Published: October 9, 2026
OpenAI's reported annualized revenue was approaching $50 billion at the end of September, according to financial documents cited by the Financial Times. That is roughly $20 billion below the $70 billion figure previously reported by several media outlets, putting renewed attention on how AI companies measure growth and how investors value the industry.
The discrepancy became a focus for markets on October 8, when technology shares fell amid concerns about AI spending, company valuations and the financial returns expected from the technology. Reuters reported the difference while noting it could not independently verify the Financial Times account.
But the headline difference needs context: the reported figures involve different ways of accounting for sales made through cloud partners. They do not, by themselves, show that OpenAI suddenly lost $20 billion in revenue or that demand for its products collapsed.
These are reported annualized figures, not audited full-year revenue totals. The comparison depends partly on accounting methodology.
Why the $20 billion difference matters
Revenue estimates are important in the AI industry because they help investors assess whether companies can turn rapidly growing use of AI products into sustainable businesses.
AI developers are committing large sums to computing capacity, data centers, specialized chips, research and product development. Investors therefore watch revenue growth closely when judging whether those investments are likely to generate sufficient returns.
When a prominent revenue estimate changes, investors may reassess not only the company involved but also the suppliers, cloud providers and infrastructure businesses connected to its expansion.
That helps explain why a reporting difference involving one AI company can influence sentiment across a much wider group of technology stocks.
Annualized revenue is not the same as annual revenue
One of the most important distinctions in this story is the meaning of annualized revenue, sometimes called an annual revenue run rate.
A run rate takes revenue generated over a shorter period and projects what the total would look like if that pace continued for a full year. It is useful for describing a fast-growing business, but it is not a record of the money actually earned over twelve completed months.
For example, if a company generated $5 billion in revenue during one three-month period, a simple annualized projection would be $20 billion. That projection assumes the pace continues; it does not prove the company will actually earn $20 billion over the following year.
Growth can accelerate, slow down or fluctuate. A run rate should therefore be read alongside the reporting period, the accounting definition and the company's actual financial results.
Three numbers that should not be confused
Actual annual revenue: Revenue recorded across a completed twelve-month period.
Annualized revenue: A projection based on a shorter period's revenue pace.
Bookings or commitments: Contracted or anticipated business that may not yet have been recognized as revenue.
How accounting differences created the comparison problem
The Financial Times report, as described by Reuters and Axios, attributed the discrepancy partly to differences between how OpenAI and rival Anthropic account for sales made through cloud partners.
Anthropic reportedly includes revenue from some sales through providers such as Amazon Web Services and Google Cloud in its figures, while OpenAI does not count certain partner sales in the same way.
Imagine a customer pays $100 for an AI service through a cloud platform. Depending on which company controls the customer relationship and delivers the service, the accounting treatment can determine whether the AI company records the full $100 as revenue and the partner's share as an expense, or records only its own share of the transaction.
Both approaches can be consistent with accounting rules when applied appropriately to the underlying arrangement. However, comparing the headline figures without explaining those differences can create a misleading impression of which company is larger or growing faster.
Axios reported that the previously cited $70 billion estimate was an adjusted figure intended to make OpenAI's results more directly comparable with Anthropic's methodology. The newer reported figure of nearly $50 billion uses a different basis.
Why technology stocks reacted
Investors are trying to determine how much future growth is already reflected in AI-related stock prices. When expectations are extremely high, even uncertainty over a company's financial metrics can prompt investors to reconsider the price they are willing to pay.
Reuters reported that U.S. technology shares came under pressure after the revenue report. Nvidia, a major supplier of AI processors, was among the companies whose shares fell. Other semiconductor and infrastructure names also weakened during the session.
The decline was not caused by the revenue report alone. Market coverage also pointed to rising oil prices, bond-market movements and concerns about inflation. Those factors can affect the broader stock market and the cost of financing large technology investments.
The result was a mix of company-specific questions and wider economic pressures rather than a single, definitive verdict on the future of AI.
What this means for AI infrastructure spending
AI models require substantial computing resources to train and operate. Providers must pay for processors, servers, networking, electricity, data-center capacity and the engineering needed to maintain their services.
Revenue expectations help determine how confidently companies and their financial backers commit to additional infrastructure. If future sales grow more slowly than expected, investors may ask whether spending plans should be delayed, financed differently or supported by stronger evidence of customer demand.
That does not mean AI infrastructure investment will automatically stop. It means financial scrutiny can become more selective, with investors paying closer attention to revenue quality, margins, computing costs and the time required to earn back capital.
Newspriint has previously covered the scale of spending on AI data centers and the demand for processors supporting the industry.
Related: The AI data-center investment boom
Related: AI chip demand and U.S. markets
OpenAI and Anthropic are difficult to compare directly
OpenAI and Anthropic compete for customers, enterprise contracts, developers and investment. But their revenue estimates cannot be compared responsibly unless the figures cover equivalent periods and use sufficiently similar definitions.
Investors also need to distinguish revenue from profitability. A company can generate rapidly rising sales while spending heavily on research, infrastructure and expansion. High revenue alone does not establish that a business is profitable or that its valuation is justified.
The same caution applies to reports of operating profit. Readers need to understand which costs are included, whether the figures are audited, and whether they cover the same period as the revenue being compared.
For the wider AI market, the key question is not simply which company reports the largest headline number. It is whether customer demand, pricing, computing costs and operating expenses can support sustainable growth over time.
What investors should watch next
The next useful signals will be clearer financial disclosures, consistent definitions of revenue and evidence that AI products are producing recurring customer spending.
- Revenue growth: Whether actual sales continue to rise over successive reporting periods.
- Accounting transparency: Whether companies explain partner sales and adjustments clearly.
- Profitability: Whether revenue growth can eventually cover computing, research and operating costs.
- Infrastructure commitments: Whether spending plans remain aligned with demonstrated demand.
- Market conditions: How interest rates, energy prices and financing costs affect expensive growth projects.
These indicators are more informative than treating one annualized estimate as a final measure of success or failure.
What is confirmed — and what remains uncertain?
Reuters reported on October 8 that the Financial Times had cited investor documents showing OpenAI's annualized revenue approaching $50 billion at the end of September, below the previously reported $70 billion figure. Reuters said it could not independently verify the report and that OpenAI had not immediately responded to its request for comment.
The reported accounting-method difference helps explain why the figures are not directly comparable. However, the report alone does not establish that OpenAI's sales have fallen, that AI demand has collapsed or that the company's long-term prospects have fundamentally changed.
The broader market reaction also reflected other economic concerns. Further company disclosures and financial reporting will be needed to assess how the revenue figures should influence expectations.
The bigger picture
The OpenAI revenue discrepancy is a reminder that the AI boom is entering a phase in which investors are scrutinizing not just product launches and funding announcements, but the definitions behind the numbers.
A reported $50 billion annualized run rate would still indicate a substantial business. The difference from an earlier $70 billion estimate matters, but understanding how the figures were calculated matters just as much.
For the industry, the lasting test will be whether strong AI adoption translates into transparent, sustainable revenue and returns that can justify the enormous cost of building the technology.
Sources and further reading
Reuters — reporting on OpenAI's annualized revenue estimate
Axios — explanation of OpenAI and Anthropic revenue accounting
Financial Times — report on OpenAI's revised annualized revenue
Hero image: “Trading Floor at the New York Stock Exchange” by Scott Beale, licensed under CC BY-SA 4.0. The image is illustrative and does not depict the OpenAI financial documents discussed in this article.
Post a Comment
Have something to add? Share your thoughts or let us know what you think about this story.