Ask what it is building for the people who will run that ecosystem in 2036, and the room goes quiet.
Entrepreneurship education is usually classified as a service delivered to students. That classification is the reason it is funded last, reported on annually, and evaluated by the number of companies it produced this year. It is the wrong classification. Education is infrastructure - and it behaves like infrastructure in the one way that matters most: the return arrives long after the budget cycle that paid for it has closed.
What education produces beyond founders
The narrow view says entrepreneurship education produces founders. It does, and that is the smallest part of its output.
It also produces informed employees - people who join early-stage companies understanding what ambiguity, runway and pivots actually feel like, which is the difference between an early hire who survives the first crisis and one who leaves during it. It produces future mentors, because today’s participant is the person who returns in eight years to coach the next cohort. It produces future investors, particularly business angels, who overwhelmingly come from the operator population. It produces community members who show up to things, connect people, and make the network dense enough to be useful. And it produces future ecosystem leaders - the program directors, policy people and community builders who will be running the institutions a decade from now.
That is not a list of side effects. It is the actual mechanism by which ecosystems compound:
Access → education → experience → validation → venture building → first revenue → community → ecosystem → compounding.
The chain closes when today’s participants become tomorrow’s founders, mentors, leaders, investors and ecosystem builders. Nothing about that chain can be accelerated by spending more in one year. Each link needs the previous one to have happened, and several of them need people to have gained real operating experience in between.
Brad Feld’s Boulder Thesis reached the same conclusion from the community side: entrepreneurs must lead, leaders need long-term commitment measured in decades rather than quarters, the community must be inclusive of anyone who wants to participate, and it must sustain continual activity across the entire entrepreneurial stack. Education touches three of those four principles directly. It is the primary inclusion mechanism, the main source of continuous activity between flagship events, and the pipeline that produces the next generation of leaders.
The long-horizon evidence is visible where commitment has been sustained. In the 2026 European Deep Tech Report, ETH Zurich leads globally with 192 alumni-founded, VC-backed deep tech startups since 2020, ahead of EPFL (94) and Cambridge (67). Together, ETH and EPFL account for 289 VC-backed spinouts, 8 unicorns, around $24.2 billion in combined enterprise value and roughly $6.4 billion raised. Switzerland’s deep tech share now sits at roughly 40% of total VC volume, among Europe’s top three.
Those 192 startups did not start in 2020. They are the visible output of institutional structures built over decades - which is exactly why the number is useful as an argument and useless as a target.
The ten-year payoff and the three-year funding cycle
Everything in the previous section runs on a decade-long clock. Almost everything funding it runs on a three-year one.
Article 5 in this series documented the German version: 77% of university startup-support budgets come from third-party funding, and 84.5% of institutions are formally asking for permanent structures. Article 8 documented the measurement version: programmes are evaluated on ventures founded within the reporting period, which is the one output the compounding chain produces last.
Put those together and the pattern is clear. The slowest-compounding asset in the ecosystem is funded on the shortest cycle and measured on the shortest horizon. That is not an oversight in any single institution. It is a systemic mismatch between the nature of the asset and the machinery that funds it.
Policy is starting to acknowledge this. The Nationales MINT Forum’s 2026 position paper proposes a continuous “Entrepreneurship Journey” competence model running from primary through upper secondary education, drawing on the Austrian approach - alongside teacher qualification in entrepreneurship and maker pedagogy, maker spaces and open workshops, a central coordinating body to connect the many existing initiatives, stronger entrepreneurship centres and incubators for STEM graduates, and nationwide competence monitoring. At European level, the EU Startup and Scaleup Strategy, adopted in May 2025, puts talent among its five pillars, with the European Startup and Scaleup Scoreboard as a monitoring instrument. And the German co-creation data suggests the collaborative substrate is already forming: projects between universities, business and government more than doubled to 1,910, and 55% of universities now cooperate with international investors.
But ecosystem builders do not have to wait for any of that.
Four things are available without a single policy change. Run multiple touchpoints across the year instead of concentrating everything into one flagship event - continuity beats intensity, and the Boulder Thesis has said so for over a decade. Build explicit alumni pathways back in as mentors, so the compounding loop actually closes instead of leaking. Keep at least one permanently open, low-threshold entry point that requires no idea, no team and no application. And measure what emerged between people - teams formed, partnerships started, mentoring relationships created - rather than how many people attended.
An ecosystem is not a place. It is a relationship system, and education is how new people enter it.
The goal was never more startups. It was more people who believe they can build one, and enough of a place for them to try.
