Lynred doubles its cleanrooms: what an infrared capacity ramp really recruits

I. What was opened, in units that cannot be improvised

Lynred's own release describes 2 June 2026 as a historic day for the company: the inauguration of the Infrared Campus combined with its fortieth anniversary, held under the high patronage of the Minister for Industry and Energy, in the presence of its two co-shareholders Safran and Thales (Lynred, 2 June 2026). Electronique News, reporting on 5 June 2026, describes the campus as a new plant on the existing Veurey-Voroize industrial site. Le Journal des Entreprises, on 5 June 2026, frames it plainly as an industrial extension.

The operative figure is not the ribbon but the cleanroom. Les Echos reported on 23 June 2026 that Lynred is doubling its cleanroom surface in order to lift annual capacity toward one million sensors. Lynred's 2025 press kit puts the new campus at a total surface of 8,200 m². Cleanroom square metres are the honest unit of measure in this industry because they are the one thing that cannot be improvised: a cooled infrared detector chains material growth, hybridisation onto a read-out integrated circuit, vacuum packaging and a cryogenic cooler, and every step of that chain needs a qualified, particle-controlled environment plus an operator population trained inside it.

II. The €85m/€100m gap, printed rather than averaged

Two investment figures circulate and both are correctly sourced — they simply belong to different dates. The €85 million figure is the company's own groundbreaking announcement of 10 May 2023, which also stated production capacity rising 50% by 2025 and 100% by 2030. The €100 million figure appears across the inauguration coverage of June 2026: La Tribune on 2 June, Le Dauphiné Libéré on 2 June, Présences Grenoble on 4 June, and Le Journal des Entreprises on 5 June.

We have found no public reconciliation of the €15 million delta — whether it reflects equipment added to the programme, construction inflation between 2023 and 2026, or a wider accounting perimeter. So we print both figures with their dates and say the explanation is not public. A reader who sees a single rounded "around €90 million" anywhere should distrust the writer, not the numbers.

III. Forty years that explain the skills mix

The company's own history, set out in its 2026 inauguration press kit, is the best predictor of what it recruits. Sofradir was created in 1986 for the cooled-detector business; Ulis was created in 2002 for microbolometers; the two merged in 2019 to become Lynred. Electrophysics (USA) was acquired in 2008, and New Imaging Technologies (NIT) in 2024. The 100,000th cooled detector was delivered in 2020.

Read as a competency map, that lineage is unusually explicit. Cooled detectors demand III-V and II-VI material physics, hybridisation, vacuum packaging and cryogenics. Microbolometers demand MEMS process work, wafer-level packaging and very different cost engineering. The NIT acquisition widens the portfolio again. An engineer joining this site is joining a house that runs at least two distinct device physics under one roof — which is why process, test and characterisation roles here transfer poorly from a CMOS-only background, and why they are unusually defensible once acquired.

IV. Why Europe treats infrared as sovereignty

Infrared detection is dual-use in the most literal sense: the same physics serves meteorological satellites, industrial thermography, automotive night vision and defence optronics. Both the 2023 groundbreaking release and the 2026 inauguration coverage frame the campus explicitly in terms of French and European industrial sovereignty, and La Tribune's defence desk read the June 2026 investment as a strengthening of Lynred's defence ambitions.

Les Echos adds the demand driver that matters for a career decision: drones. A capacity expansion pulled by defence and drone programmes is funded on budgetary rather than consumer cycles, which is a durability argument. It also carries a distinct risk profile — programme slippage, export-control friction and political dependency — that a consumer-electronics ramp does not.

V. The labour market this lands in

Veurey-Voroize sits inside the Grenoble basin, which is simultaneously the advantage and the constraint. The same valley hosts STMicroelectronics at Crolles, CEA-Leti on the Presqu'île, Soitec at Bernin and Schneider Electric's Grenoble R&D campus. For cleanroom operators, process engineers, test and characterisation engineers and equipment maintenance specialists, that concentration means a genuinely deep pool — and direct, simultaneous competition for it from employers with far larger balance sheets.

Three asymmetries follow. First, a capacity ramp hires operators and technicians before it hires architects: the highest-probability entry into a newly commissioned cleanroom is the shift floor and the metrology bench, not the design office. Second, doubling cleanroom surface is a multi-year qualification exercise, so equipment installation, qualification and maintenance are load-bearing roles for years, not a construction phase. Third, a company owned fifty-fifty by Thales and Safran offers internal mobility into two of Europe's largest aerospace and defence groups — a structural advantage no independent SME can match.

VI. What we could not verify, and the questions to ask

We found no published, dated hiring target attached to the Infrared Campus — no "N jobs by 2028" figure exists in the sources we checked, so we print none. The frequently quoted revenue of about €197 million and headcount of about 536 come from a company-data aggregator profile rather than an audited filing; both are labelled unverified here and excluded from the analysis. We also found no primary source substantiating the claim that Lynred competes directly with Soitec and CEA-Leti for identical profiles; overlap is plausible at the cleanroom-technician level, and it stays marked unverified.

So ask the three questions the public record does not answer. How much of the doubled cleanroom is qualified and in production today, as opposed to commissioned on paper? Is the one-million-sensor figure nameplate capacity or demonstrated output? And how is the NIT-derived uncooled line staffed relative to the cooled heritage? Anyone who can answer those three is telling you what the site actually is — and anyone who cannot should not be quoting you a hiring plan.

VII. Methodology and disclosures

Every figure above is attributed to a dated source. Where figures differ across dates — €85 million in May 2023 versus €100 million in June 2026 — both are printed with their dates and the missing reconciliation is stated. Where a number originates from an aggregator profile rather than an audited filing, it is labelled unverified and kept out of the argument. Where a claim is widely repeated but unsourced — a campus hiring target, and direct talent competition with named neighbours — it is excluded and named here so that its absence is visible.