- Lead. The Financial Times reported on October 8 that OpenAI’s annualized revenue stood at approximately $50 billion at the end of September — about $20 billion below the figure investors had been working with — triggering a sharp selloff in AI stocks.
- Fact. The gap is not a shortfall in OpenAI’s actual revenue but a consequence of competing accounting methods: investors applied a gross revenue calculation — counting what enterprise cloud customers spend through AWS and Google Cloud — while OpenAI reports on a net basis, excluding its cloud partners’ share.
- Stake. The episode illustrates a structural problem in how private AI companies are valued before listing: because OpenAI has no audited accounts and no disclosure obligations, investors were applying a methodology borrowed from a competitor to extrapolate numbers OpenAI itself never confirmed.
The Financial Times published a report on October 8 revealing that OpenAI had told investors its annualized revenue run rate reached approximately $50 billion at the end of September 2026 — roughly $20 billion below the figure that had been circulating in financial markets since earlier in the quarter. The article, sourced to a person familiar with the matter, prompted an immediate and sharp reaction: Nvidia fell nearly 3 percent, AMD dropped around 4 percent, Broadcom declined 4.35 percent, and Microsoft, Amazon, Alphabet and Meta each fell more than 1 percent. One estimate cited by Meyka put the total one-day AI sector loss at $2.7 trillion in market capitalisation. The Nasdaq composite declined by more than 1.25 percent, and Nvidia alone was estimated to have shed approximately $170 billion in market value.
Where the $20 Billion Came From
The discrepancy did not represent a revenue shortfall at OpenAI. Rather, as the FT explained, it reflected a mismatch between accounting methods applied to two different companies. Anthropic, OpenAI’s closest rival, counts gross revenue — meaning it includes the full amount enterprise customers spend on its AI models through platforms such as Amazon Web Services and Google Cloud, before those platforms take their share. Investors had apparently been extrapolating OpenAI’s trajectory using the same gross method, producing the $68–70 billion estimate. OpenAI, however, reports net revenue: it excludes the portion that flows to its cloud distribution partners. The $50 billion figure is OpenAI’s own net number, and the two are not directly comparable. Tech Times described investors as having built the higher estimate “using the wrong method.”
This is consistent with the earlier reporting that OpenAI’s enterprise run rate was nearing $70 billion as of mid-2026 — a figure that, read using the gross methodology, would be close to the circulated number. The FT’s correction reanchored market expectations to the net figure, producing what Euronews called “a revenue scare that rattled the AI trade.”
The Pre-IPO Disclosure Problem
OpenAI’s situation highlights a specific vulnerability in how private AI companies are priced. Because OpenAI has not filed a prospectus, has no Securities and Exchange Commission disclosure obligations, and has not completed a public listing — its IPO has been flagged for late October or early November — investors rely on leak-and-extrapolate cycles rather than audited statements. When the FT’s sourced figure contradicted the extrapolated one, the market treated the revision as a demand signal rather than an accounting clarification. The actual demand for OpenAI’s products does not appear to have changed in any way between October 7 and October 8.
By October 10, AI and technology stocks had partially recovered, with the S&P 500 closing near its second-highest level on record and the Nasdaq advancing for a fourth consecutive week. The rebound reflected investors re-engaging with the original bull case — that AI infrastructure spending remains structurally elevated — after concluding the FT’s disclosure was about accounting, not demand.
A Signal for Disclosure Standards
The episode has renewed calls from regulators and analysts for clearer disclosure standards for large private AI companies before they access public markets. The FTC, which opened a consumer-safety investigation into OpenAI, Anthropic and METR earlier this year over autonomous AI agent risks, has not addressed revenue disclosure directly. But the two-day $2.7 trillion swing in listed AI companies based on unaudited, leaked figures from a private company raises questions about market information standards that existing securities law was not written to handle.