Key Notes
- OpenAI’s revenue run rate approached $50 billion at the end of September, below the roughly $70 billion figure previously reported.
- Differences in partner-sales accounting explain the comparison gap, while OpenAI reported strong third-quarter enterprise growth.
- AI infrastructure shares fell as investors weighed the revenue supporting large computing commitments and ambitious valuations.
OpenAI’s annualized revenue approached $50 billion at the end of September, the Financial Times reported on October 8, about $20 billion below the figure circulated late last month.
The gap puts the ChatGPT developer’s financial disclosures under scrutiny at a sensitive moment for AI valuations. However, subsequent reporting points to differences in the treatment of partner sales, rather than a demonstrated $20 billion decline in revenue or a missed full-year sales promise.
That distinction matters beyond OpenAI’s private shareholders. Publicly traded chipmakers and cloud providers fell as investors reassessed the revenue supporting the industry’s enormous computing investments.
Why the Earlier Figure Was Higher
The approximately $70 billion figure, which we reported in September based on Axios and Reuters, needs to be understood alongside the newly disclosed accounting basis.
Axios explained that investors had adjusted OpenAI’s revenue upward to make it more comparable with Anthropic’s treatment of sales through cloud partners. Both companies’ approaches comply with generally accepted accounting principles, according to the report, but they recognize partner-related revenue differently.
That means a comparison requires a consistent definition of whose revenue is being counted. Customer spending through a distribution partner and revenue recorded by the AI developer can be different measures, even when they relate to the same underlying demand.
The important change is therefore the basis of the headline number. The reports do not establish that customers canceled $20 billion of business, or that OpenAI had earned that amount and subsequently lost it. They also do not provide a complete reconciliation of every component in the two calculations.
Revenue Run Rate Is Not Full-Year Sales
CNBC separately confirmed the roughly $50 billion figure with a person familiar with OpenAI’s investor presentation. Its report compared that amount with an earlier $68 billion estimate that included partners’ gross revenue.
The same presentation indicated that OpenAI’s total revenue run rate grew 77% during the third quarter, while its enterprise run rate increased 107%, CNBC reported. Those growth figures concern the pace of business over the quarter, rather than revenue earned across an entire year.
An annualized run rate extends a recent sales pace over twelve months. It can help describe the scale of a rapidly expanding business, but it is neither the revenue accumulated during the previous twelve months nor a guarantee of sales over the next twelve.
The distinction is particularly important when a company is growing quickly. Sales generated earlier in the year may have been substantially lower than the latest pace, so multiplying recent activity into an annual figure does not reconstruct the income statement.
OpenAI’s own historical disclosures show how quickly its commercial scale has changed. Chief Financial Officer Sarah Friar said in January that annual recurring revenue exceeded $20 billion in 2025, compared with $6 billion in 2024. Those earlier figures provide background, rather than a substitute for a consistent comparison of the latest investor metrics.
AI Stocks React to the Revenue Clarification
Nvidia (NASDAQ: NVDA), Oracle (NYSE: ORCL) and CoreWeave (NASDAQ: CRWV) all declined during Thursday’s session. CNBC’s intraday report put their losses at approximately 3%, 6% and 8%, respectively; those were observations during trading, not closing returns.
The reaction illustrates how a private AI company’s disclosures can affect listed businesses supplying its infrastructure. Investors have to assess whether the demand supporting new chips and data centers will produce enough revenue and cash to sustain further expansion.
A lower revenue measure does not translate mechanically into an equivalent reduction in infrastructure spending. The latest reports do not announce canceled computing contracts or a cut to OpenAI’s capacity plans. They do, however, make the definition of customer revenue more important when evaluating those commitments.
The question is also about profitability. Revenue alone does not reveal the cost of serving customers, the cash required to train models or the timing of payments to infrastructure suppliers. Rapid enterprise growth and substantial financing needs can coexist.
Funding and Valuation Face a More Demanding Test
OpenAI announced on March 31 that it had closed a financing with $122 billion in committed capital at an $852 billion post-money valuation. At that time, the company said it was generating about $2 billion in revenue per month.
The March announcement establishes the terms of that completed financing. It does not establish OpenAI’s current cash balance or mean the company has already secured every dollar it may need for future expansion.
As we previously wrote, Bloomberg reported in September that OpenAI was seeking at least $30 billion in another round at a proposed valuation of approximately $1.4 trillion before the investment. That remains a reported fundraising target, rather than a completed transaction price.
The latest revenue clarification does not automatically reset either valuation. It changes the sales measure investors should use when assessing the price, alongside growth, operating costs and the capital needed to deliver future services.
A fuller financial disclosure would need to reconcile direct and partner-related sales, show revenue recognized over a defined period and explain the resulting margins and cash flows. Until then, the central distinction is between a reported revenue pace near $50 billion and the larger comparison figure previously circulated to investors.
