OpenAI's annualized revenue is approaching $50 billion, according to financial documents shared with investors and reported by the Financial Times on October 8, 2026. The figure is large in itself, yet it sits some $20 billion below the roughly $70 billion reported by several outlets in late September — a gap big enough to pressure US tech stocks the same day. OpenAI declined to comment on the reports.
Most of the $20B gap is arithmetic, not lost sales
Annualized revenue is not audited full-year income; it is a run rate that multiplies one month's revenue by twelve, so it moves whenever the chosen month changes. The bigger factor is method. As the FT relayed from a person familiar with the matter, the earlier $70 billion was produced by investors trying to compare OpenAI directly with Anthropic. Anthropic counts sales through cloud partners such as AWS and Google Cloud on a gross basis, while OpenAI's figure excludes that pass-through and is calculated net. Same company, same month, different ruler — a roughly $20 billion difference. The revision is about measurement, not about business vanishing in a month.
Why markets sold first
After the report landed mid-session on October 8, selling spread across the AI supply chain: the Nasdaq Composite closed down 1.25%, its worst day since mid-August; Nvidia fell 2.9%, Intel 5.3%, and Oracle — holder of a huge cloud deal with OpenAI — dropped 5.5%. The investor logic is simple: the story behind enormous compute spending needs revenue growth to support it. When the sector leader's run rate is re-marked, the chip, cloud, and power contracts tied to it get re-priced too, even when the issue is accounting method rather than orders.
$50B is still an extraordinary pace
The other side matters too. OpenAI reportedly raised $122 billion in March, and its 2025 revenue was in the low tens of billions; a run rate near $50 billion is still rare growth for a company of this size. The real questions are not about any single month's annualized number. They are whether revenue growth can outrun compute and staffing costs, and whether, once Anthropic and OpenAI face public-market disclosure with audited figures, investors finally get a comparable yardstick. Until then, every annualized figure should be read together with its method — taken alone, it can easily overstate or understate the company.
