Invest1 publisher2 min readPublished
Altman pushes OpenAI's earliest listing into 2027 on safety grounds
OpenAI's chief executive told Fortune that an offering now is "ill-timed" given the safety concerns and will wait until 2027. The condition he attached to that date is an industry slowdown pact his peers would have to sign too.
The Investor · Invest desk
What happened
- Altman called an OpenAI public offering "ill-timed" given the safety concerns around the technology, in an exclusive interview with Fortune.
- He said he would have no problem standing up to his investors if he needed to pause or stop AI development altogether, and that no gamble with humanity is OK.
- The week's alarm followed an Anthropic employee's post on X warning that all of humanity could be destroyed by the technology and that major frontier labs were "gambling with our lives."
Compiled by The InvestorSomething wrong?How this is made
Why it matters
- constraint The condition attached to the timing is an industry-wide slowdown, so the date turns on what companies other than OpenAI agree to do.
- decision Anyone whose plan pays out on an OpenAI listing now funds the position at least into January 2027, and 2027 is as precise as the timing gets.
- exposure Investors in a company heading for a public offering have been told in advance, in print, that its chief executive may pause or stop development.
- precedent A frontier lab citing its own safety exposure as grounds for delay invites the same question at every rival with listing ambitions.
The wording sets an earliest date and leaves the real one open. Altman said the listing will not take place until 2027 [2]. That rules out the rest of this year and leaves twelve months inside which any month qualifies [8].
The reason he gave matters more than the year. An offering now is "ill-timed" given the safety concerns, he said [1], and the fix he pointed to is a pact he is making with industry peers to slow down while safety and alignment solutions can catch up to model capabilities [5]. OpenAI controls its own filing calendar; the pact needs the peers to sign it too.
The safety exposure does have public numbers attached, and they come from other principals. The probability of doom, or p(doom), has been discussed openly in the AI community for years [15]: Elon Musk's has hovered between 10% and 20% [9], and Anthropic's Dario Amodei's has been slightly higher at 10% to 25% [10], five points above Musk's ceiling on the same floor [11]. What set the week off was an Anthropic employee's warning on X that all of humanity could be destroyed by the technology and that major frontier labs were "gambling with our lives" [6].
Fortune did not publish a valuation, an underwriter or any financial figure for OpenAI in its account of the hour-long sit-down [12]. Anyone repricing a position off Friday's interview [3] has the date and the condition to work with. There is one further term in there for holders: Altman told Fortune he would have no problem standing up to his investors if he needed to pause or stop AI development altogether [4].
The same sentence supports more than one reading. Safety may be the binding constraint, with the date following from it. Or a company not yet ready to file needs a reason for the wait, and safety costs nothing to give. There is also this: a floor set by a judgement moves in both directions, because the judgement can be revised as fast as it was issued.
In my view the part of this that binds is the exclusion of 2026, and I would treat 2027 as a floor. A registration filing before January would show the safety framing was cover for a company that was not ready to list anyway.
What to watch
- Whether any peer lab is named in, or signs, the slowdown pact Altman described to Fortune.
- Whether OpenAI or Altman attaches a quarter to the 2027 date, turning a floor into a schedule.
- Any employee tender or secondary sale run before then. Such a sale would price what the wait costs insiders.