On 23 October 2024 the salt stopped. Vencorex, the chlor-alkali and isocyanates producer at Pont-de-Claix, had been placed in receivership on 10 September at the request of its Thai shareholder PTT GC, and six weeks later its deliveries of salt to Arkema's nearby Jarrie site ceased. On 21 January 2025 Arkema presented a project that would close chlorine, soda, methyl chloride and technical fluids production at Jarrie and cut 154 jobs. Four months later, on 23 May 2025, Michelin announced a EUR 60 million unit on the other Isère chemical platform, at Péage-de-Roussillon, to make a molecule almost nobody produces at scale. The same department, the same seven months, money leaving and money arriving.
This is the first piece of the chemicals and process chapter of the CareerOn Industry Atlas. It asks the question every chapter opens with: where is the money? In chemicals the honest answer is not a total. No admitted document prints one figure for investment across Isère chemistry, and we do not build one by adding figures that different publishers measured in different ways. What the record does show, in eight dated documents, is the direction money moved after the 2024 break, and the reason it moved that way.
In Isère chemicals, money no longer follows volume: it follows the units a platform cannot lose and the new molecules that public co-funding makes bankable, which is why Pont-de-Claix lost a chain for want of salt while Roussillon kept building.
A platform is a chain, not a list of plants
To read the money you first have to read the plumbing. A chemical platform looks like a collection of separate factories behind one fence. It works like a single machine. One unit's by-product is another unit's raw material, and the utilities that keep them all safe and running (steam, water treatment, effluent handling, emergency response) are shared.
The regional environment authority, the DREAL, printed the Isère version of this chain in plain words in its report on an inspection of Vencorex on 22 September 2023. Vencorex made chlorine, soda and hydrochloric acid; chlorine was the raw material for its isocyanates, with a stated capacity of 80 kt a year, and the report adds: "Enfin, l'HCl, coproduit de la fabrication d'Isocyanates, alimente le site de Jarrie pour le chlorure de méthyle utilisé pour la fabrication de Silicones à Roussillon puis Saint-Fons." In English, the hydrochloric acid co-produced by isocyanates fed Jarrie's methyl chloride, which in turn went into silicones at Roussillon and then Saint-Fons.
Read that sentence slowly, because it is the whole story of 2024 in advance. Pont-de-Claix, Jarrie and Roussillon were not three industrial sites that happened to share a department. They were links. When the first one failed, the question for every other link was not "is my market healthy?" but "can I still get my input, at a price and a specification I can live with?"
Seven months in which one chain broke and another platform kept building
10 September 2024
Vencorex in receivership
At the request of its shareholder PTT GC.
23 October 2024
Salt to Jarrie stops
Arkema’s historical supplier ceases deliveries.
21 January 2025
Jarrie refocus project
Chlorine, soda, methyl chloride and technical fluids to stop; 154 jobs.
10 April 2025
Lyon commercial court
Partial takeover by Wanhua: 54 of 450 jobs kept, per Le Figaro and AFP.
23 May 2025
Michelin 5-HMF at Roussillon
EUR 60 m, partly subsidised; about 30 direct jobs.
October 2025
Construction starts
Per the platform history page, read 11 October 2026.
14 January 2026
DREAL inspection at Pont-de-Claix
Making-safe of stopped units checked, targeted not exhaustive.
Sources 1 Arkema · 4 Agence France-Presse · 6 Michelin · 8 GIE Osiris · 5 DREAL Auvergne-Rhône-Alpes
Jarrie: keeping the lines that do not need the chain
Arkema's release of 21 January 2025 is unusually precise about why it made the choice it made. Since the receivership, it says, it had looked for an alternative, because "the lack of an alternative supply of salt, in quantity and of a quality comparable to that of Vencorex, was a strong threat to maintaining the activity of the entire site." It found one, partly. A salt that needs no further purification could be used directly by some of the existing units, and it "ensures the sustainability of the site's hydrogen peroxide, chlorate and perchlorate units"; it was not compatible "in terms of specifications, cost and quantity" with keeping everything else.
That is the decision rule in one paragraph. Arkema did not keep the largest units or the oldest. It kept the units whose input could be replaced, in businesses where it describes itself as "one of the world leaders", and it proposed to close those whose input could not. The group's 2024 Universal Registration Document, filed with the French market authority on 27 March 2025, repeats the plan on page 62 in the same terms, including the 154 jobs, and states its status at filing: a project "subject to the information and consultation of Arkema's works councils." Both documents are Arkema's own, so the 154 is one publisher's figure printed twice, not two confirmations.
The same report shows the scale this decision sits against. On page 292 Arkema prints its recurring capital expenditure for the whole group at EUR 761 million in 2024, against EUR 608 million in 2023, with no exceptional capital expenditure in 2024. That is a group-wide figure across every country Arkema operates in, and the report prints no Jarrie-specific investment, so we do not present any part of it as money spent in Isère. Its use here is narrower. A group spending on that order chose, at Jarrie, to stop rather than to rebuild the chain. Money was available. It did not go to the units that depended on a supplier that no longer existed.
Pont-de-Claix: what was bought, and for how much
On 10 April 2025 the Lyon commercial court ruled on Vencorex. As reported by Le Figaro with AFP the same day, it attributed a partial takeover to the Chinese group Wanhua, through its Hungarian subsidiary BorsodChem: "Elle ne gardera cependant que 54 emplois sur 450. Le prix de cession est de 1,2 million d'euros, et le repreneur s'engage à investir 19 millions d'euros sur le site d'ici 2027, selon le tribunal." The competing cooperative bid built by employees and local authorities was ruled inadmissible. The ruling itself is not public online; we cite the dated report of it and credit the figures to the court as the report does.
Job counts at Vencorex differ by date and by publisher, and we print the one our admitted source prints, with its date: 54 kept out of 450, per Le Figaro and AFP on 10 April 2025. We do not average it with other counts, and we do not net it against Jarrie's 154 or anyone's hiring. Each figure measures a different thing at a different moment.
Nine months later the regulator came back. The DREAL's report on its inspection of 14 January 2026 describes Vencorex France as an upper-tier Seveso and IED installation on a 120-hectare site, and states that the aim of the visit was to verify the making-safe of the upstream TDI units and chlorine units stopped in 2017 and of the "établissement" part stopped in 2025. Because of the size and complexity of the site, it says, the check was targeted, not exhaustive. That document is the regulated proof of what the price tag of EUR 1.2 million meant: most of what remained at Pont-de-Claix was being made safe, not run.
Roussillon: money arriving, and who helped pay
On the Les Roches-Roussillon platform, in the north-west of the department, the record reads differently. On 6 November 2025 the platform's utilities operator, the GIE Osiris, presented its annual sustainability day and printed two sentences that belong side by side. The first: "Entre 2021 et 2025 près de 200 M€ auront été investis sur la Plateforme chimique, 80 % de ce périmètre étant soutenu par France Relance." Close to EUR 200 million invested between 2021 and 2025, with 80 % of that perimeter supported by France Relance, the national recovery programme. The second is about the business: activity for most operators had not recovered, sitting between minus 30 and minus 50 %, and there was no prospect of recovery by 2026.
The 80 % share is Osiris's own figure. The government decision that would let us check it against a programme document has not opened from our probes, so it stays credited to Osiris. What it says, if taken at its word, is that the investment wave at Roussillon was not market confidence alone.
The single costed operator decision in the record is Michelin's. Its release of 23 May 2025 announces a demonstration unit for 5-HMF, a bio-sourced platform molecule made from fructose that can replace fossil-derived ingredients, on the Osiris platform: an annual capacity of 3,000 metric tons, "a total investment of EUR 60 million, partly subsidized by the ADEME in France and the CBE JU at European level", about 30 direct jobs, and a start planned during 2026. Michelin describes the site as the largest production site in the world for this molecule; that is Michelin's claim, not ours. The platform's own history page, read on 11 October 2026, dates the start of construction to October 2025. Whether the unit has started production is not in any admitted document, and we do not imply it.
New money went to new molecules and shared utilities, not to rebuilding the old chain
| Where | Decision | Money printed | Public money |
|---|---|---|---|
| Pont-de-Claix (Vencorex) | Partial takeover, 10 Apr 2025 | EUR 1.2 m price; EUR 19 m pledged by 2027 (per the court) | Not printed |
| Jarrie (Arkema) | Refocus project, 21 Jan 2025 | No site figure; group recurring capex EUR 761 m in 2024 | Not printed |
| Roussillon (Michelin) | 5-HMF demonstrator, 23 May 2025 | EUR 60 m total | Partly subsidised by ADEME and CBE JU; share not printed |
| Roussillon (platform, per Osiris) | Projects 2021 to 2025 | Close to EUR 200 m | 80 % of the perimeter supported by France Relance |
Sources 4 Agence France-Presse · 6 Michelin · 7 GIE Osiris · 2 Arkema
Why the money went where it went
Put the four documents together and a pattern appears that no single one states.
First, money stayed where the input was secure. Jarrie's retained units are the ones a substitute salt can feed. Roussillon's projects sit on pooled utilities whose operator reports non-fossil energy consumption above 75 %. In both cases capital followed the shared infrastructure that still worked, not the size of the market.
Second, new money went to new molecules, not to rescuing old chains. Nobody in the admitted record invested to restart chlorine at Pont-de-Claix; the buyer committed EUR 19 million by 2027 on a site whose remaining units were being made safe. Michelin committed EUR 60 million to a molecule with no industrial European supply.
Third, the public purse sits under the new money. Osiris credits France Relance with 80 % of its 2021 to 2025 perimeter; Michelin's own release says ADEME and a European partnership subsidise part of its unit, without printing the share. Read without this, Roussillon looks like proof that the market believes in Isère chemistry. Read with it, Roussillon is also proof that the state chose where it wanted chemistry to survive.
The counter-case: subsidy is not demand
The strongest objection to this piece's thesis is that it may describe public policy rather than economics. If 80 % of a platform's investment perimeter is supported by a recovery programme, the investment shows where subsidy was available, not where a business case exists on its own. Osiris's own sentence about activity, minus 30 to minus 50 % with no recovery in sight for 2026, sits uncomfortably next to EUR 200 million of capital.
We print that objection because the evidence supports it. What it does not do is reverse the direction. Subsidy did not save the Pont-de-Claix chain, even though, as Le Figaro reported, the government had pledged one public euro for each private euro behind the cooperative bid, and it did not pay to rebuild Jarrie's chlorine. It went, with operators' own money beside it, to utilities and new molecules. That is still a statement about where the money is: in the pieces of a platform that both a company and a state judge worth keeping.
Every job count keeps its publisher and its date, and none is netted against another
| Count | What it measures | Publisher, date |
|---|---|---|
| 154 | Jobs to go at Jarrie under the refocus project | Arkema, 21 Jan 2025 |
| 54 of 450 | Jobs kept by the Vencorex buyer | Le Figaro with AFP, 10 Apr 2025 |
| About 30 | Direct jobs planned at the Roussillon 5-HMF unit | Michelin, 23 May 2025 |
Sources 1 Arkema · 4 Agence France-Presse · 6 Michelin
What this means for your career
If you are looking at chemicals and process work in Isère, this record tells you three practical things.
The durable work sits in what a platform cannot lose: utilities, effluent and water treatment, process safety, and the specialty units a company chose to keep. Jarrie's retained lines and Roussillon's pooled utilities are the two clearest examples in the record. A candidate who can explain how a shared utility system works, and what happens to every user when one input stops, is speaking the language these sites now run on.
The new work is small in headcount and specific in skill. Michelin's unit is about 30 direct jobs. Roles that start a first-of-a-kind unit (commissioning, process engineering on a new route, quality on a new molecule) are rare, and they are the ones that funding follows.
The decommissioning work is real work too. The DREAL inspection of January 2026 is a record of a site being made safe across 120 hectares. Environmental compliance, site remediation and safety inspection are not the end of a chemicals career; in this department they are a current part of it.
Practice that matches these tasks is being commissioned for the CareerOn catalogue. Until it exists, this piece links to no simulation, because a link to the wrong exercise would teach the wrong job.