Sam Altman learned at Y Combinator that powerful leaders can govern—but they can’t rule

Sam Altman thinks one of Silicon Valley’s most powerful jobs came with a useful limit. At Y Combinator, he learned that even a leader at the center of the action doesn’t get to simply command everyone else. “At best you can govern—you don’t get to, like, rule,” he told Fortune’s Alyson Shontell.

That lesson seems to be following him into OpenAI. Altman said the company sits under pressure from governments, competitors, investors, and the broader AI world, all pulling in different directions on how fast the technology should move. He described the work as a series of tense negotiations, not a clean chain of command.

His YC years helped him see how to handle that. He first came through the accelerator as a founder, then returned as president in 2014 and stayed until 2019. Because YC was an investor, not the owner of every startup it backed, he had to deal with companies as independent actors, sometimes by telling them they were making a big mistake and needed new norms rather than issuing orders.

Altman says that approach matters even more at OpenAI, where he defended the company’s “incredibly complicated structure” and argued it may need to do things that are not obviously in the interest of shareholders. He also said now would be an “ill-advised moment” for OpenAI to go public, given AI safety concerns, and said he’d be willing to tell investors the company had to take a path “extremely against your financial interest right now” if the public good required it.

He went further, suggesting the biggest AI safety decisions may have to be coordinated across companies and eventually governments. He floated the idea that AI leaders could reach some form of shared agreement, and even said the U.S. and China could do something worthy of a Nobel Peace Prize if they settled on common standards and testing. For Altman, the old YC lesson has turned into a much bigger test: govern the ecosystem, don’t pretend to rule it.