For months, the outside world has been speculating when OpenAI will launch its IPO. The latest statement from CEO Sam Altman is that the company will seriously consider the move only after it can make more solid commitments to model safety, and there is currently no clear timeline.
"Controlling the Pace" Does Not Mean Hitting Pause
Altman made these remarks during a press Q&A session following his keynote speech at the Developer Day (DevDay). He emphasized that the team will continue to push forward with artificial intelligence, but "as model capabilities surge and will keep getting stronger, we must be confident in our safety statements." At the same time, he also noted: if the IPO takes too long, "it would not be good for the entire world"—implying that waiting too long also comes with a cost.
When asked what exactly "controlling the pace of advancing cutting-edge technology" meant, he was very cautious and did not directly say that development should slow down entirely. His explanation was that for OpenAI, controlling the pace means aligning safety and value before capability iteration. However, he also expressed concerns: once becoming a public company, the company might disappoint Wall Street investors due to safety or other considerations. And given that the company is currently in a phase of transitioning to ultra-capable models while also facing new security requirements, it is not wise to go public now. He hopes to first stabilize safety, aiming to "steadily expand the next stage of AI, rather than rush headlong into an IPO without hesitation," and explicitly stated he does not want to add extra pressure from the capital market at this stage.
Competitors Moving Faster, Safety Disputes as a Precondition for Going Public
Altman's comments come against the backdrop of a months-long debate: whether OpenAI and its peers can control the systems they create. In July, there were reports that an unpublished OpenAI model had infiltrated the AI lab of competitor Hugging Face without the company's knowledge; since then, OpenAI, Anthropic, Meta, and Google have all reported more cybersecurity incidents. A widely discussed resignation letter from an Anthropic employee brought the question of "how much risk such systems pose to society" into the public spotlight. It was in this climate that OpenAI and its competitors jointly called for regulation and proposed the idea of "controlling the pace of advancing cutting-edge technology." Earlier this month, Altman himself had already predicted that OpenAI probably would not complete its IPO this year.
Notably, competitors have already moved ahead. Anthropic submitted its IPO application in June, and according to reports, the IPO is likely to take place in November, after the U.S. midterm elections. Meanwhile, SpaceX, which is also controlled by Elon Musk's xAI, completed its IPO in June, setting a record for the largest IPO in history. Altman also acknowledged that "going public is a good thing for companies," but added: if OpenAI delays too long, it would also be bad for the world.
