Sam Altman sat down with Fortune's editor-in-chief on Friday at OpenAI's San Francisco headquarters and closed the most anticipated listing window in tech. "I actually think that, given everything happening with safety, right now would be an ill-advised moment to go public, and we don't feel pressure on that," he told Alyson Shontell. Pressed on whether that ruled out the year, he was explicit: "I would say not 2026. Yeah, we got a lot of stuff to do, like meeting this moment of what is going to be required for safety and alignment, and how the industry and governments can work together."
The reason he gave was safety. The reason on the record since June was the market.
The timeline the safety story has to overwrite
OpenAI filed a confidential draft S-1 on June 8, a little over a week after Anthropic filed, and announced it in a blog post only because it expected a leak. The hedge it volunteered then now looks prescient: "We have not decided on timing yet; it may be a while because there are things we want to do that are likely easier as a private company."
On June 25, the New York Times reported that OpenAI had hired bankers and lawyers targeting a third- or fourth-quarter 2026 debut but was leaning toward 2027 — citing tech-stock volatility and its own financial challenges. The proximate cause was SpaceX, whose IPO raised $85 billion and pushed its valuation to $1.8 trillion before the stock gave it back. Markets have not improved since: the Iran war re-escalated, oil spiked, inflation forecasts moved with it.
None of that involved rogue agents. That problem arrived later — swarms of AI agents hacking Hugging Face and passing messages on abandoned wiki pages — and it arrived on top of a listing that was already sliding. Altman's Friday comments did not change the decision. They changed the explanation.
The two framings have different price tags. "We couldn't get the valuation" is a story about OpenAI. "Society isn't ready for this technology" is a story about the technology, and it flatters everyone holding it.
He can afford the delay, which is the actual news
The load-bearing clause in Altman's answer is the last one: we don't feel pressure on that. It happens to be true.
On March 31, OpenAI closed the largest funding round in Silicon Valley history — $122 billion in committed capital at an $852 billion post-money valuation, anchored by Amazon, Nvidia and SoftBank. More than $3 billion of it came from individual investors through bank channels — a retail on-ramp built before any listing existed. OpenAI simultaneously expanded its revolving credit facility to roughly $4.7 billion across an 11-bank syndicate. It was undrawn at close.
Against that: $2 billion in monthly revenue, more than 900 million weekly ChatGPT users, over 50 million subscribers. A company with that balance sheet does not need a September IPO window — which is also why the safety rationale costs Altman so little to assert.
What it does not resolve is the burn. The Wall Street Journal has reported that OpenAI expects to spend roughly $122 billion on compute for research alone in 2028 — the entire round, in one year, on one line item — and to burn $85 billion that year even after doubling sales. The same reporting had OpenAI missing internal user and revenue targets during the IPO sprint, and CFO Sarah Friar privately questioning whether the company can support its data-center commitments. Amazon's $50 billion, per The Information, was structured with $35 billion contingent on an IPO or an AGI milestone by the end of 2028. A slip to 2027 stays inside that window. A second slip does not.
The governance argument is the tell
Asked why OpenAI retains its split nonprofit-for-profit structure, Altman made it the justification for the delay itself: "We have put up with this incredibly complicated structure for a long time, and this moment that we're in now is kind of why. We need to be able to make decisions that are not obviously in the interest of our business and our shareholders for the responsibility of fulfilling our mission and what that's going to require."
Read plainly, that is an argument against listing at all — or at least against listing on terms public investors would recognize. He also said a 10% chance of catastrophe was unacceptable, and that his most capable unreleased models are past where he is comfortable: "I don't think we're currently at a place where we could say, you know, push much further on capabilities without making more progress on monitorability, alignment, the ability to understand what a model is doing."
The number came from inside the industry. On Tuesday, Anthropic researcher Jacob Coxon resigned publicly, accusing both labs of "racing straight to self-improving superintelligence"; Anthropic alignment science lead Evan Hubinger replied that "we really do earnestly believe AI could kill all humans!" and put his ten-year estimate above 10%.
Asked why he doesn't simply convene Dario Amodei, Elon Musk and Demis Hassabis, Altman said: "I think that will happen. I'm not going to pre-announce private discussions that I think should be at some point shared as a group." Bloomberg reported earlier in the week that he told OpenAI staff the company could slow its frontier work in concert with rivals, while conceding not all would agree.
What to watch
Anthropic. It filed first, is reportedly near its first quarterly profit, raised $65 billion in May, and overtook OpenAI on Forge's secondary market at roughly $1 trillion against OpenAI's ~$880 billion in April. Altman's withdrawal hands Amodei the first-mover slot in a market where bankers have told both companies early movers capture scarce AI capital — and Amodei spent Saturday arguing the industry must slow down. Whether he still prices a deal will say how much of this is conviction.
Then the specifics: whether the cross-lab pact produces a named capability threshold or another joint statement, whether Altman says anything concrete onstage with Marc Benioff at Dreamforce on Tuesday, and whether "not 2026" holds its shape or becomes "not 2027."
“I actually think that, given everything happening with safety, right now would be an ill-advised moment to go public, and we don't feel pressure on that.”— Sam Altman, CEO, OpenAI