The calls began going out on the evening of September 15. Investors, not OpenAI, made the first move, according to The Wall Street Journal, approaching the company with offers that implied a valuation above $1.2 trillion. Within a day the number had climbed higher in a rival account: The New York Times reported that OpenAI believes it is worth $1.5 trillion or more.
The gap between the two figures — roughly $300 billion — is the space in which the negotiation will be fought. It is also a measure of how fast the company’s perceived value has moved. OpenAI last raised money in March, at an $852 billion valuation, pulling in $122 billion in a single round. The new numbers imply a gain of 41 to 76 percent in six months.
The case for the higher figure rests on two products the company has spent the year pushing. The first is a new generation of models, the frontier systems that anchor ChatGPT. The second is Codex, the programming agent that has become one of the company’s fastest-growing lines. Customers, the company argues, are paying for both in numbers that justify the price.
The timing is no accident. OpenAI has been telling investors for months that it is still in the steep part of its growth curve, and the March round gave it the balance sheet to keep spending while it waited for the market to catch up. Now the company is asking the market to catch up all at once.
People familiar with the discussions said the talks are still early. No terms have been set, and the valuation could settle anywhere in the range the two papers described — or outside it, if the demand the company’s bankers are testing does not materialize. A number that looks inevitable in a headline is, in a term sheet, still a guess.
The IPO question hangs over the whole exercise. OpenAI had been expected to list in 2026, but that timeline has now slid to 2027, according to people briefed on the plan. A private round at a $1.2 trillion valuation does two things at once: it funds the company’s enormous compute bill, and it sets a reference price that any future offering will have to clear.
Analysts said the delay is a statement in itself. Companies that expect their valuation to keep rising have every reason to wait, and OpenAI’s leadership has made clear it believes the company is still under-accounted by public markets. The risk, which some investors have begun to voice privately, is that the private price runs ahead of what public buyers will accept.
The compute spending is what ties the whole thing together. OpenAI’s valuation is a bet on artificial intelligence becoming the dominant layer of the economy, and every dollar of that bet flows through data centers. The March round bought the company time; the next round buys it scale, and the price of scale is set by the chip market, not by any product roadmap.
The investor base has shifted as the numbers have grown. Early rounds were led by venture firms and strategic partners; a round of this size draws sovereign funds, state pension systems, and the largest asset managers. Those buyers are less patient about exits, and a delayed IPO sits awkwardly with their need for liquidity.
OpenAI has not commented on the reports, and the company rarely does during fundraising. The silence is part of the process. But the existence of two competing valuations, leaked on the same evening, is itself a signal — the company wants the higher number in circulation, and the buyers want the lower one as an anchor.
The backers from March have a direct stake in the outcome. The funds that bought in at $852 billion will see their paper gains multiply if the new round prices higher, and they have little reason to argue against the company’s own arithmetic. That dynamic can push a valuation up in ways that have little to do with what a neutral buyer would pay.
The $122 billion March round already ranks among the largest private financings ever, and the new talks suggest it was a waypoint rather than a destination. If the round closes at the top of the range, OpenAI would become the most valuable private company on record by a wide margin, and the IPO question would get harder, not easier.
For now, the numbers are doing the talking. A $1.2 trillion floor and a $1.5 trillion ceiling are both enormous, and the difference between them is larger than the market value of all but a handful of public companies. The negotiation will come down to a single question, one that has followed OpenAI since the beginning: how much is a lead worth when the race is this expensive?
