OpenAI Earns $20 Billion Less Than Projected, Investors Say

The annualized revenue estimate of US$ 70 billion that circulated over the past two months may have been significantly inflated, and OpenAI itself appears to have agreed. According to the Financial Times, the company reportedly informed its investors that its annualized revenue is "approaching $50 billion" — a US$ 20 billion drop from the US$ 70 billion figure that news outlets including Axios and Reuters had reported on September 29. The difference is not trivial: it equals the annual revenue of a mid-size technology company, and raises questions about the true quality of OpenAI's growth and what investors really know about the financial health of the record-funding foundation.

OpenAI has not issued any official statement confirming or denying the figures. What is known is that the company has maintained a policy of selective transparency with its investors, sharing internal estimates in update rounds that have never been audited. This creates a curious scenario: a company valued between US$ 852 billion and US$ 1.4 trillion, depending on which valuation round you discuss, operates with financial data that undergoes no independent verification. The Wall Street Journal had already reported in April 2026 that OpenAI's revenue and growth projections had fallen short of internal targets, raising questions about the company's ability to fund its expansion into data centers and compute capacity. Now, the FT suggests the gap between what was promised and what was delivered may be even larger than previously thought.

What "annualized revenue" really means

To understand the magnitude of this discrepancy, it is important to understand the concept of annualized revenue — also known as "run rate." The metric is straightforward in theory: you take the revenue from a recent month and multiply by twelve to project the full year. In practice, it is a highly volatile measure, especially for companies facing seasonality or abrupt pricing changes. Reuters had already warned when reporting the US$ 70 billion on September 29 that this figure was based on unaudited data from an anonymous source. Multiplying a single month by twelve assumes that month is representative of the entire year — an assumption that rarely holds, especially in the AI sector where enterprise contracts can generate irregular peaks in revenue recognition.

The difference between US$ 70 billion and US$ 50 billion is not just numerical; it is qualitative. If OpenAI's annualized revenue is truly closer to US$ 50 billion, the company is still an extraordinary money-making machine. But the 28% drop from the previous estimate suggests that growth, despite being impressive, may not meet the expectations of some investors — especially considering that OpenAI burned approximately US$ 44 billion in 2025 and projects US$ 75.6 billion in operating expenses for 2026. The annual burn rate is approaching the company's own revenue, raising a fundamental question: how dependent on continuous capital injections is OpenAI to sustain its operations?

The context: rapid but uncertain growth

The macroeconomic landscape for OpenAI in 2026 is one of accelerated growth, yes. According to Reuters, the company's enterprise sales more than doubled since July, and consumer revenue in Q3 exceeded that of all of 2025. The company reached US$ 25 billion in revenue in February 2026, jumped to US$ 40 billion in August, and by the annualized metric approached US$ 70 billion in September. This progression suggests consistent acceleration — but the "run rate" metric is not recognized revenue. GAAP revenue (revenue recognized under accrual accounting) can be significantly lower, as part of subscription and enterprise contract billing is recognized over the period of service, not at the moment of payment.

Furthermore, OpenAI operates in an increasingly competitive market. Anthropic, its main competitor, reported an annualized revenue of US$ 65 billion in July 2026 — a comparable figure, also based on unaudited estimates. The revenue scale of both companies is enormous, but the cost of training and inferring next-generation artificial intelligence models continues to rise. NVIDIA, the GPU chip supplier for both companies, reported record quarterly revenue in 2026, indicating that a significant portion of OpenAI's revenue is being paid back to hardware suppliers. OpenAI's gross margin — never publicly disclosed — may be well below what the market imagines.

Investors and the IPO risk

One of the most interesting aspects of this story is the timing. Until recently, OpenAI was in talks to raise an additional US$ 30 billion in a new round at a valuation estimated at US$ 1.4 trillion. If the actual revenue is US$ 20 billion lower than projected, that valuation may be overly optimistic. Sam Altman himself said in October 2026 that an IPO would not happen that year — suggesting the company is still navigating uncertain waters in terms of financial preparation for the public markets.

OpenAI's IPO had been expected for 2026, but successive delays have become a pattern. Each delay raises new questions: does the company need more capital? Are financial reports not ready for SEC standards? Or are current investors simply not willing to take the company public at such a high valuation in a volatile market? The fact that OpenAI is talking to new investors to raise an additional US$ 30 billion, even with revenue that may be lower than announced, suggests that cash flow is more critical than the general public imagines.

What to expect

What remains uncertain is how long this gap between projected revenue and actual revenue can persist without straining investor confidence. OpenAI has resources to operate for years at the current burn rate — but the math begins to change when the company goes public and must justify its numbers quarterly. Until then, analysts will continue to debate whether US$ 50 billion or US$ 70 billion is the right number. The answer, probably, will be revealed when OpenAI finally enters the stock exchange — and investors can read the audited reports that, until now, do not exist.

The big question is: when a company as large as OpenAI operates with relative financial opacity, what is the true cost of that secrecy for the investment market? And to what extent can the culture of "move fast and break things," taken to the extreme of a trillion-dollar scale, create systemic risks that we have not yet identified?

Sources: Financial Times, Bloomberg, ROI Central

✓ Independent sources cross-checked and verified before publishing