YC data: 454 repeat founders, and no proof second time is better

Y Combinator has handed over a dataset covering 454 founders who have been through its programme more than once, plus 935 founder-company records running from 2005 to 2026.

The headline claim attached to repeat founders is that the second company goes better. This dataset does not test that. It counts appearances, not outcomes.

What it does show is that coming back is almost always a two-part affair. Some 428 founders, or 94%, went through YC exactly twice. Only 25 turned up three times.

Justin Kan, a co-founder of Twitch and Stash, is the only founder in the data to have been through four times.

The average gap between appearances is 5.1 years. That average hides two different habits.

Nearly 30% of returns happened within two years of the previous company, including 38 in the same calendar year. At the other end, 61 returns came a decade or more later.

Repeat-founder numbers peak in the most recent data, at 65 in 2025. Crunchbase News cautions that YC's cohorts have grown considerably, that the 2025-26 figures include newer batch formats, and that the recent data may be incomplete. A rising count is not the same as a rising rate.

Whole teams come back too. The founding groups behind Layer by Layer, Voodoo Manufacturing, Ultra, Blair and Fastgen each returned together for a subsequent company.

Aaron Epstein, a YC general partner who worked the spring 2026 batch, says he had a number of second-time founders he had advised before. He is careful about the framing: the alumni base keeps growing, so more people are eligible to return.

His account of what repeat founders do differently is about spending, not brilliance. He says the mistake they avoid is overhiring and overspending before product-market fit.

Sherwood Callaway is one of them. His first company, Opkit, an insurance verification and revenue-cycle software business, went through the fully remote summer 2021 batch and was later acquired by 11x AI.

His second, Sazabi, is an observability platform pitched against incumbents including Datadog. It announced an $8m seed round in late June led by J2 Ventures, Village Global and Y Combinator, with Orange Collective and more than 60 angels from companies including Vercel, Cursor and OpenAI.

Callaway says he treated the second batch differently, deferring it to build more of the product first and using YC as a go-to-market event rather than a starting gun. That is a use of the programme you can only plan if you have already been through it.

For a founder reading this, the useful figure is 5.1 years, and the 30% who came back inside two years. The data says nothing about whether any of them did better the second time. Anyone selling you a second-time-founder premium on the strength of this dataset is selling you a count.