FeedKind realised — which Calysseo has never published.
FeedKind competes with fishmeal. The World Bank’s benchmark series for fishmeal, Peru 65% protein CIF, stood at USD 1,836.55 per tonne in March 2026, the most recent month in the published series. Calysta’s own CEO said in earlier reporting that aquaculture would not pay far above fishmeal parity. Using that as the upper anchor and a premium pet-food price as the lower-volume case gives a range:
| Assumed realised price | Implied tonnes sold in 2025 | Share of 20,000 t nameplate |
|---|---|---|
| USD 1,836.55/t (World Bank fishmeal benchmark, Mar 2026) | about 387 t | about 1.9% |
| USD 3,473/t (EUR 3,000/t at ECB 2 Sep 2026 rate) | about 204 t | about 1.0% |
So on any plausible price, Calysseo sold somewhere between roughly 1% and 2% of its first-phase nameplate capacity in 2025 — the second full year after China’s Ministry of Agriculture and Rural Affairs cleared FeedKind for aquafeed.
Practical consequence: nameplate capacity in this sector is an engineering statement, not a supply statement. When a competitor, a supplier or an offtake counterparty cites a tonnage figure, ask which of the two it is. A plant that has been commissioned, approved and shipped internationally can still be running at one-fiftieth of the number in its press release.
The plant did everything except run
What makes Calysseo instructive is that it cleared every gate the sector treats as the hard part.
| Date | Milestone | Source |
|---|---|---|
| 6 Jan 2021 | Groundbreaking, Changshou, Chongqing; 20,000 t first phase | Calysta |
| Feb 2024 | MARA approves FeedKind for use in aquaculture feeds in China | PPTI |
| May 2024 | First dog treats containing FeedKind Pet launched, Germany | PPTI |
| Aug 2024 | First major international shipments from Chongqing to Europe | PPTI |
| 7 Feb 2025 | Adisseo shareholders approve loan of up to USD 16.25m | Adisseo 2026-031 |
| 31 Dec 2025 | Revenue USD 0.71m; net loss USD 15.52m | Adisseo 2026-031 |
| 7 Mar 2026 | Bad debt provision of RMB 82.8218m recognised | Adisseo 2026-016 |
| 18 Jun 2026 | Adisseo discloses it will not continue financial assistance | Adisseo 2026-031 |
| 30 Jun 2026 | Loan matures, not renewed | Adisseo 2026-031, PPTI |
| 1 Jul 2026 | Berend Jan Kingma starts as Calysta CEO | AgFunderNews |
Regulatory approval, commercial launch, an export shipment, a branded consumer product on shelf in Germany. None of it was the constraint. Adisseo names the constraint precisely: “greater-than-anticipated difficulties in production scale-up phase,” and a facility that “requires technical transformation to meet its plant’s design standards.”
That is the same shape as Believer Meats, which cleared FDA and USDA and shut anyway. Approval is not the gate. Throughput at designed cost is the gate.
The provision tells you what Adisseo thinks it will recover
Adisseo recognised a bad debt provision of RMB 82.8218 million against the receivable. Divide that by the USD 11.893 million outstanding and the implied rate is about 6.96 RMB to the dollar. That is a rate consistent with the renminbi around the end of 2025, which means the provision is not partial. Adisseo has written the loan down to approximately zero.
It can afford to. The filing states the exposure is 0.41% of Adisseo’s audited net assets at 31 December 2025. Work backwards and Adisseo’s net assets are about USD 2.9 billion. The venture that was Calysta’s flagship industrial asset is four-tenths of one percent of its partner’s balance sheet.
Practical consequence: in a joint venture between a startup and a listed corporate, the corporate’s tolerance for a stalled asset is set by its balance sheet, not by the project’s merits. A partner for whom the plant is a rounding error can walk away on “the principle of prudence” — Adisseo’s phrase — long before the technology has been disproved.
What Calysta owes, and what nobody has disclosed
The loan condition is worth reading closely. Adisseo’s assistance was made “subject to the other 50%-shareholder of Calysseo having provided a proportionate financial assistance to Calysseo under the same terms and conditions, in proportion to its shareholding in Calysseo.”
Adisseo’s outstanding principal is USD 11.893 million. If the proportionality condition was satisfied, Calysta’s should be approximately the same, putting combined shareholder loans at roughly USD 23.8 million. We are inferring that from the stated condition. Calysta is private, has published no equivalent figure, and we have not seen one.
That matters for reading Calysta’s position. Calysta is not merely a technology licensor watching a JV struggle; on the face of the condition it is a lender to the same entity, for a similar amount, with the same recovery risk Adisseo has just provided against in full.
The counter-argument
The strongest case against reading this as a failure is that the technology worked and the downstream did not.
Calysta cofounder Alan Shaw, now non-executive chair, told AgFunderNews that “Calysseo has proven full productivity in upstream processing” and that the current focus is “on debottlenecking downstream processing. This will enable higher throughput at designed capacity and correspondingly lower the cost of goods produced.”
That is a specific and testable claim, and it is consistent with Adisseo’s own language about technical upgrades rather than a technology dead end. Fermentation itself — growing the organism on methane at scale — may genuinely be solved. Separation, dewatering and drying are conventional unit operations, and conventional unit operations can be fixed with capital.
Neither company has said the technology has ceased development. Adisseo says it “continues to work with Calysta to seek solutions.”
The difficulty is that this argument still requires money, and the disclosed sequence is that external financing was sought and not obtained: Adisseo’s filing describes “unsuccessful attempts to secure external funding” before the decision not to renew. The upstream may work. The capital to fix the downstream is the thing that is missing.
What we could not establish
- Realised FeedKind price. Never published. Our tonnage range rests on two assumed prices, both stated above. Treat the range as an order of magnitude, not a measurement.
- Whether the 20,000-tonne nameplate was ever revised. The 20,000 t figure comes from Calysta’s January 2021 groundbreaking release. We found no later company statement restating it, and no statement withdrawing it.
- Calysta’s shareholder loan balance. Inferred from the proportionality condition in Adisseo’s filing, not disclosed by either party.
- The exact FX rate used for the RMB 82.8218m provision. We divided the provision by the stated USD balance to get an implied 6.96. We did not find Adisseo’s stated conversion rate, so we cannot confirm the provision is exactly 100% rather than approximately so.
- The World Bank fishmeal series stops at March 2026. Trade reporting through June and July 2026 describes Peruvian fishmeal at record highs in China following an El Niño-hit anchovy season. The March print is therefore the last published benchmark, not the current market.
- Whether Calysseo has since secured financing. The filing says it is “actively seeking funding solutions.” We found no announcement of a completed raise as at 4 September 2026.
What to watch
Four things, each checkable against a document rather than a press release.
- Adisseo’s next interim or annual report. Any further provision, or a reversal, will appear there, and Adisseo files. A reversal would be the strongest possible signal that financing landed.
- Whether Calysseo Chongqing resumes production at all in 2026. Adisseo has tied resumption explicitly to completing technical upgrades. Resumption without a disclosed raise would be the surprising outcome.
- The Chinese pet food approval. Shaw said Calysta expected clearance to sell into the Chinese pet food market in Q3 2026. That quarter closes this month. Pet food is the higher-price channel, and the one that most changes the revenue arithmetic above.
- Whether anyone else publishes a utilisation number. Calysseo’s figures are public only because a listed company had to disclose them. Solar Foods discloses because it is listed; we tracked how its Factory 02 denominator moved for the same reason. Every privately held gas fermentation plant in the world is under no such obligation, and it is worth being explicit that this is why the sector’s utilisation data is so thin — not because the plants are running well.
The claim we will hold ourselves to: on the evidence filed to date, no gas fermentation plant anywhere has published a food or feed volume consistent with its own nameplate. If one does, we will say so.