OpenAI has held early talks with large investors about a capital raise that would value it at roughly $1.2 trillion ahead of a public offering, the Financial Times reported, picked up by Reuters on 15 September. The talks are at an initial stage and the figure may not hold. No amount, terms, investor list or closing date has been set.
The detail that makes it interesting is who started the conversation. Investors approached OpenAI, not the other way around.
The timing is the other thing. On Saturday 12 September, Sam Altman said the company would not pursue an initial public offering in 2026, calling it an ill-advised moment to go public and citing frontier AI safety. That same weekend he backed Dario Amodei’s call for the industry to slow its rate of capability growth. Three days later the $1.2 trillion figure was in the press.
For scale: OpenAI’s March round involved $122 billion of committed capital at a post-money valuation of about $852 billion. $1.2 trillion would be roughly 41 percent higher, in under six months. At the time of that March round the company disclosed $2 billion of monthly revenue, with enterprise customers accounting for more than forty percent of it.
The obvious reading is that this is hypocrisy. It is not, and understanding why is more useful than enjoying the contradiction.
Pacing capability development and raising capital are different activities that happen to sit in the same company. In fact the slowdown argument, taken seriously, makes the raise more necessary rather than less. Embedded independent evaluators, safety cases, interpretability work and monitoring all cost money and none of them ship a product. A company that genuinely intends to go slower on capability and heavier on assurance needs a larger balance sheet to do it, not a smaller one.
There is a cafe version and it is the least glamorous conversation in hospitality. An operator decides to stop opening new sites for eighteen months and fix the ones they have, because service has slipped and three managers have left. That decision is correct and it requires more cash, not less, because refits, retraining and better pay all land before any of the benefit does. Slowing down is a capital-intensive choice that looks like a modest one.
What the reported valuation actually measures is something else again, and it is the part worth watching. The IPO has been pushed out and the private number went up 41 percent. That is value accruing in private markets rather than public ones, which means the people who capture it are the people already in the room.
Anthropic, per Reuters, could begin marketing an IPO as early as mid-October, with a listing shortly before the November midterms. If that happens, the two companies will have taken opposite routes at the same moment, and the comparison will be instructive: one testing whether public markets will pay these numbers, the other deciding it does not need to ask.
Investors initiated the talks. That is the most telling line in the reporting, because it means the demand is not being manufactured by a company that needs money.
Pacing is about how fast the model gets better. Nobody said anything about the valuation.
Source: Quartz, “OpenAI is exploring a new funding round that would value it at $1.2 trillion”
