In May 2025, Gourmey’s chief executive Nicolas Morin-Forest explained why his company did not need a large bioreactor. “Most of the scale effects on production cost are already delivered at the 5,000L scale, allowing us to reach $3.43/lb, or €7/kg in a commercial setup, without the need for even larger, unproven bioreactors,” he told Green Queen. “Our production process is so efficient that scaling beyond 5,000 litres simply isn’t necessary.”
In July 2026, the same company — now Parima, after acquiring Vital Meat — announced that running its process on a 22,000-litre line operated by Australia’s Vow had removed the “final barrier” to scaling cultivated meat.
Both statements can be true. Circumstances change, and a company that finds a cheaper route than the one it planned should take it. But between those two dates, four separate cost figures entered circulation for essentially the same platform, and they are routinely quoted alongside one another as though they were commensurable. They are not. They differ in what is being measured, in what fraction of the product is cultivated, in whether the figure is modelled or observed, and in whether it describes today or the end of the decade.
This is the recurring problem with cultivated meat and precision fermentation economics, and we have written about it before from the other direction: our analysis of the four public cost models for precision-fermented protein found estimates spanning under $20/kg to $15,000/kg. Here the issue is not an absence of numbers. It is an abundance of them without a stated basis.
The four figures
| Figure | Source and date | What it measures | Status |
|---|---|---|---|
| €7/kg ($3.43/lb) | Morin-Forest, May 2025, on an Arthur D. Little techno-economic analysis | Finished product ~50% cultivated cells, in a modelled 5,000-litre commercial setup | Modelled. Contingent on process optimisations described as “achievable and clearly defined” but not yet delivered |
| “below €7/kg” | Parima press release, 15 October 2025 | Not restated | Described as “independently verified production costs” |
| 99% lower | Parima via Green Queen, 2 July 2026 | Cost of a 22,000-litre run relative to “earlier runs” | Relative. No base figure disclosed, so no absolute cost is recoverable |
| Low €40s/kg (~US$45–50/kg) | Arthur D. Little, 2026 Viewpoint | Finished-product cost in 2026 | Illustrative and industry-representative. Explicitly not company-specific |
| €5–10/kg | Morin-Forest, July 2026 | “The next phase, and the path to” | Target. Explicitly forward-looking |
The €7/kg and the low-€40s/kg figures come from the same consultancy. They are separated by a factor of roughly six. Both are correct within their stated terms, and the terms are the whole explanation: one is what a specific optimised 5,000-litre commercial configuration could cost once defined improvements land, and the other is what the industry’s finished product illustratively costs today.
Read together without those terms, they simply look like a contradiction.
Practical consequence: when a cultivated meat cost is quoted to you, three questions resolve almost all of the apparent disagreement in this sector. What percentage of the finished product is cultivated cells? Is this a measured cost of something produced, or a model of a configuration? And is it today’s cost or a target?
What the €7/kg actually rests on
The May 2025 analysis is unusually well documented for this industry, and Gourmey deserves credit for stating its basis when almost nobody else does.
Morin-Forest told Green Queen that the analysis “was conducted on a finished product containing about half cultivated cells, with the rest made up of plant-based ingredients like fats or proteins.” That single sentence is the most important disclosure in the whole sequence, because it means the €7/kg is not a cost per kilogram of cultivated meat in the sense most readers assume. It is the cost of a composite product, roughly half of which is conventional plant-based ingredients. On our own arithmetic — not a figure either company published — if the split is by mass, the cultivated fraction cannot cost more than about €14 per kilogram at that finished-product price, and costs less than that to the extent the plant-based inputs cost anything at all. The source does not state whether “about half” is measured by mass or by volume, so treat that as an order-of-magnitude reading rather than a derived cost.
The model’s other stated assumptions: capital expenditure under €35m per facility, an output of 1,700 tonnes from six 5,000-litre bioreactors, and a cell feed at around 20 cents per litre made possible by cells that “thrive without proteins or growth factors”. Gourmey said at the time that these benchmarks could be met via “achievable and clearly defined process optimisations” — which is a clear statement that they had not yet been met.
So the €7/kg was, in May 2025, a projection of a configuration that did not yet exist, on a finished product half of which is not cultivated, published with its assumptions attached.
Where it changed status
Five months later, the merger press release describes Gourmey’s contribution to Parima as “premium cultivated duck products validated by Michelin-starred chefs and independently verified production costs below €7/kg”.
“Independently verified” is doing a great deal of work in that phrase. It is defensible in one sense: the analysis was performed by a third party rather than in-house, and Arthur D. Little is independent of Gourmey. But “independently verified production costs” reads, to any ordinary reader, as a statement that costs were incurred and then checked. What was verified was a model. No production at that cost had occurred, and the press release does not restate the ~50% cultivated basis, the 5,000-litre configuration, or the contingency on unrealised optimisations.
Nothing here suggests bad faith. This is a standard compression that happens as a number travels from a technical analysis into a corporate communication and then into trade coverage: qualifiers are shed at each step because they do not fit the sentence. But the qualifiers were the information.
Practical consequence: for diligence purposes, “independently verified” should trigger the question verified as what — a measured cost, or a model produced by an independent party. The two carry entirely different weight in an investment memo.
The 99% has no denominator
The July 2026 announcement is a genuine engineering result. Parima ran its duck process at tonne scale in a single run on Vow’s 22,000-litre line, reportedly the largest food-grade cell culture bioreactor in operation, and did so “on the first attempt, with no performance loss compared to smaller scales” according to CTO Victor Sayous. Arthur D. Little’s 2026 Viewpoint treats this as one of three developments that materially clarified the sector’s cost path.
The cost claim attached to it is that Parima achieved this “at a 99% lower cost than earlier runs”, driven by cell-culture yield gains, the removal of growth factors, albumin and insulin from the medium, and volume efficiencies.
A 99% reduction relative to an undisclosed base yields no absolute number. If earlier runs were pilot-scale batches at several hundred euros per kilogram — entirely plausible for 400-litre research runs — then a 99% reduction lands somewhere in the single-digit-to-low-tens euros per kilogram. If the base was higher, so is the result. The figure is unfalsifiable as published, and it cannot be reconciled with either the €7/kg model or the low-€40s/kg industry figure because there is nothing to reconcile it against.
Practical consequence: a percentage reduction without a base is not a cost disclosure. Treat it as evidence of direction, not of level.
What is actually measured
Strip out the modelled, the relative and the aspirational, and a small set of figures in this story appear to be observed rather than projected:
- Growth media approaching €0.2/litre (~US$0.22/L), down from roughly €1–€1.5/L a year earlier, per Arthur D. Little. Gourmey separately stated around 20 cents per litre for its own feed in May 2025.
- Cell densities of roughly 55–100 g/L across leading players — though Arthur D. Little attaches an important caveat that reported densities “vary by reactor mode and remain, in part, model-derived”, and that techno-economic literature identifies physical ceilings such as CO₂ inhibition in the largest vessels.
- Vessel scale: Vow’s 22,000-litre food-grade line, with a second factory extended to 35,000 litres; Parima’s own capacity at 400-litre and 2,000-litre bioreactors plus a 5,000-litre fermenter, with in-house manufacturing “up to 2,000-litre scale”.
- Output: Vow has claimed a 1,500 kg harvest as the largest in the industry’s history; Parima describes tonne-scale in a single run; Arthur D. Little says both companies “have reported reaching multi-ton production.”
Even that last line is three formulations of the same underlying event, and we could not determine whether Vow’s 1,500 kg harvest and Parima’s tonne-scale run are the same production campaign counted twice.
Note that the media cost — the input Arthur D. Little identifies as “the single largest expense” and “the most direct determinant of finished-product economics” — is also the one with the most consistent reporting across sources. That is not a coincidence. It is a purchased input with a supplier invoice. The numbers that resist disclosure are the ones that require an allocation decision.
The counter-argument
The strongest defence of the industry’s practice here is that this is what pre-commercial cost reporting looks like in every capital-intensive sector, and demanding audited unit costs from companies that have not yet been permitted to sell in most of their target markets is a category error.
Parima has approvals for cultivated chicken and duck in Singapore and is awaiting decisions from Food Standards Australia New Zealand, the EU, the UK, Switzerland, the US and one undisclosed market. It has nine active regulatory filings. A company in that position genuinely does not have a commercial cost per kilogram, because it does not yet have commercial production at meaningful volume. Techno-economic modelling is the honest available substitute, and Gourmey published its assumptions more fully than most.
There is also a real argument that the modelled and observed figures are converging rather than diverging. Arthur D. Little’s 2026 assessment is that sub-€10/kg has moved from “an aspiration” to “a matter of timing rather than feasibility”, on the basis that media cost, cell density and demonstrated scale all moved in the same direction over twelve months. If that holds, the €7/kg model will eventually be vindicated as early rather than wrong.
That may well be right. It does not change the position of someone trying to compare two companies today, which is that the published numbers do not support comparison.
What we could not establish
- The base for the 99% reduction. No absolute figure for “earlier runs” was disclosed, and we did not find one. No absolute cost for the 22,000-litre run can be derived.
- Whether “below €7/kg” in the October 2025 press release uses the same basis as the May 2025 analysis — the ~50% cultivated finished product in a 5,000-litre commercial setup. The release does not say, and we have not assumed it does.
- Whether the €7/kg model has been re-run since the merger or at 22,000-litre scale. The 2026 Arthur D. Little Viewpoint says Parima “has a line of sight toward sub-€10/kg economics” but does not restate €7/kg.
- The exact figure behind “low €40s/kg”. Arthur D. Little states a range descriptor, not a point estimate. Our structured record uses 40 as the lower bound of that descriptor and flags it as such.
- Whether Vow’s 1,500 kg harvest and Parima’s tonne-scale run are the same event.
- Parima’s FSANZ timeline. Coverage indicates an August 2026 target appears to have slipped; we found no determination and no revised date, and have not inferred one.
- What Parima pays Vow. The tolling or contract rate is undisclosed, which matters because Arthur D. Little’s cost model treats contract-manufacturing cost per kilogram as a direct function of media consumption at “an unchanged tolling rate”.
What to watch
- Whether Parima publishes an absolute cost from the 22,000-litre run. It is the one number that would collapse most of the ambiguity in this piece, and the company now has the production event to base it on.
- Whether the ~50% cultivated basis persists in future cost claims. If the cultivated fraction rises, a constant headline cost per kilogram represents a large improvement; if it falls, the reverse. Without the fraction, the headline is uninterpretable.
- Whether tolling rates become visible. The capital-light model Arthur D. Little describes converts fixed capital into variable cost — which means the tolling rate becomes the dominant term in unit economics, and it is currently invisible in every disclosure we found.
- Whether Vow’s co-manufacturing pivot attracts a second named customer. Parima is the first public validation of the strategy shift we covered in Vow’s move from own-brand production to manufacturing for others. One customer is a partnership; two is a business line.
- Whether anyone in cultivated meat publishes a cost per kilogram with a stated basis and a measured denominator. Meatly disclosed a £12,500 bioreactor and £0.22-a-litre medium and still no cost per kilogram. Parima has published four cost figures and, on the evidence available, still no measured cost per kilogram either.