The Scale-Up Trap

The European innovation system is not empty. Horizon Europe, the European Innovation Council, the EIT, national agencies, ESA, EUSPA and CASSINI, the Innovation Fund, InvestEU and the EIF all move real money. A founder can be busy with funding for years and still have no company that a customer will renew.

The trap is the crossing. Public instruments are built to prove a technology, staff a consortium, or award a prize. Growth capital is built to own a share of a company that can sell the same thing, again, in more than one country. Between those two logics sits a firm that is too proven for another grant and too small for a lead investor who can write a growth round without a foreign partner.

I have watched that crossing from the jury side of CASSINI and from bid work on ESA-funded Earth Observation. The pattern shows up before the cap table does. A space or deep-tech team can win a challenge, an academy place and a pilot, and still have nobody in Europe who will lead the round that turns the pilot into a line of business.

Research, roughly TRL 1 to 4, is where Europe is generous. Collaborative grants buy papers, prototypes and doctoral time. They do not buy a company. Treating them as startup capital is how good labs learn the wrong habit: a proposal calendar instead of a customer calendar.

Spin-out and early venture, TRL 4 to 6, is where most of what people call innovation funding actually sits. EIC Pathfinder and Transition, EIT communities, ESA BIC, national incubators. The money is competitive and slow relative to a startup clock. It is still the right money if the job is to leave the lab with a team, not a slide.

Seed to the first serious round is the EIC Accelerator’s job: a grant sized in the low millions, plus an equity component through the EIC Fund that is supposed to crowd in a private lead. When the private lead is real, the instrument works. When the grant arrives alone, it buys eighteen months of runway into the same gap.

Scale-up is the hole. The European Investment Bank’s 2024 stocktake is blunt. By year ten, EU scale-ups raise about half as much capital as San Francisco peers. US venture investment runs six to eight times the EU figure in a typical year. More than four in five EU scale-up deals have a foreign lead or a sole foreign investor. The scarcity is not of ideas. It is of European leads who can set the terms.

Space and Earth Observation have a public good the rest of the world envies: Copernicus, Galileo, EGNOS. The commercial layer on top is thin. CASSINI was built to create ventures and a habit of entrepreneurship around EU space data. It was not built to replace a Series B. Downstream companies die between an agency pilot and a contract a finance director in another member state will sign.

Climate and hard tech sit on a different instrument, the Innovation Fund, fed by the ETS and sized in the tens of billions through 2030 for demonstration and first industrial deployment. That is not startup venture money. By 2026 the public record already showed a portfolio under strain: projects slipping their own financial-close dates, and a rising share terminated before close. A grant that cannot become a bankable offtake is a different failure from a missing seed round, and it should not be managed with the same slide.

Digital and AI companies move faster, which makes the trap shorter and sharper. Enterprise buyers are still national. A product that is legal, procured and referenced in one member state starts again next door. Hyperscalers do not have that problem. They buy distribution while the European firm is still translating a security questionnaire.

Quantum, chips and dual-use look strategic in every communiqué. The buyer is still national. A twelve-person firm cannot sell “Europe”. It sells a ministry, a prime, or nobody. Defence innovation and space entrepreneurship are starting to share a stage. They do not yet share a purchase order.

A lab spin-out has intellectual property and, often, a technology transfer office. It rarely has an operator who has shipped. Funding exists. The scarce input is a company, not a project.

A venture-backed startup can combine angels, a national fund and an EIC ticket. The trap opens at the round that needs a lead above the local fund’s concentration limit. That is usually the round where a US or UK investor appears, and with them a drift in where the company is willing to incorporate, hire and exit.

An industrial incumbent captures the consortia and the procurement. The startup is a work package in the proposal and a line item afterwards. Maturity, in this system, is too often measured by how well you fit a call text, not by whether a customer would be in trouble if you disappeared.

A challenge, a hackathon or a CASSINI-style venture programme is cheap per company and high leverage on deal flow. I have seen it produce founders. It does not produce a scale-up. Using it as a substitute for growth equity flatters the programme and strands the company.

A multi-million grant is worth it when it retires a technical risk a private investor cannot price. It is expensive when it retires no risk and adds a reporting culture. The benefit is not the headline euro amount. It is whether a private lead will follow, on terms the founder can live with, inside a year.

A scale-up fund in the low single-digit billions — public anchor, private manager, the shape of the Scaleup Europe Fund — is the right order of magnitude only if it can lead, and only if the company can sell across borders afterwards. Capital without a market is a more comfortable trap. The cost of doing nothing is already in the EIB numbers: half the capital, foreign leads, exits that leave.

The Scaleup Europe Fund is moving from design to first cheques. The useful test is not assets under management. It is whether the fund leads rounds in strategic technologies, and whether those companies still sell in Europe three years later. If it behaves as a large private return vehicle with a public badge, the trap moves one stage later and gets a better logo.

Capital markets union remains the binding constraint. More public money on top of twenty-seven shallow venture markets recreates the gap at a larger ticket. Defence and dual-use demand will pull space, sensing and cyber startups toward procurement that is slower than venture capital and larger than grants. Teams that cannot survive a two-year public-buyer cycle should not be told that EUDIS is their Series B.

The next Earth Observation companies will sell decisions, not pixels. Funding models built for satellite hardware, and consortia built for sensors, will miss them. AI does not fix a fragmented customer. It makes a single-country reference account look even smaller.