Garry Tan on YC's first Spring batch: 12% revenue growth, 110% more young founders, and the last window to start a company before superintelligence
Jun 11, 2025 with Garry Tan
Key Points
- Y Combinator's Spring 2025 batch reports 12% revenue growth, up from roughly 10% in prior four batches, with 90% of companies working in AI.
- Applications from founders aged 18-to-22 surged 110% year-on-year, driven by a perception that superintelligence will lock in dominant players and close the window for new entrants.
- YC expanded its partner team to 15 and now accepts 0.8% of applicants, with partner capacity rather than deal flow as the constraint on batch size.
Summary
Y Combinator's Spring 2025 batch, the first time YC has used a spring cohort format and rebranded what was previously called X25, runs at 90% AI companies and 11% hard tech. Batch-wide revenue grew 12% over the program, up from roughly 10% across the prior four batches.
The demographic shift is more striking. Applications and acceptances from 18-to-22-year-olds are up 110% year-on-year. Tan attributes this partly to a generational bet that now is the last window to build a company before superintelligence locks in the winners. Once AI reaches that threshold, dominant businesses will hit their terminal value almost immediately through network effects, brand, and cornered resources, leaving little room for new entrants to establish moats. Young founders appear to be racing the clock.
Tan also cites a circulating data point that CS graduate unemployment is currently running at 2x the rate of art history majors, though he hasn't verified it. Whether or not the number holds, it fits the broader pattern he's describing: traditional credentials are losing signal value, and agency and taste are becoming the scarcer inputs.
Platform neutrality
Tan's near-term policy concern is AI platform access. Siri today gives users no choice of underlying model—no Perplexity, no Anthropic, no ChatGPT. Mandatory platform neutrality would be the critical next unlock. If that opens up, hundreds of thousands of companies could each reach $1 billion in net revenue with teams as small as 10 people.
ARR reporting
YC pushes founders to report contracted ARR, with a visible asterisk if the contract includes an opt-out clause such as a 30-day opt-out. That disclosure should happen explicitly rather than buried. Basic honesty here avoids securities fraud, though these are non-GAAP figures and new metric definitions are still being worked out as the market evolves.
Batch dynamics
About 30% of companies change their idea during the batch. Tan views this as a feature rather than a problem. Some of the most significant companies coming out of YC have pivoted within the final two to three weeks. AI tooling has accelerated that cycle, allowing founders to build, test, and redirect faster than in prior cohorts.
Partnership and scale
YC now has 15 partners after adding Andrew Nicholas, John Shu, and a third partner this batch. All three built companies with exits ranging from hundreds of millions to north of a billion dollars, including one to PagerDuty. With each partner working 10 to 25 companies, that implies a batch ceiling somewhere in the 150-to-375 company range. YC's current acceptance rate is 0.8%. Tan says that makes him feel "behind the eight ball" because the constraint is partner capacity, not deal flow. He describes YC as a 10-to-100-year institution-building project, with Airbnb's Brian Chesky, Paul Graham, Jessica Livingston, and Carolyn Levy on the board.