46 companies in one year: how research becomes business.
ETH Zurich recognised 46 new ventures in 2025 and its companies raised CHF 540 million. The more instructive figure is 661: the cumulative number of ventures the university has produced since 1973, almost all of them small Swiss firms.
Gli Insights sono pubblicati in inglese.
ETH Zurich announced on 25 February 2026 that it had recognised the foundation of 46 new ETH Ventures during 2025, the highest number in its records. Twenty-four were classified as ETH Spin-offs, whose business is based on research findings from the university, and twenty-two as ETH Start-ups, founded by ETH members using knowledge acquired while studying or working there but not reliant on ETH research. In 2024 the university recognised 37 spin-offs.
Companies created at ETH Zurich secured CHF 540 million across 41 financing rounds reported in 2025, 27 per cent more than the year before. Twelve of those rounds exceeded CHF 10 million; the largest, CHF 103 million, went to Auterion AG, which develops software for drones. Technology and software, including in particular artificial intelligence and machine learning, accounted for 26 per cent of the new ventures, ahead of biotech and pharmaceuticals at 15 per cent.
The record is the least useful part of the announcement. What matters for an operating company is the machinery underneath it: a transfer process that rewrote its own rules in July 2025, and whose cumulative output is a population of small Swiss suppliers.
The count rose partly because the definitions changed
ETH Zurich says as much in its own release: the 2025 figures “cannot be directly compared to those from previous years”, because the new regulations for creating ETH Ventures moved companies that would once have been recognised as spin-offs into the new start-up category.
Those regulations — the Business Creation Regulations, in force since 10 July 2025 — replaced the earlier spin-off guidelines with a documented procedure. Two labels now exist. ETH Spin-off applies to companies built on ETH research results, where the founders must settle intellectual property and exploitation rights with ETH transfer. ETH Start-up applies to ventures founded on knowledge gained at the university rather than on its research. ETH takes a baseline equity share of 2 per cent in spin-offs, which can rise according to the patent licensing option chosen — more equity against lower royalties, or the reverse — and according to participation in ETH incubator programmes. The university describes 2 per cent as very moderate both internationally and within the ETH Domain. A standard case with its documentation ready can be incorporated within six to eight weeks through an express route offering three pre-defined licensing options.
A second change followed in November 2025 with UPortunity, an accelerator for ETH Start-ups enabled by UBS, whose fellows receive a monthly salary, a project budget, office and laboratory space, and mentoring from experienced founders.
Published terms and a fixed timetable distinguish a transfer process from a series of one-off negotiations.
Five decades of transfer have produced a population of small Swiss firms
By the end of 2025, 661 ventures had carried an ETH label since 1973 — 639 spin-offs and 22 start-ups. Of those, 484, or 73 per cent, remain independent and privately held, with six more listed; 79, or 12 per cent, have been acquired or merged; and 92, or 14 per cent, have ceased operations.
Their durability is unusual. The Ventures Report 2025 puts the five-year survival rate at 95 per cent and the ten-year rate at 84 per cent, against a Swiss average it cites of about 50 per cent after five years and 35 per cent after ten.
Their size matters more. Across the 490 independent ventures, 60 per cent employ between one and nine people, 31 per cent between ten and forty-nine, and 8 per cent between fifty and 249. One per cent employ between 250 and 1,999, and 0.2 per cent more than that; the largest of them is u-blox AG. Ninety-nine per cent operate from Switzerland, although the founders of the past decade came from 49 different countries.
The output of this transfer process is therefore overwhelmingly small and medium-sized enterprises, based in Switzerland, in the same size class as the companies that might buy from them. That is a different proposition from the one usually implied by the word deep tech.
The route to a research result now runs through a company more often than through a licence
ETH Zurich concluded 37 intellectual property licences in 2025, of which 29 were with its own spin-offs. That share, 78 per cent, compares with 52 per cent in 2020, when 17 of 33 licences went to spin-offs. The balance goes to other industry partners, and it has narrowed: eight licences in 2025, against seventeen of a larger total of 49 in 2024.
The consequence is practical. Licensing an ETH result at the source is becoming the exception; more often the result already sits inside a young company that holds the licence. The counterparty for a manufacturer looking for a new sensor, material, inspection method or model is therefore increasingly a firm of two to ten people, and the instrument required is commercial — a supply or co-development agreement — rather than a technology transfer negotiation with a university.
Scale is worth keeping in proportion. The Swiss Venture Capital Report 2026 counts CHF 2.95 billion invested in Swiss start-ups in 2025 across 354 rounds, 23.9 per cent more than in 2024, of which CHF 1,192.5 million in the canton of Zurich. The CHF 540 million raised by ETH ventures is the equivalent of roughly 18 per cent of that national total, from one institution, though the two counts use different perimeters.
Three Swiss and European instruments make the first step inexpensive
A company that wants to test whether any of this is usable does not have to start with an investment case. Three instruments define the cost in advance.
- The Innosuisse innovation cheque. Worth up to CHF 15,000, it covers 100 per cent of the research partner’s costs for a preliminary study: a concept development, a feasibility question, an assessment of market potential. Applicants must be based in Switzerland with a valid UID and fewer than 250 full-time equivalents, counted across the group. Applications are open at any time, evaluation takes four to six weeks, and the money goes directly to the research partner.
- Innosuisse innovation projects with an implementation partner. For a full project, the company self-finances 40 to 60 per cent of total costs and contributes at least 5 per cent of total project costs in cash towards the research partner’s expenses. Applications go in at least six weeks before a decision meeting of the Innovation Council, a decision follows within six to eight weeks, and the project must start within three months of the agreement taking effect.
- Horizon Europe. Following the EU Programmes Agreement signed on 10 November 2025, Switzerland is associated to Horizon Europe retroactively from 1 January 2025, together with Euratom, the Digital Europe Programme and ITER. Swiss entities can participate as beneficiaries from the 2025 programme year, including in European Innovation Council and European Research Council calls; restrictions remain in the Digital Europe Programme’s cybersecurity and semiconductor objectives.
The sequence is short. Write down one technical question the company cannot answer with its own staff. Check the published venture lists, ETH’s and other Swiss institutions’, before opening a supplier search. Establish the company’s own share of the cost before applying, since both Innosuisse instruments require it.
What the next reporting cycle will show
Three publications will test whether 2025 was a turn or a peak. ETH Zurich issues its Ventures Report each February; the next edition will show whether the spin-off and start-up split stabilises, and whether licences to industry partners outside the spin-off portfolio recover from eight. The Swiss Venture Capital Report follows the same rhythm; in its 2026 edition, 74 per cent of the investors surveyed expected Swiss investment volume to rise by at least 25 per cent in 2026, an expectation the next edition will test. WIPO publishes the Global Innovation Index each September; in 2025 Switzerland ranked first, as it did in 2024, while global R&D growth slowed to 2.9 per cent in 2024 from 4.4 per cent a year earlier.
None of this is a matter of national standing. The measurable question is narrower: how reliably research becomes a company able to issue an invoice. Forty-six in one year, 661 across five decades, and 95 per cent still trading after five years describe an infrastructure rather than an event, and it is open to whoever approaches it.