What's trending, explained.
What's trending, explained.
Documents shared with OpenAI's investors put the company's annualized revenue at about $50 billion as of the end of September 2026, according to the Financial Times, which first reported the figure on October 8. That was roughly $20 billion lower than the $68-70 billion number that had been circulating in markets just weeks earlier.
The shortfall was not a sudden collapse in sales. OpenAI's overall run rate actually grew 77 percent quarter over quarter in the third quarter, and its enterprise business grew 107 percent, according to reporting on the disclosure. The company still expects to reach at least $70 billion in annualized revenue by the end of 2026.
The gap came down to accounting. Anthropic, OpenAI's main rival, books the full value of sales made through its cloud distribution partners — Amazon Web Services and Google Cloud — as revenue. OpenAI has historically recorded only its own net share of partner sales. To make the two companies comparable, investors had "grossed up" OpenAI's numbers to match Anthropic's method, producing the $70 billion figure. The $50 billion number was OpenAI's own measure.
Markets reacted sharply anyway. The Nasdaq closed 1.4 percent lower on the Thursday the story broke. Nvidia fell about 3 percent, Oracle nearly 6 percent, and CoreWeave almost 8 percent, as investors questioned whether AI revenues can justify the industry's enormous capital spending.
This is less a story about OpenAI's business — which is still growing fast — than about what investors thought they knew. A $20 billion swing that turned out to be a bookkeeping difference shows how much of the AI boom's narrative rests on numbers that are hard to compare.
That comparison matters because the two labs are now competing for the same public-market investors. Anthropic confidentially filed for an IPO earlier this year and is reportedly targeting a listing before Thanksgiving, with talk of a valuation around $2 trillion. OpenAI is not expected to go public until at least 2027, but it is reportedly discussing a further $30 billion-plus financing round and carries a private valuation of roughly $852 billion. As these companies approach public markets, standardized revenue reporting is becoming a serious issue — investors need to know whether the figures they are shown are gross or net.
The episode also fed the broader anxiety hanging over the AI trade: enormous infrastructure spending — one reported internal estimate put OpenAI's cumulative negative free cash flow at $278 billion from 2026 through 2030 — must eventually be justified by revenue. Any sign that the revenue story is softer than advertised moves the whole sector.
The next data point is December, when OpenAI's year-end run rate will show whether it really can bridge the gap to $70 billion. Watch too for whether OpenAI and Anthropic adopt a common revenue disclosure basis — pressure is building for one, especially with an Anthropic IPO approaching.
The deeper test is the one investors keep circling: whether commercial AI revenue can grow fast enough to cover the capital expenditure behind it. This week's sell-off was a reminder that when the numbers disappoint — even over an accounting convention — the market is quick to punish.