OpenAI isn't going public anytime soon, and Sam Altman is blaming the models, not the market.
Altman told reporters Tuesday, after his DevDay keynote, that OpenAI won't pursue an IPO until it can make "confident safety claims" about its systems. He said AI capability has surged recently, with more gains expected, and that the company's ability to vouch for safety needs to keep pace with that progress. He also conceded the opposite risk: waiting too long to go public would be "bad for the world." He gave no timeline for either the safety bar or a listing.
That's Altman threading a needle he's threaded before. Public markets want predictable quarterly disclosures, and safety claims about fast-moving AI systems are inherently hard to make with confidence, let alone defend to shareholders. An IPO would also open OpenAI's finances and governance up to public scrutiny and lawsuits in a way its current investor structure doesn't.
Anthropic and most other frontier labs have stayed private too, so for now the entire industry is betting that investors will keep funding AI at today's scale without a stock ticker demanding answers.