OpenAI is reported to be in early talks over a share sale that would value it at about $1.5tn. Its last round valued it near $730bn. That is roughly a doubling, with no obvious event in between to explain it.
At the same time, executives have made clear a public listing is not close.
The two numbers are not the same kind of number
A private valuation is the price at which a small number of chosen buyers agreed to take a small slice of stock. It is negotiated, it is often structured with protections that do not apply to ordinary shares, and it prices a few percent of the company.
A public valuation is what anyone can pay or receive each morning, on every share, with audited quarterly accounts behind it.
The two are quoted in the same units, which is the source of endless confusion. Multiplying a negotiated slice by the total share count produces a headline, not a market price.
Why defer
There are respectable reasons. Quarterly reporting imposes a rhythm at odds with research spending that pays off over years, and a company whose costs are enormous and whose revenue curve is steep would spend every earnings call explaining the gap.
There is also the plainer reason. A private company discloses what it chooses. A listed one publishes revenue, cost of revenue, and the actual economics of serving each customer — the numbers the entire debate about this industry turns on, and which nobody outside has seen.
What the multiple implies
No published revenue figure makes $1.5tn arithmetically ordinary. That is not an accusation; it is the definition of an expectations-driven valuation. The buyers are not pricing this year's business. They are pricing a belief about what the business becomes.
Sometimes that belief is right. Amazon looked absurd on earnings for a decade. But the same structure — capital committed today against returns assumed later — is what is currently producing record default rates in private credit, and the Federal Reserve has just raised the price of capital for the first time since 2023.
Expectations-driven valuations are the most sensitive thing there is to a discount rate. When money costs more, profits promised further out are worth less today, and the companies priced entirely on distant profits move the most.
Who sells into a round like this
The reported vehicle is an employee share sale, which is worth separating from a fundraising round. The company is not necessarily raising money here; existing holders are being given a way to sell.
That serves a real purpose. Staff paid substantially in equity have been unable to convert any of it for years, and a tender offer relieves the pressure that would otherwise push them toward an early listing or the door.
It also quietly transfers risk. The buyers are institutions taking a position at the highest price the company has ever carried, from people who know the business best and are choosing this moment to take some money off the table. That is not a scandal — diversification is rational for anyone with a single concentrated holding — but it is information, and it is the kind that a public market would price immediately.
What to watch
Whether the round closes at the reported figure, and on what terms. Structure matters more than headline: liquidation preferences and ratchets can make a high nominal valuation cheap for the buyer and expensive for everyone holding common stock underneath it.
And whether the deferral holds. Companies that can list on their own terms usually do so when conditions are good. Postponing in a strong market is a decision worth revisiting if the market turns.
Reports of early talks over a share sale at a valuation near $1.5tn, the approximately $730bn valuation of the prior round, and statements indicating that a listing is not imminent are as reported by Bloomberg, Reuters and the Financial Times in mid-September 2026. OpenAI has not confirmed the figures. The analysis is our own.
Topics technologyopenaivaluationventure capitalipo
