Executive brief
A company that builds roughly 15,000 medical devices a year and reported €200 million of revenue in 2024 told a regional business paper that it refuses the volume fight and will grow to €300 million within three years on premium detectors alone (Les Affiches de Grenoble et du Dauphiné, 16 January 2025). That is a strategy statement, and it is testable. This dossier reconstructs what Trixell at Moirans actually is — a 1997 joint venture between Thales (51%), Philips Healthcare (24.5%) and Siemens Healthineers (24.5%), operating an 8,100 m² plant with 2,000 m² of cleanrooms — and then examines the more unusual thing happening on that site: an industrial accelerator, incorporated as a cooperative on 12 July 2024, whose founding shareholders include a university hospital, two engineering firms and a trade union. We print two divergent dated headcounts rather than averaging them, and we name the eight claim families we refused to assert.
I. The entity, precisely
The operator is Trixell SAS, registered at Moirans (38430) under SIREN 411 195 043, incorporated on 26 February 1997, classified under NAF 2660Z — the manufacture of irradiation, electromedical and electrotherapeutic equipment (French business register, consulted 3 September 2026). Ownership is unusually explicit for a European industrial asset: the company's own shareholder disclosure describes a joint venture between Thales (51%), Philips Healthcare (24.5%) and Siemens Healthineers (24.5%) — three firms the disclosure itself places in the top five of the medical imaging market (Trixell, shareholders page, undated, consulted 3 September 2026).
Readers should hold that structure in mind, because it explains behaviour that would otherwise look irrational. Trixell sells large-format digital X-ray detectors to original equipment manufacturers, and two of its three shareholders are among the largest such manufacturers in the world. A supplier part-owned by its customers does not chase market share against them; it becomes the standard they all build around. The company describes precisely that outcome — technology that "has become the market standard" — alongside the physical means of producing it: a dedicated 8,100 m² production plant including 2,000 m² of clean rooms (Trixell, about page, undated, consulted 3 September 2026).
II. What the site makes, and out of what
The physics is worth stating plainly, because it decides which skills the site hires for. A flat-panel detector converts X-rays into light in a caesium iodide scintillator, and reads that light out through an amorphous silicon semiconductor array. Trixell's managing director describes mastery of exactly those two materials, and splits the product line the way clinical use splits it: static detectors for radiography, and dynamic detectors that guide a surgeon's hands in real time (Les Affiches, 16 January 2025).
The company's own chronology shows how long that competence took to build. A significant investment in a new production line is dated 2007; the world's first WiFi flat-panel detector, the Pixium Portable 3543, is dated 2008; surgical radiography detectors 2011; a new portable family 2013; more than 60,000 Pixium detectors in service worldwide 2014; an expanded interventional family 2016; and 100,000 detectors delivered at the joint venture's twentieth anniversary in 2017 (Trixell, history page, undated, consulted 3 September 2026).
Contemporary trade reporting of the 2007 expansion adds a capacity number — production capacity raised by 75%, with roughly 30,000 square feet built for an automated manufacturing line, to meet OEM demand for digital radiology (BioWorld / Diagnostics & Imaging Week, 13 September 2007). We deliberately do not print an investment figure for that expansion: the trade item carries an internally inconsistent currency pair, and a separate French trade outlet describes a €15 million production-line investment without a date we could verify (imedicale.fr, undated, consulted 3 September 2026). Two unreliable numbers do not average into one reliable number.
III. The strategic choice: premium instead of volume
In January 2025 the company set out its position without euphemism. Asian competition is described as formidable; Trixell's answer is to decline the volume contest and hold the premium segment. The scale that follows from that choice is small by semiconductor standards and large by medical-device standards: approximately 15,000 medical devices a year, built to what the managing director calls extreme reliability, on the argument that an imaging failure during surgery can be fatal (Les Affiches, 16 January 2025).
Two growth vectors are declared alongside it. The first is technological: ImaSpiiR-X, a spectral-imaging project supported by France 2030, with partners including the CEA and the Lyon university hospital — colour X-ray images intended to reduce diagnostic uncertainty. The second is sectoral: the same detector physics is being pointed at automotive battery inspection, airport security and agri-food control. A third, nomadic imaging aimed at medically underserved areas, is described as a concept entering industrialisation (Les Affiches, 16 January 2025). We print no funding amount for ImaSpiiR-X and no revenue split by application, because no source we opened states either.
The diversification claim deserves a mechanism rather than applause. A detector qualified for interventional radiology is over-engineered for battery inspection — which is precisely why it can enter that market at a premium instead of competing on price. The transferable asset is not the sensor; it is the qualification discipline behind it. That is also, as the next section shows, the asset the site has begun to rent out.
IV. The unexpected shift: an accelerator co-founded by a trade union
On 12 July 2024, in the Thales/Trixell premises at Moirans, stakeholders met to create AXEL, an accelerator for prototyping and industrialising medical imaging devices, constituted as a société coopérative d'intérêt collectif (SCIC). Its partners include industry, the Grenoble university hospital, research bodies — and, the account notes as a remarkable fact, the CGT trade union. The initiative is presented as the culmination of twelve years of union work begun after a 2012 fight to keep medical imaging inside the Thales group, a fight that ended three months later with the disposal being called off (Imagerie d'Avenir, 18 October 2024).
The regional development agency lists the founding constellation: industry (Thales, Fortil Group, Sherpa Engineering), higher education and research (Université Grenoble Alpes, Grenoble INP, CHU Grenoble Alpes) and social partners (CGT), with local authorities — Grenoble Métropole, Le Grésivaudan and Pays Voironnais — expected to join. AXEL's declared service is concrete rather than promotional: assembly of industrial pre-series on pilot lines, certification and market access, supply-chain construction, and mastery of manufacturing cycles (Auvergne-Rhône-Alpes Entreprises, published 9 September 2024, updated 22 September 2025).
The operating detail matters for anyone judging whether this is real. AXEL is directed by Laurent Chevallier, previously a programme manager and head of partnerships and funding at Trixell; it started with €280,000 of capital and, on 4 December 2024, exactly one employee — its director — who intended to spend several months identifying which competences client firms actually need before hiring experts. All profits are reinvested; break-even is targeted at roughly two years. Its registered seat is the shared Thales–Trixell–Lynred site at Centr'Alp (ESSOR Isère, 4 December 2024).
Read structurally, AXEL is an attempt to convert a single site's medical-device qualification capability into shared regional infrastructure. The scarce resource in European medtech is not ideas; it is the ability to take a working prototype through pre-series manufacture and regulatory certification. Moirans has done that since the 1980s, when the site built image-intensifier tubes, and since 1997 through digital detectors — a lineage AXEL's director states explicitly as the reason the accelerator exists there (ESSOR Isère, 4 December 2024).
V. Two headcounts, both dated, both printed
In March 2022, the strategic committee of the health industries sector met at Moirans. The contemporaneous account reports the industrial site employing 520 people, of whom 120 for Thales and 400 for Trixell, and names Xavier Caillouet as director of Thales's radiology activity and CEO of Trixell (Présences Grenoble, 29 March 2022). In January 2025, a profile of the company under its new managing director — Kaïs Mnif, appointed in January 2024 — refers to Trixell and its 320 collaborators (Les Affiches, 16 January 2025).
We do not reconcile those figures, and we do not average them. They may count different perimeters, or they may record a real change; no source we opened states which. Both are dated and attributed, and a reader is better served by that ambiguity than by a false single number. A third figure in circulation — a platform profile asserting 130 employees and an annual revenue in the billions of US dollars (LinkedIn company page, consulted 3 September 2026) — is arithmetically incompatible with the company's own reported revenue and is not used here.
VI. The pipeline behind the hiring
The 2022 account also records Nemoxis, a project led by Thales with partners including CEA-Leti, aimed at proximity and mobile imaging, using new-generation carbon-nanotube X-ray sources combined with image-processing software and artificial intelligence. Two things were stated then: that the group was moving from component supplier to supplier of complete equipment, and that production was expected to start in 2024–2025 and would require recruiting around one hundred people (Présences Grenoble, 29 March 2022).
That forward-looking recruitment figure is a statement of intent from 2022, not a measured outcome. No source we opened confirms how many of those posts were filled, or whether the production start held. We print the intention with its date and refuse to convert it into a present-tense hiring claim.
VII. What we refused to print
Eight claim families were available in loose form and rejected: the 2007 expansion's investment amount; Trixell's market share in any segment; any detector unit price; ImaSpiiR-X or France 2030 funding amounts for this company; salary levels at Moirans; manufacturing yield or cleanroom utilisation; AXEL's revenue, client list or pipeline beyond the "around ten" prospective start-ups mentioned in January 2025; and any present-day headcount stated as a single number. Each was refused for the same reason: no source we opened supports it at the precision the sentence would imply.
What remains is enough to judge the place. A part-owned-by-its-customers joint venture, three decades deep in two specific materials, has chosen the premium end of a market under Asian price pressure, is extending the same physics into industrial inspection, and is renting its qualification competence to the wider ecosystem through a cooperative that a trade union spent twelve years bringing into existence. For a career reader, that last fact is the most operationally useful one on the page: it means the site's scarce skill is not building detectors. It is proving they can be built.