The two leading AI labs are both behind the IPO timetable. One said so on the record.

One of the nine conditions this map tracks is simple to state: do the frontier AI labs actually come to the public markets? A listing matters for more than the headline. It is the moment a lab's revenue, costs and commitments stop being leaks and become disclosure, and it is a new source of money for a build-out that needs a great deal of it.

For most of this year, the working assumption was that both leading labs would list in 2026. Both filed confidentially in June, according to reports. In the past two weeks, that assumption has slipped at both.

Start with the firmer of the two. On 12 September, Fortune published an interview in which OpenAI's chief executive, Sam Altman, was asked about timing. His answer: "I would say not 2026." He called the present "an ill-advised moment to go public" and said the company does not feel pressure to do it. He tied the decision to work on safety and alignment. He did not give a new date. In August, OpenAI's finance chief had told staff it would be public "in 2027 or sooner," so 2027 is the obvious reading, but it is not what he said.

That is the CEO on the record, which is why it moves the map. OpenAI turns yellow, for the first time on something other than an estimate.

Anthropic is different in both substance and sourcing. On 18 September the Wall Street Journal reported, from people familiar with the matter, that its listing has moved from October to November. The reported reason is that advisers want to show third-quarter results first. The Information reported the same. Anthropic has not confirmed a date, and there is still no public filing.

A one-month move with a stated reason is not much of a delay, and we have filed it that way: as a minor signal that records the change and colors nothing. Anthropic stays gray.

Why record it at all? Because the pattern is the point. The market was pricing two lab listings in 2026. It now has, on the record, at most one. And the reason both companies give is the same one: the current argument over AI safety. It is unusual for a company to say that the public mood about its own product is a reason not to list.

There is a second thread worth watching. On 16 September, Fortune, citing the Financial Times, reported that OpenAI is in talks on a new private round valuing it at about 1.2 trillion dollars. If that closes, private money is standing in for public money. That keeps the build-out funded, but it also keeps the numbers private for longer. For a map that tries to separate disclosure from reporting, that is a real cost.

And a third, on the Anthropic side. On 25 September The Information reported that Anthropic's seven co-founders are asking shareholders to approve special shares giving them 50.1 percent of the vote on most matters. Founder voting control is common in technology listings. It is not a timetable change, so it is not a signal here. If it is approved, or appears in a public filing, it becomes disclosure and we will add it.

Now the case against reading too much into this. Neither company has withdrawn anything. Anthropic is still aiming for this year. OpenAI raising privately at a higher valuation is hardly a sign that capital has dried up. A postponed listing is not a failed one, and a company choosing its moment is ordinary.

It is also worth being clear about what this does not touch. An IPO date tells you nothing about how much revenue a lab is earning, how fast it is growing, or whether demand for AI is holding up. Those are separate conditions on the map, and this changes none of them. Prediction markets have moved sharply on both listings, but odds are prices, and we do not color anything on a price.

What would change the picture is observable. A public S-1 from Anthropic, which would put audited revenue and costs on the record for the first time. A further move beyond November, or beyond 2026. OpenAI naming a date, or closing the private round. Shareholders approving, or rejecting, the Anthropic voting structure.

On the map: OpenAI turns yellow. Anthropic stays gray. The IPO condition moves from not enough evidence to yellow. Nothing turns red.