The companies that will define the next decade are taking shape.Most of them are in one room.
That room is Y Combinator, the world's most successful startup accelerator. Its companies are together worth $1.3 trillion.1
Airbnb, Stripe, Coinbase and DoorDash came out of it. So did Groww, Razorpay and Meesho.
It is the only place we invest, and this is what we believe.
Selection is everything.
More than 10,000 companies apply to Y Combinator every three months, and about one in a hundred gets in.2 That makes it roughly four times more selective than Harvard.3 We start where the hardest filter in startups ends.
A record is worth more than a pitch.
In twenty years Y Combinator has funded more than 6,700 companies.4 PitchBook estimated in 2023 that, of those funded since 2010, about one in twenty had become a billion-dollar company, the highest rate of any leading accelerator.5 Airbnb went public at over $100 billion and Coinbase at $86 billion. Stripe is valued at $107 billion.1
The money is made early.
Y Combinator's chief executive has said that investors who back its companies at Demo Day, batch after batch, have made a median of five times their money.6 He did not state the period covered or how fees are treated, so read it as a direction and not as a benchmark. For scale, the median US venture fund raised between 2018 and 2020 has not yet doubled its investors' money after fees.7
AI is being built by people you have not heard of yet.
We back Y Combinator founders building across AI, from chips and computing infrastructure to data, model evaluation, robotics, and software for security, science and business. We would rather meet them now than read about them later.
Access is earned, not bought.
The best rounds close within days of Demo Day, when the companies first present to investors, and often before most investors have heard the company's name. Getting in takes relationships with founders, a structure that can invest in US companies, and the time to meet a few hundred teams every batch. We do that work.
Nobody can pick the one winner.
Most startups fail, a few do well, and a single winner can be worth more than all the others combined. So we do not try to pick one. Each of our vehicles holds many companies, and one investment is never one bet.
Mettle decides the rest.
After the idea, the market and the money, what is left is whether the founders keep going when it gets hard. That is what our name means, and it is what we look for in every founder we meet.
If you believe this too, we would like to hear from you. Tell us what you are building, or what you would like to invest in, at comms@mettlevp.com.
- 1Y Combinator, ycombinator.com, accessed 10 October 2026. Combined valuation of its companies, and the Airbnb, Coinbase and Stripe figures, as Y Combinator states them.
- 2Y Combinator, ycombinator.com/investors, accessed 10 October 2026: over 10,000 companies apply every three months, with a typical acceptance rate of 1%.
- 3Harvard College admitted 4.2% of applicants to the Class of 2029. Harvard Fact Book, accessed 10 October 2026.
- 4Count of companies in the Y Combinator directory on 19 September 2026.
- 5PitchBook, Y Combinator leads accelerators in unicorn-creation rate, 2023: an estimated 4.5% of startups that went through Y Combinator since 2010. The rate describes that sample only, not all companies funded to date.
- 6Garry Tan, chief executive of Y Combinator, post on X, 19 November 2025: a median multiple of 5x for recurring Demo Day investors. The post does not state the period, the number of investments per batch, or whether the figure is before or after fees.
- 7Carta, VC Fund Performance, Q4 2025 data, published March 2026: median net TVPI of 1.60x, 1.33x and 1.14x for the 2018, 2019 and 2020 fund vintages, across 2,906 US venture funds. Net means after fees and carry. A fund vintage is the year the fund was raised.