Grenoble Deep Tech Startups: Three Pools, Three Screens

Most Grenoble screens start the same way: a list of 475 still-active startups in the Grenoble Alpes region in 2025, 44% of them deeptech, sorted by raise size. That list will not tell you which of those companies you can actually buy.

Grenoble deep tech startups sit in three pools that need separate screens, and for a power electronics, sensors or quantum-adjacent mandate the middle pool is where completed deals live. Transfer-office spinouts out of CEA-Leti, Spintec and Linksium are licence positions before they are companies. The Minalogic and inovallée tier of profitable design houses and subsystem suppliers is what semiconductor strategics have bought. The Hydro21 hydro-engineering tier is capacity you buy from, not invention you buy.

Three Grenoble company pools, and the five questions that sort them

The three pools are the transfer-office spinouts, the Minalogic and inovallée SME tier, and the Hydro21 hydro and energy engineering tier. Every public map of the region merges them, so they look like one ecosystem. In diligence they behave nothing alike.

Run the same five questions across all three:

Question one is Grenoble-specific, and the reason funding-stage filters fail here. CEA's intellectual property charter states that the commission generally retains ownership of the IP it develops and licenses it to industrial partners through post-R&D licences, with preferential terms for startups exploiting its technology. A CSIS analysis of the CEA-Leti model counts roughly 80 spinouts, about 75% still in business and three public (Soitec, Kalray, Tronics), and notes there is no single IP template: arrangements are negotiated project by project. So the first diligence item on any CEA-Leti or Spintec company is the licence file, not the cap table. Skip it and you may buy a team with conditional access to the thing you wanted.

CEA-Leti spinouts, Minalogic design houses and Hydro21 suppliers side by side

The three pools diverge most sharply on who owns the technology and on what a closed deal there has historically looked like.

The pattern in the right-hand column is the claim page one will not make. Grenoble's realised exits have gone to semiconductor strategics buying a process or a device line, not to acquirers buying a venture story. Rank the region by raise size and you systematically sort away from the companies that have been bought.

What you inherit when you buy a Quobly, Diamfab or SteerLight

Transfer-office spinouts hold the technology the region is famous for and the weakest ownership position of the three pools. Core IP in a CEA-Leti, Spintec or CNRS Institut Néel spinout frequently sits with the institute under a negotiated licence, so the question is scope: fields of use, exclusivity, sublicensing, territory, and what happens to the licence on change of control. Equity brings a second negotiation: French deeptech rounds here routinely include Bpifrance and institutional deeptech funds, so check whether a CEA-affiliated investor such as Supernova Invest is on the register before you model a clean 100% purchase.

Activity here reads from industrial attachments, not announcements. Quobly, the CEA-Leti and CNRS spinout building silicon-based quantum computers, raised €115M in June 2026 with cloud access in 2026 and an HPC deployment targeted for 2027. Diamfab, a 2019 CNRS Institut Néel spinout making diamond substrates for power electronics, raised €8.7M from Asterion Ventures, Bpifrance's French Tech Seed and Kreaxi, and works with CEA, Schneider Electric, Soitec and STMicroelectronics. SteerLight sits in the same tier on silicon photonics lidar. Named industrial partners are the signal, because a power-electronics startup that Schneider and ST are working with has passed a screen no directory runs.

Two deals in this pool show what strategics pay for. STMicroelectronics agreed on 5 March 2020 to take a majority stake in Exagan, the Grenoble GaN-on-silicon company founded in 2014 with CEA-Leti and Soitec support, with the minority to follow 24 months after closing. Allegro MicroSystems closed its $420M cash acquisition of Crocus Technology on 31 October 2023, a 2006 Spintec spinout with R&D in Grenoble, to accelerate its TMR sensing roadmap. Both buyers wanted a device line inside their own roadmap. Neither bought a brand.

Linksium, the Grenoble SATT, reports roughly 87 to 91 launched deeptech startups against a target of 15 a year, backed by a €57M ten-year state commitment. Treat that roster like a national programme cohort: selected against transfer and policy goals, so it is a pipeline, not a short list. The same logic applies to sourcing the anchor institution rather than the city brand.

The Minalogic and inovallée tier that startup lists leave out

Grenoble's profitable design houses and subsystem suppliers own their IP, and that is exactly why they are missing from the lists. They rarely raise venture money, so a funding-stage filter cannot see them, and a self-service directory shows them next to branch offices and service firms with no way to tell a 10-person incubator resident from a 100-person ASIC team. The inovallée technopole in Meylan and Montbonnot alone describes roughly 380 companies and 12,000 jobs, with about thirty startups across two incubators. Most of that count is not acquirable and a small slice of it is the whole point.

Ownership here is clean. IP was built on paid customer programmes, cap tables are usually founder or family held, and there is no institute licence to unwind. Activity proof is commercial rather than promotional: repeat design wins, tape-outs, qualified processes, headcount that has grown for three years. Custom ASIC and SoC houses working in nanopower, analogue interfaces and power management, such as IC'Alps in the Grenoble area, sit squarely in this tier and map directly onto a power-electronics or sensors mandate.

This is the pool where a desk screen breaks down, because the companies that fit it publish the least. Chibit exists for that gap: describe the mandate to Innovation Scout and the output is companies that are currently active and relevant to it, not a tenant roster to sort by hand. Score what comes back on relevance, activity and fit before anyone sends an email, and the same approach works along any deep-tech supply chain where the suppliers outnumber the startups.

Why Hydro21's 110 member firms belong on a vendor list

Hydro21, founded in Grenoble in 2002, brings together more than 110 companies in hydroelectric engineering, including GE Hydro, EDF Hydro, CNR and Artelia, with over 10,000 people working in hydro across Grenoble, Chambéry and Lyon. IP in this tier is engineering know-how and an installed base. Equity sits with corporate parents or private owners, and the large names are acquirers themselves. Activity proof is framework contracts and plant references.

Hydro power is why the valley industrialised and why the cleanrooms have cheap firm power, so it keeps showing up in ecosystem copy. The tier is still capacity. If your mandate is grid or generation equipment, it produces suppliers, joint-venture partners and the occasional specialist subcontractor worth owning. It does not produce the device-level technology a semiconductor or sensors buyer came to Grenoble for.

Picking a Grenoble pool against your mandate

Match the pool to what the mandate is short of. A power electronics mandate short on device technology should screen the transfer-office spinouts first, with GaN and diamond substrate work as the entry point, and budget legal time for the licence before the LOI. A sensors mandate short on a shipping product line should screen the Minalogic and inovallée SMEs first, because Crocus is the precedent and design houses are the asset class that trades. A quantum-adjacent hardware mandate belongs in the spinout pool by necessity, as a watch-and-partner position rather than an acquisition in 2026. An energy or grid mandate should treat Hydro21 as procurement.

Grenoble is the wrong move when the mandate is software, digital health or anything that needs a 50-person commercial team attached. Regional funding is heavy, with €1.2bn of deeptech investment in 2023 that put the city level with Munich for fourth in Europe, and nearly €4bn raised since 2000. That capital went into physics, not go-to-market. A buyer who needs revenue on day one will spend six months learning that lesson.

One timing note worth building into a 2027 plan. The €830M FAMES pilot line, coordinated by CEA-Leti with 11 research organisations from eight countries, was inaugurated in Grenoble on 30 January 2026, adding 2,000 m² of cleanroom for 14,000 m² total and targeting FD-SOI More-than-Moore devices for computing, connectivity, sensing and power management with 40-plus industrial partners including STMicroelectronics, Soitec, Siemens, Nokia, GlobalFoundries and ASML. Access to a FAMES process window is a harder activity signal than a seed round, and the first companies to use it are the ones to watch next.

FAQ about sourcing Grenoble deep tech companies

Who owns the IP in a CEA-Leti spinout?

CEA generally retains ownership of the IP it develops and licenses it to industrial partners, according to its own intellectual property charter, with preferential terms for startups exploiting its technology. Terms are negotiated per project, so a buyer has to read the specific licence rather than assume a standard template.

Are Linksium startups ready to acquire?

Linksium's roster of roughly 87 to 91 launched companies is a transfer pipeline selected against technology-maturation goals, not a short list matched to a corporate mandate. Expect pre-revenue teams, institutional shareholders and licence dependencies; use the roster to find technology, not a transaction.

Which Grenoble deep tech companies have actually been acquired?

STMicroelectronics took a majority stake in GaN power specialist Exagan in March 2020, and Allegro MicroSystems completed a $420M purchase of Spintec spinout Crocus Technology in October 2023 for its TMR magnetic sensing. Both buyers were semiconductor strategics acquiring a device line, which is the shape most Grenoble deals take.

How do you tell an operating Grenoble spinout from a dormant one?

An operating Grenoble spinout shows current industrial attachments: named development partners, cleanroom or pilot-line time, customer programmes, and hiring in the last two quarters. A dormant one shows a funding announcement, an unchanged website and no verifiable customer, which is the distinction a startup database cannot make.

Does Grenoble fit a mandate outside semiconductors and energy?

Grenoble fits mandates in power electronics, sensors, photonics, quantum hardware and hydro and grid engineering, where the research base and the supplier tier are deep. For software-led or commercially-led mandates, other European ecosystems carry more of what you need.

Power electronics and sensors buyers get the most out of this region when they walk in with the licence question, the ownership question and a named pool, not a city-level list. Tell Find Startups what your industry and region are, and Chibit comes back with active, vetted companies matched to it, so the Grenoble work starts from a short list worth acting on.