On 7 August 2026, the French state, the Hauts-de-France region, Amiens Métropole, the commune of Poulainville and the Banque des Territoires announced what happens next at the site of what was once billed as the world’s largest insect farm. The site enters a “Fast Track” process as strategic economic land, and it “pourra ainsi bénéficier d’une puissance pouvant atteindre 700 MW” — it can be provided with up to 700 megawatts of electrical capacity.
No company name was disclosed. No timetable was disclosed. The reconversion project is described as “en cours de finalisation.”
That is the residual value of a vertical farm designed for 200,000 tonnes of insect ingredients a year: a large industrial parcel with an exceptional grid connection, being offered to whoever wants to draw 700 MW. Eight months after liquidation, no buyer has been named for the plant, and none of the sources we read states a price for any Ÿnsect asset — including the one that did sell.
First, the record needs correcting
English-language trade coverage of the liquidation says the plant that closed is near Dole, in the Jura. It is not.
AgFunderNews, in the report that most other outlets followed, states: “Ÿnsect will close its industrial-scale facility near Dole. The Keprea facility will continue, with a new focus on producing fertilizers from insect waste.” Two paragraphs earlier the same piece says that “In April 2025, Ÿnsect’s ex-CEO and cofounder Antoine Hubert acquired the company’s pilot facility near Dole, France via his new venture Keprea.” Both cannot be right: the same article has the Dole site closing and continuing.
The hyperlink AgFunderNews attaches to “will continue” resolves the ambiguity against its own sentence. It points to France 3 Bourgogne-Franche-Comté, under the headline “Ÿnsect c’est fini, le site de Dole repris par Keprea lui continue” — Ÿnsect is finished, the Dole site taken over by Keprea continues.
Every French source we read agrees. Maddyness reports “la fermeture de sa ferme verticale de Poulainville (Somme).” The Hauts-de-France regional government, Action agricole picarde and Nord France Invest all place the industrial-scale plant at Poulainville, near Amiens, in the Somme — roughly 500 kilometres from Dole. France 3 reports the Jura plant operating under new ownership: “Depuis sa reprise par Keprea, l’usine de Damparis s’est réorientée en produisant des engrais pour les sols agricoles à partir de déjection d’insectes.”
This is a slip in one report rather than a pattern, and AgFunderNews is a serious outlet whose work we rely on throughout this publication. But VerticalFarmDaily republished the same sentence, and other outlets carry variants of “flagship mealworm facility near Dole.” The error matters because it points readers at the wrong asset: the site that failed to find a buyer, that consumed the bulk of the capital, and that is now the subject of a public reindustrialisation effort, is in the Somme.
The sites
| Site | Where | Purpose | Announced capacity | Status |
|---|---|---|---|---|
| The vertical farm (“Ÿnfarm”, a name used by only one source we read) | Poulainville, Somme — described by Ÿnsect itself as “in Amiens, France” | Molitor mealworm breeding and processing; roughly one-third animal feed, two-thirds fertiliser | “up to 200,000 tonnes of ingredients in 2023”; restated in 2025 as 200,000 tonnes a year “à terme” | Inaugurated 2021, never completed. No buyer. Being marketed with up to 700 MW of power |
| Industrial demonstrator | Damparis, near Dole, Jura | Pilot / demonstrator, commissioned 2016 | Not stated in any source we read | Operating, acquired by Keprea, now making insect-frass fertiliser. 12 of 35 staff retained |
| Protifarm site | Ermelo, Netherlands | Buffalo mealworm ingredients for human food | “more than 1,000 metric tons of food ingredients annually” | Acquired April 2021, closed 2023 with 35 job cuts. Subsequent disposal unknown |
| Hatchery | Omaha, Nebraska (ex-Jord Producers) | Hatchery; backyard chicken feed market | Not stated in any source we read | Acquired March 2022. Subsequent disposal unknown |
| Second US plant | Location never specified; synergies with Ardent Mills were to be “explored” | Not stated in any source we read | Not stated in any source we read | Construction was to begin “by end of 2023”. No evidence it began |
Two numbers in that first row do the most work.
200,000 tonnes. Nord France Invest, writing in October 2020, described a farm that “will breed and process molitor beetles for animal feed (1/3) and fertiliser production (2/3)” with a target of up to 200,000 tonnes of ingredients in 2023. Maddyness, in December 2025, gives the same figure as the eventual design capacity. We have found no source stating what the plant actually produced.
€130 million. Hebdo39, citing Les Echos, reports that Ÿnsect needed “environ 130 millions d’euros pour achever sa giga-usine de Poulainville” — roughly €130 million to finish it. The plant was inaugurated in 2021. Four years later it still required nine figures of capital to complete.
That is the actual shape of this failure, and it is not the shape the headline capital figure suggests. This was not a company that built a working plant it could not run profitably. It was a company that inaugurated a plant it had not finished building, and then ran out of money before it could.
Practical consequence. For anyone assessing a scale-up: “inaugurated” is not a milestone. Commissioned, at rate, at yield, at spec — those are milestones. An inauguration is a photograph.
Nobody agrees how much was raised
We counted seven distinct published totals for Ÿnsect’s lifetime fundraising, across three currencies and one figure with no currency at all.
| Figure | Currency | Source |
|---|---|---|
| over $600 million | USD | TechCrunch, December 2025 |
| more than $500 million | USD | AgFunderNews, December 2025 (and outlets republishing it) |
| over €600 million | EUR | AgTechNavigator, December 2025 |
| plus de 600 millions d’euros | EUR | Maddyness, April 2025 |
| plus de 550 millions d’euros | EUR | Maddyness, December 2025 |
| 600 millions de dollars | USD | Hebdo39, February 2025 |
| “dépassait 600 millions” | none stated in the source | Région Hauts-de-France press release |
| around $450 million | USD | Ÿnsect’s own website, December 2022 |
Maddyness contradicts itself between April and December 2025. The regional government — the closest thing here to an official statement — gives a bare “600 millions” with no currency symbol, and attaches it to “la première levée de fonds,” which does not obviously mean the lifetime total. Ÿnsect’s own last public figure, from December 2022, was around $450 million alongside 360 employees.
At current rates the gap between “$500 million” and “€600 million” is on the order of two hundred million dollars. That is not a rounding difference. It is larger than most rounds in this sector.
We are not going to pick one. The point is that the single most-cited fact about the largest failure in insect protein is not established, and every “raised $600m and failed” retrospective is resting on a number nobody has reconciled. Readers who have followed our work on alternative protein consolidation counts and on an offtake book quoted three ways will recognise the pattern: the currency is the missing field, again.
One figure is firmer than the rest, because it comes from the state. France 3 reports that on 30 January 2026 the AFP confirmed with the Ministry of Finance that “environ 148 millions d’euros avaient été investis par l’État parmi les 600 millions levés depuis la création de la société en 2011.” Roughly €148 million of public money.
TechCrunch adds the operating picture: revenue from the main entity peaked at €17.8 million in 2021, “a figure reportedly inflated by internal transfers between subsidiaries,” and by 2023 the company had a net loss of €79.7 million.
The timeline, and why the land was never in the estate
| Date | Event |
|---|---|
| September 2024 | Procédure de sauvegarde opened after declaring insolvency |
| 17 February 2025 | Deadline for offers with the court-appointed administrator |
| 24 February 2025 | Company requests redressement judiciaire at the tribunal de commerce d’Évry |
| March 2025 | Placed in redressement judiciaire; observation period follows |
| 7 April 2025 | Historic investors reinject €10 million |
| 8 April 2025 | Keprea revealed publicly; Emmanuel Pinto named CEO; around 200 jobs reported at risk |
| 16 June 2025 | Hearing at Évry; Keprea takeover of the Dole demonstrator agreed; ruling on the continuation plan deferred to 30 June |
| Spring–June 2025 | Redundancy plan covering 137 of 200 posts |
| 1 December 2025 | Liquidation judiciaire pronounced. 43 employees remaining |
| 30 January 2026 | AFP and the Ministry of Finance confirm ~€148m of public money |
| 7 August 2026 | Poulainville designated strategic economic land; up to 700 MW; no buyer named |
Headcount across that period tells its own story: 360 employees stated by the company in December 2022; around 200 at risk in April 2025; 137 made redundant by June; 43 remaining at liquidation.
The most consequential structural fact is one almost no coverage mentions. The Hauts-de-France communiqué calls on the Caisse des Dépôts et Consignations, “propriétaire de la future friche via la SCI constituée avec Ynsect” — owner of the future brownfield site through the property company formed with Ÿnsect — to examine the conditions for reindustrialising it.
The state investment bank owned the real estate. It was held in a separate property vehicle, not on Ÿnsect’s balance sheet. So when the company was liquidated, the single largest physical asset was not in the estate to be sold; it was already owned by a public financial institution, which is now trying to find it a second life. That is why the site can be repositioned by a public-sector consortium eight months after liquidation, and it is why creditors will see nothing from it.
Practical consequence. When a scale-up is financed through a property vehicle co-owned with a public lender, the plant is not collateral for the operating company’s creditors. Investors underwriting the operating entity should read the SCI structure before assuming the building backs their position.
The one asset that sold
Keprea — a company created in August 2024, based in Paris, founded by former Ÿnsect people including co-founder Antoine Hubert, and directed by Béatrice Vassy — took over the Damparis demonstrator and 12 of its 35 staff. Its stated business is “la production et le négoce de tous produits destinés à l’agriculture durable ou à la protection de l’environnement.” The site now produces fertiliser from insect frass.
Two things about that transaction are worth stating plainly.
The price is not public. No source we read gives a figure, and we tried.
The date is disputed. AgFunderNews puts the acquisition in April 2025. Agra places the enabling court hearing on 16 June 2025, and France 3, writing in December 2025, says the Jura plant “avait été reprise par Keprea en juin dernier.” Maddyness’s April 2025 scoop described Keprea as a venture Hubert was preparing to launch — “s’apprête à lancer” — with a possible launch by September, which is not consistent with an acquisition having already completed that month.
The substance, though, is the same in every version and it is the finding: the only Ÿnsect asset with a buyer was the 2016 pilot; the buyer was assembled by the company’s own founders; and the new use is fertiliser, not protein. The protein thesis did not find a purchaser at any price.
The case that this is a French story, not a sector story
Worth putting properly.
Ÿnsect built the most capital-intensive plant in its industry, in a high-cost jurisdiction, with heavy public co-funding, before demonstrating unit economics at the pilot it had been running since 2016. That is a specific, avoidable set of choices. A company that scaled through tolling agreements and existing infrastructure would not be in this position, and several are not.
The sector-level counter-evidence is nonetheless hard to wave away, and AgFunderNews assembles it in the same report: Agronutris filed a safeguard procedure in 2025; Aspire Food Group sold its Ontario cricket farm after buckling under debt; South Africa’s Inseco ceased operations and sold its assets to industry partners; Denmark’s ENORM was declared bankrupt after a six-month restructuring. Dr. Dustin Crummett of the Insect Institute gives the economic reason: “for animal feed, where insects cost 2 to 10 times as much as soy or fish meal, the sector struggles to become competitive.”
A cost multiple of two to ten times the incumbent is not a scale problem. Scale closes a gap of tens of percent. It does not close 10x, and no amount of vertical-farm automation was going to.
That is the same conclusion we reached about Believer Meats, which cleared FDA and USDA and shut down anyway, and about the supply-layer shutdowns whose developers had raised far more than their suppliers. Approval was not the constraint. Capital was not the constraint. Cost was.
What we could not establish
- A price for any Ÿnsect asset. Not the Keprea transaction, not any element of the estate. No source we read states one.
- Whether Keprea acquired real estate or only plant, business assets and staff. Agra describes it as taking over “son démonstrateur industriel de Dole” without specifying the perimeter.
- Any total realised from the liquidation estate.
- What the Poulainville plant ever actually produced. Only the 200,000-tonne design figure is published.
- Its floor area. 40,000 m² per Nord France Invest in 2020; 45,000 m² per Maddyness in 2025.
- What happened to the Ermelo, Netherlands site. Confirmed closed in 2023; no source we read says whether the site or its equipment were sold, to whom, or for how much.
- What happened to the Omaha, Nebraska hatchery. Not addressed in any source we read.
- Whether the second US plant was ever built. Only the December 2022 statement of intent is documented.
- The capacity of the Damparis demonstrator. Not stated anywhere we looked.
- Whether “Ÿnfarm” was the plant’s official name. Only TechCrunch uses it.
- Whether Poulainville will be demolished. No source says so; the 700 MW framing implies reuse of the connection, which does not settle the buildings.
- The identity of the prospective occupier, explicitly withheld in the August 2026 communiqué.
- Whether the observation period began in April or September 2025. The Région says April; AgFunderNews says September.
- The outcome of the call for asset acquisition launched at liquidation. As of the most recent source we read, dated 7 August 2026, nothing beyond the Keprea transaction had been announced.
What to watch
- Whether the 700 MW connection finds a taker, and what for. If the buyer is a data centre or an electrolyser, the honest conclusion is that the most valuable thing Ÿnsect built was a grid connection. That would be a useful data point for every alternative protein project currently justifying a site on its power and utilities.
- Whether any insect-protein equipment is sold separately. Specialised rearing, sieving and processing equipment has almost no secondary market outside the sector, and the sector is contracting. A disclosed equipment sale — at any price — would be the first real mark on what this class of asset is worth, in the way the Believer Meats plant and IP auctions were.
- Whether Keprea’s fertiliser business works. Frass was two-thirds of the Poulainville plan by design. If the by-product turns out to be the viable product, the sector’s original thesis was inverted, and the twelve people in Damparis are running the experiment that tests it.
- Whether anyone reconciles the fundraising total. A court-appointed liquidator’s report would settle it. Until then, treat every “$600m” retrospective as an estimate.