Air Liquide in Isère: the gas layer is changing customers, not leaving

I. What this dossier claims, and what it will not

Two stories about the same department are both true at once, and each is misleading alone. In the first, an industrial-gas major is compounding: half-year revenue of €13,828m, an operating margin of 20.9%, a record order backlog of €6bn, and an Electronics business line growing while European industry stalls. In the second, thirty kilometres from the semiconductor valley that buys those gases, a chlorine platform that once carried roughly 14% of French chlorine capacity is in liquidation and being taken down.

The temptation is to make one story the cause of the other. We decline. Nothing in our evidence shows Air Liquide withdrawing from Isère, and nothing shows Air Liquide's growth as a consequence of the platform's collapse. What the evidence supports is narrower and more useful: the gas layer of this department is changing customers. Its historical demand came from bulk inorganic chemistry — chlorine, caustic soda, chloromethanes — sold by the tonne. Its documented forward demand comes from electronics and quantum computing, sold by the part-per-billion. Those two businesses need the same physical discipline and a different qualification ladder, which is why the second dossier in this pair is about people.

What is refused, explicitly and everywhere below: any figure for Air Liquide's investment in Isère (we found none); any headcount for Air Liquide at Jarrie, Pont-de-Claix or Crolles (we found none); the size of ALIAD's participation in Quobly (not disclosed in the sources we opened); any claim that Air Liquide owns, operates or is liable for Vencorex or Arkema assets (the hearing records it as a co-tenant, which is not the same thing); and any EMEA-only Electronics revenue line (the group does not publish one).

II. The dismantling, as the record states it

The primary document here is not a press release. It is the record of a hearing before the French National Assembly's economic affairs committee on 13 May 2025, taken during the Vencorex insolvency. Parliamentary testimony is a specific kind of evidence: it is dated, attributable and on the record, and it is not audited. Every figure in this section is therefore printed as what the hearing recorded, not as a verified accounting fact.

The hearing describes a single 120-hectare chemical platform in the south of the Grenoble conurbation, classified under the Seveso regime, with several distinct operators sharing utilities and flows — Vencorex, Arkema and Air Liquide among the co-located companies — supplied with salt brine through a pipeline of about 80 kilometres. It puts the employment at stake in the basin at roughly 6,000 jobs, and Vencorex's share of French chlorine capacity at about 14%. It records that Arkema's restructuring at Jarrie followed the loss of the salt and chlorine supply on which its own chloromethanes and technical-fluids production depended, with about 150 positions expected to go.

Thirteen months later, local reporting of 9 June 2026 closes the other half. The Exalia project — an attempt to restart activity at Pont-de-Claix that had been presented as creating some 250 jobs — was abandoned after the Lyon commercial court favoured dismantling the installations. The same account reports roughly 50 employees remaining on the site against 450 previously, and open questions about demolition and remediation.

Read together, these two documents describe something more structural than a company failure. A shared-utility platform is a physical system: brine arrives, chlorine is produced, chlorine feeds neighbours, neighbours' by-products feed back, and industrial gases, steam and effluent treatment are shared overheads. Remove the anchor tenant and the shared overheads do not shrink proportionally — they become unaffordable for whoever remains. That is the mechanism behind a restructuring that touched a company which never lost a customer of its own.

III. The other Isère: a €115m round with a gas company in it

On 3 June 2026, in Grenoble, Quobly — a CEA and CNRS lineage company building silicon-based quantum processors — closed a €115m Series A. The round was led by Bpifrance (through the Deep Tech 2030 fund managed for the French state under France 2030), SEALSQ and STMicroelectronics, with participation from the European Innovation Council Fund, Blast, ALIAD — Air Liquide Venture Capital, and existing investor Innovacom; existing shareholders include the CEA, CNRS, Quantonation and Supernova Invest. It follows a €19⁷m seed phase run from 2023 to 2025.

Two details in that release matter more than the headline. First, the technology: Quobly's approach uses FD-SOI on 300 mm wafers, deliberately inside established semiconductor manufacturing processes, to attack scalability, yield and reproducibility. Second, the partner list. The company states that it works with industrial leaders including STMicroelectronics, Air Liquide, Soitec and Orano, and it names what those partnerships bring: process control, materials engineering, cryogenics and yield optimisation.

That sentence is an Isère supply chain written out in full. FD-SOI wafers are a Soitec speciality; 300 mm FD-SOI manufacturing is what the valley's fabs do; extreme cryogenics and ultra-pure materials are Air Liquide's; isotopically controlled materials are Orano's domain. The commercial roadmap Quobly publishes — Alloy Pioneer accessible through the cloud by end-2026, deployed inside HPC infrastructure in 2027 — is what turns those partnerships into recurring industrial demand rather than research collaboration.

We record the limit of this evidence. Air Liquide's own half-year report confirms the ALIAD participation and describes the group as supplying specialty gases and Advanced Materials expertise while deploying its extreme-cryogenics technologies along the quantum value chain. It does not disclose the amount invested, does not commit a euro of Isère capital expenditure, and does not describe a site. A venture ticket is a claim on the future, not a plant. It is printed here as a positioning signal, and nothing more.

IV. The historical thread, dated and kept in its place

Air Liquide's relationship with the valley's fabs is not new. A trade report of 15 May 2003 records that STMicroelectronics, Philips and Motorola selected Air Liquide to supply, install and operate the production and distribution of gases and chemical fluids for Crolles 2 — the model the industry calls total gas and chemical management, where the gas company runs the fluid infrastructure inside the customer's fab.

That document is twenty-three years old and is used for one purpose only: to establish that the on-site operating model in this valley is long-standing rather than prospective. It is not evidence of current scope, current volumes or current staffing, and none of those are claimed. The current-state evidence we have is a single recruitment listing for an Air Liquide gas-distribution technician on a permanent contract at Crolles, requiring a two-year post-secondary qualification, with no salary stated — a labour-market signal about one role, treated as such, and developed in the companion dossier.

V. Where the capital actually went in the first half of 2026

The half-year results published 28 July 2026 let us test the claim that electronics gases are where this group's money is going. Group revenue was €13,828m, up 0.8% as published, +4.3% excluding currency and energy effects and +2.6% comparable. Gas & Services contributed €13,408m. Recurring operating income reached €2,893m, up 5.7%, with a group operating margin of 20.9%, up 100 basis points as published and 110 excluding energy and purchase-price-allocation effects. Net profit was €1,823m. The group reports about €300m of efficiencies generated in the half, a record order backlog of €6bn, and investment decisions of nearly €3bn — of which €1bn in Electronics alone, which the chief executive frames as supporting artificial intelligence.

The business-line split for the first half of 2026 is where the asymmetry appears. Electronics revenue rose from €1,224m to €1,317m, up 7.6% as published and 6.2% comparable — the fastest-growing of the four lines, in the six months to 30 June 2026, against Large Industries at €3,703m (−0.6% comparable), Industrial Merchant at €6,144m (+3.2%) and Healthcare at €2,244m (+4.2%).

Geography, in the same half of 2026, is where Europe's position becomes uncomfortable. The Americas delivered €5,225m (+5.4% comparable), Asia-Pacific €2,768m (+2.0% comparable, and +11.9% excluding currency and energy effects with the DIG Airgas acquisition in South Korea), and EMEA €5,414m — up 0.2%. In the first half of 2026 inside EMEA, Large Industries fell 2.6%, Industrial Merchant was flat at +0.2% (+1.9% excluding helium), and the group states that volumes fell principally in Automotive and Chemicals while rising in pharmaceuticals, manufacturing and technologies. In the same half of 2026, Asian Electronics sales grew 9.3% in the half and 13.2% in the second quarter; in the Americas the Electronics line fell 2.3% because weaker Equipment & Installations sales masked growth in carrier gases and advanced materials. No EMEA-only Electronics figure is published, and we do not construct one.

The investment decisions of the half are listed by region, and the list is the argument. In the United States: over $170m in Indiana to build, own and operate two industrial-gas units supplying ultra-pure gases to SK hynix's first American chip-packaging plant; over $160m in Arizona for high-purity gas capacity commissioning in 2028 at the extension of a plant operated by a global semiconductor leader; over $150m in Idaho for a memory-chip leader's capacity increase; over $200m in Texas for a partial-oxidation unit starting in 2029; and $350m for oxygen, nitrogen and argon to a low-carbon steel plant in Louisiana. In EMEA, the named items are different in kind: two contracts with ArianeGroup covering cryogenic propulsion equipment to the launcher's 42nd flight and three years of propellant gases and fluids at the Guiana Space Centre; €70m in Kazakhstan for nitrogen and dry compressed air; a carbon-capture agreement with Holcim whose final investment decision remains subject to additional partnerships and public-sector support; the group's largest Spanish home-healthcare contract, for 90,000 respiratory patients in the Valencia region; and the Quobly round.

Set beside that, the European electronics reference point is still the commitment of 24 July 2025: more than €250m in Dresden's "Silicon Saxony" to build, own and operate three air-separation units, two hydrogen production units and associated infrastructure, supplying ultra-pure nitrogen, oxygen, argon, hydrogen, helium and CO₂ on a major customer's site, operational in 2027, and described by the company as its largest investment to support the development of the electronics sector in Europe. The same release sizes the Electronics business at €2,510m of 2024 revenue with more than 4,000 employees worldwide, inside a group of about 65,000 employees serving 4.3 million customers and patients in 59 countries, with 2025 revenue of nearly €27bn.

VI. The mechanism, stated plainly

Industrial gases are the least discretionary input in advanced manufacturing and the most local. Ultra-pure nitrogen cannot usefully be trucked across a continent at fab volumes; the economics push the gas company to build on the customer's doorstep and operate for fifteen or twenty years. That is why the map of Air Liquide's electronics investments is, quite literally, the map of where new fabs and packaging plants are being built — Indiana, Arizona, Idaho, Dresden, East Asia — and why the absence of a comparable Isère line item is informative without being an accusation.

The mechanism this dossier files is therefore not "Air Liquide is leaving Isère" and not "Air Liquide is investing in Isère". It is this: in this department, the gas layer is being repriced from tonnage to purity, and the capital that follows purity is being committed where new capacity is being built. Isère's claim on that capital rests on the valley's existing fabs and on the FD-SOI-plus-cryogenics stack that the Quobly round has just capitalised — not on the platform whose demolition the courts have now favoured.

Three things would falsify or extend this reading, and each is a document we could not obtain. A named Air Liquide capital-expenditure line for an Isère site would turn a venture signal into an industrial commitment. A disclosed remediation and dismantling plan for the Pont-de-Claix platform would establish who carries the closure cost — and utilities co-tenancy is precisely the relationship where those costs become contested. And an EMEA Electronics revenue disclosure would settle whether Europe's ultra-pure demand is growing at Asian rates or merely holding. Until those exist, the honest version of this story is the one printed above.

VII. What a reader should do with this

If you are choosing a career in this basin, the operative fact is not which employer is expanding this quarter. It is that the same competence — running continuous, high-pressure, high-purity fluid systems safely, under a Seveso regime, on someone else's critical path — is what both the industry being dismantled and the industry being capitalised require. The chlorine platform priced that competence in tonnes. The fabs and the cryostats price it in parts per billion and in uptime. The qualification ladder differs, the physics does not. That transfer, its evidence and its limits, is the subject of the companion dossier.