Student Startups Are Outlasting the Accelerators That Launched Them

The institutions that taught students to build are quietly closing their own doors, while the companies born inside them keep going.

In Bristol, a hub that spent thirteen years catalysing the city’s innovation ecosystem will close in December 2026. In Andhra Pradesh, a university incubation centre that nurtured roughly forty ventures over four years faces dismantlement. In crypto, MegaETH has shuttered its flagship accelerator after backing startups that collectively raised $80 million—only to watch most of them migrate to competing networks. And at Princeton, the university’s new AI accelerator launched months behind the students it hoped to retain, several of whom had already left for Y Combinator.

The pattern is uncomfortable but increasingly hard to ignore: the accelerators, incubators and innovation hubs that taught a generation of students how to build companies are proving far less durable than the companies themselves.

Part of this is arithmetic. Accelerators are fixed-cost infrastructure—buildings, staff, cohort cycles, reporting lines. Startups, once launched, are mobile. They can change cities, networks and even blockchains. When MegaETH declined to take equity in the teams it incubated, it bet on alignment. The bet failed. Its most successful projects left for Base, Monad, or their own chains. The accelerator closed; the startups did not.

But the deeper story is about what student founders actually need—and what institutional programmes are structurally unable to provide. Research on student entrepreneurship suggests that students’ greatest weakness is not ambition or technical skill but network poverty: they begin with almost no professional connections, which makes early customer discovery and stakeholder commitment brutally difficult. A well-run university entrepreneurship programmes can partially compensate. But it can also become a trap. One longitudinal study of venture-creation programmes found that student founders who became too embedded in university networks failed to reach out to outside business contacts, and their ventures suffered after the programme ended.

The accelerators that are closing were, in many cases, victims of their own success at the wrong metric. They counted cohorts, demo days and capital raised. They did not count what happened three years after graduation, when the mentor email address bounced and the desk in the incubator had been reassigned. A widely shared critique from within the university commercialisation world puts it bluntly: programmes produce cohorts that consume a grant, build a prototype, and close within a year—because they provide capital and mentorship while skipping the operating discipline that sustains a business through its first eighteen months.

The student startups that outlast their accelerators tend to share a few traits. They are lean enough to survive the loss of free office space. They have at least one founder who can sell, not just build. And crucially, they have begun replacing the programme’s network with their own before the programme ends.

This is where a different model of university entrepreneurship becomes instructive. ESCP , the world’s first business school, established in 1819, has approached the problem from the opposite direction. Its Extension School, launched in March 2025, does not run a classic accelerator at all. Instead, its new entrepreneurship programme—delivered mainly online over six months and 150 hours—is open without selection based on academic background, awards a certified credential recognised in France, and is designed for working professionals and career changers who cannot afford to pause their lives for a campus cohort. The partnership with Qui Veut Être Mon Associé ? Amplify brings television investors into the classroom to share failures as well as successes, and participants are pushed to test ideas against real-world challenges rather than perfect a pitch for a demo day.

The logic is subtly different from the accelerator model now in retreat. Rather than housing founders in an institution and hoping the network sticks, the programme teaches them to build a network that exists independently of any building. Rather than optimising for a cohort’s graduation, it optimises for what a founder can do once the programme is over. That distinction—between producing a cohort and producing a capability—is precisely what separates the startups still trading today from the hubs that launched them.

There is no tragedy in an accelerator closing, provided the companies it launched are still alive. The tragedy would be the reverse: institutions that endure while the ventures they were meant to serve quietly disappear. The evidence suggests the opposite is happening. The hubs are going. The startups, at least some of them, are staying. The lesson for universities is not to build more accelerators, but to build programmes that make themselves unnecessary—and to measure success by the founders who no longer need them.