News
FT: OpenAI's Annualized Revenue Revised Down $20B to Near $50B
The Financial Times reported on October 8 that OpenAI told investors its end-of-September annualized revenue was near $50 billion — about $20B below the $70B figure previously reported in media. Differences in accounting methodology are the main cause.

The Financial Times reported on October 8 that OpenAI recently told investors its end-of-September annualized revenue was near $50 billion — about $20B less than the $70 billion figure media had previously reported. Reuters and TechCrunch both followed up on the FT story.
$50 Billion, Not $70 Billion
The $70B number first appeared in media reports last week, claiming OpenAI’s annualized revenue was near that level — enough to go toe-to-toe with Anthropic’s revenue run rate. But the FT obtained documents OpenAI gave directly to investors, and the two numbers differ by $20 billion. OpenAI did not respond to Reuters’ request for comment.
Why the Two Numbers Differ So Much
It comes down to methodology.
The earlier $70B version was calculated by OpenAI’s own investors. Per FT-cited sources, those investors wanted a number they could benchmark against Anthropic.
And the two companies count revenue differently to begin with. Anthropic includes revenue resold through cloud partners (like AWS and Google Cloud); OpenAI doesn’t count that line. So even on the same basis, a direct comparison stays unfair.
Annualized revenue isn’t a strict accounting concept either. The usual method takes one month’s revenue times 12 to project the year. Startups use it routinely, fast growers especially. But when the number also mixes in methodology differences, distortion comes easy.
Markets Reacted Fast
After the news, stocks across OpenAI’s supply chain fell. Oracle dropped 4.9% that day, AMD 3.3%, Broadcom — which makes custom chips with OpenAI — 3.2%, TSMC 3.3% (WSJ market data). Microsoft fell 1.2%, Dell 0.8%.
It makes sense. OpenAI’s revenue number directly bears on whether it can sustain the current hundred-billion-dollar-scale investment wave. In the AI bubble debate, compute leasing, data center construction, and chip orders all hang on the assumption that “top AI companies’ revenue keeps growing fast.” Shrink the number and the whole chain’s valuation logic gets rechecked.
Background: Fundraising and IPO at Critical Junctures
OpenAI’s March round raised $122 billion in one go — the largest financing in AI history.
But its books aren’t easy. Leaked 2025 financials from earlier this year showed OpenAI’s 2025 revenue around $13 billion, with spending clearly higher. Meanwhile both giants are preparing to list: Anthropic reportedly IPOs around November, OpenAI’s listing plan slipped from this year to early 2027.
Once public, financials are public. The “flexible” annualized-revenue framing stops working then, and revenue growth sustainability goes under a magnifying glass. This revision is an early shot for the market: OpenAI’s revenue isn’t as high as outsiders thought.
What to Watch Next
This won’t shake OpenAI’s product position — ChatGPT and its APIs remain developers’ most-used AI services. But it changes an expectation: the AI industry’s most valuable company, whose revenue growth is already too slow for investors to “buy with eyes closed.”
Two things worth tracking: Anthropic’s IPO pricing and market reception, and how OpenAI negotiates valuation in its next round. Plus whether the two companies unify revenue methodology — on the eve of going public, that gap only breeds more suspicion.