OpenAI CEO Sam Altman has stated plainly that the company will not take its shares public in 2026. He made the remark in an interview with Fortune, and the reason he gave was not market conditions but safety. OpenAI is reported to have already filed confidentially for a listing, and the deal has drawn attention as a roughly 1 trillion USD offering. That clock has now been pushed back again.
"An ill-advised moment to go public"
The interview was conducted by Fortune editor-in-chief Alyson Shontell at OpenAI's San Francisco headquarters and published on September 12. Altman said the company is not rushing into an IPO, and that given everything happening around safety, right now would be an ill-advised moment to go public.
Pressed on whether a listing could still happen this year, he answered that it would not be 2026, and that there is a lot of work to do. What he meant by that work is meeting the standard now required for safety and alignment, the task of keeping a model's behavior consistent with human intent. The framing puts the bottleneck not in the technology itself but in the organization handling it.
Altman also touched on OpenAI's complicated structure, which combines a non-profit arm with a for-profit one, describing it as the shape needed to make decisions that stay tied to the mission. Going public would import a quarterly clock into that structure. Grounding the delay in safety reads, in part, as an acknowledgment of how poorly the two fit together.
Three months after the filing, the timeline keeps slipping
Reports that OpenAI had filed confidentially for a listing appeared in June 2026. The second half of 2026 was initially in view, but later reporting pointed to 2027 as volatility in tech stocks and the weight of the company's finances took hold. Altman's remarks now put those reports in the principal's own words.
The valuation attached to the offering has been reported at roughly 1 trillion USD (about 153 trillion yen). A deal at that size would rank among the largest public offerings ever, which is why the disappointment on the market side is not small. Since the start of the year, 2026 had been described repeatedly as the year both OpenAI and Anthropic would head for the public markets.
※1 USD = 153 JPY (based on the New York close on September 11, 2026)
Calling extinction risk unacceptable
The more striking part of the interview was not the IPO but the way Altman talked about safety. He said a risk of human extinction would be unacceptable even at 10 percent. Whether the number is 10 or 8 or 6, he argued, the point is that everyone involved carries an enormous amount of responsibility. He acknowledged how hard such probabilities are to pin down, and still took the position that the work should proceed as if the risk cannot be tolerated.
He was equally direct about the possibility of AI moving beyond human control, saying it absolutely could happen. Safety standards, in his view, are not yet at a level that justifies pushing capabilities further, and the company has been discussing the option of pausing at the edge of a new capability level to let safety work catch up.
A leader who has driven the pace of development is now the one raising the idea of a speed limit. That reversal says a good deal about the mood across the industry in recent weeks.
No single company settles this
The other new element is the assumption, now stated openly, that slowing down is not a decision one company can make alone. Altman said he agrees with Anthropic CEO Dario Amodei on the need to pace the frontier, and noted that the subject has been a leading topic inside OpenAI in recent weeks. Reporting suggests the major labs may be close to an arrangement to coordinate on development speed.
What that arrangement would bind, and which companies it would cover, has not been spelled out. Critics have long pointed out that a handful of firms falling into step means little while competition continues elsewhere, including among Chinese developers.
On the political side, calls for tighter rules have grown across party lines following incidents of autonomous AI agents going off course and a string of departures among safety researchers. The decision to hold off on a listing was made in that atmosphere. Safety as a stated reason and a market that is not welcoming to a listing can both be true at once, so whether the remarks are taken entirely at face value is another question.
Summary
In a Fortune interview published on September 12, Sam Altman made clear that OpenAI will not go public in 2026. He pointed to safety and alignment work as the reason, and went as far as calling an extinction risk unacceptable even at 10 percent. On the pace of development, he aligned himself with Anthropic's Dario Amodei, bringing coordination among the major labs into view as well. For an offering valued at roughly 1 trillion USD, the timeline now points to 2027 at the earliest.
