Sam Altman delays OpenAI IPO to 2027, citing AI safety risks

OpenAI will not go public in 2026. CEO Sam Altman told Fortune that a listing this year would be ‘ill-advised’ and pushed the timeline to 2027, citing safety concerns.

The decision came out of an exclusive, hour-long conversation for Fortune’s Titans podcast, recorded in San Francisco. According to Fortune’s reporting, the interview took place on September 12, 2026, during a month when public anxiety about AI had already reached a boiling point.

What Altman actually said

Altman framed the IPO delay as a way to firm up safety and alignment work before OpenAI faces the pressures of public markets. Alignment is the effort to make AI systems reliably do what humans intend, rather than something adjacent and possibly worse.

He was also candid that waiting has costs. Altman acknowledged that an extended delay could carry its own risks, so this is a postponement rather than an open-ended pause.

He laid out two outcomes he considers unacceptable. The first is humans losing control over AI systems. The second is an extreme concentration of power in the hands of a few entities.

The starkest moment involved probabilities. Altman said that even an approximately 10% chance of AI causing human extinction by the end of the decade should be treated as intolerable.

The money behind the pause

The company has reportedly shifted its focus toward large private funding rounds to bridge the gap until a 2027 listing. Speculation puts that raise at around $30 billion, based on a valuation of approximately $1.4 trillion. Neither figure has been confirmed by OpenAI.

For comparison, OpenAI’s last private valuation was approximately $852 billion as of March 2026. If the reported terms hold, the company’s private-market price tag would climb sharply in just a few months.

OpenAI had previously filed confidentially for an IPO. A confidential filing lets a company prepare its paperwork with regulators without publicly disclosing financials, which makes this a retreat from a process already underway rather than a plan that never left the whiteboard.

A month of alarm bells

The interview landed in a charged environment. Earlier in September, a former Anthropic researcher posted a warning on X that AI could destroy humanity and accused major frontier labs of “gambling with our lives.”

Researchers at Anthropic and figures like Dario Amodei, the company’s chief executive, have raised their own warnings about how rapidly AI capabilities are advancing.

Altman took a similar message to a much larger stage on September 23, when he addressed the UN Security Council. He warned that AI progress could outpace society’s ability to understand it or step in when something goes wrong.

His proposed answer was coordination. Altman called for labs, governments, and other stakeholders to work together to capture AI’s benefits while containing its risks.

What this means for investors and the AI race

For investors hoping to buy OpenAI shares on a public exchange, the wait just got longer. Exposure to the company will remain limited to those with access to private rounds, which tends to mean large institutions and well-connected funds.

Altman identified losing human control and concentrating power as the two worst outcomes, and both are problems that public market pressure could plausibly make harder to manage.

Regulators will also be paying attention. Altman’s call at the UN Security Council for coordinated oversight puts governments on notice that at least one major lab wants them in the room, and it gives lawmakers a public commitment to hold him to.