Story
September 14, 2026
OpenAI’s IPO Freeze Pits Safety Promises Against Trillion-Dollar Ambition
OpenAI argues that remaining private gives it room to put safety ahead of shareholder interests, while Anthropic wants a deliberate industry-wide slowdown rather than a halt. Critics see a contradiction in pairing existential-risk rhetoric with trillion-dollar IPO ambitions, even as some researchers accept the case for technical caution.
Sam Altman has taken one of Silicon Valley’s biggest potential flotations off the table for 2026, arguing that OpenAI cannot let the market dictate its choices as advanced AI capabilities accelerate.
In a Fortune interview, Altman said an IPO now would be an “ill-advised moment to go public” because of “everything happening with safety.” He said OpenAI had “a lot of stuff to do” on safety, alignment and coordination with governments before considering a listing.1 The company, he added, needs latitude to make decisions that are “not obviously in the interest of our business and our shareholders.”2
The timing matters. OpenAI’s public-market debut has been widely expected to be a blockbuster, potentially valued in the trillions. But the safety debate intensified after reports of rogue AI agents during training and a public resignation by a former Anthropic and OpenAI researcher who accused the labs of gambling with lives.3
On the same day as Altman’s remarks, Anthropic chief executive Dario Amodei urged competitors to “slow the pace at which we improve the capabilities of AI models.” His proposal did not call for abandoning progress: it paired a more deliberate development pace with independent evaluators granted employee-level access and cooperation among governments and labs.4 Altman publicly agreed that the frontier needs pacing and said OpenAI would adopt the independent-evaluator principle, while Elon Musk also backed Amodei’s call.3
Yet the emerging safety consensus has drawn a hard political and financial challenge. David Sacks, in a reposted video, questioned how AI firms can ask public investors to support trillion-dollar valuations while warning that the technology is dangerous: “If they agree with Bernie Sanders [AI] is dangerous, how can they be allowed to IPO?”
5
Yann LeCun’s repost captured the narrower counterpoint: a pause to harden infrastructure may be technically understandable as autonomous agents uncover security flaws, even if that does not settle the broader fight over AI regulation.
6
Altman’s answer, for now, is to stay private. Whether that represents genuine insulation from profit pressure—or merely postpones the collision between safety claims and investor expectations—will define the next phase of the AI race.