OpenAI Annualised Revenue Falls $20bn Short of Reported Figure

OpenAI’s annualised revenue is roughly $20bn below the figure that has been widely reported in recent weeks, according to financial documents shared with investors, a discrepancy that raises questions about how the market measures demand for artificial intelligence.

The company has told investors that its revenues were approaching $50bn on an annualised basis at the end of September, far short of the $70bn figure reported by major media outlets late last month based on information provided to investors. OpenAI declined to comment on the matter.

The annualised revenue metric is the most closely watched indicator of overall demand for AI services. It underpins the vast infrastructure spending by cloud providers and technology companies, and it has become a driver of public equity market valuations as investors price in the commercial potential of generative AI. A correction of this magnitude in the headline number for OpenAI, the sector’s most prominent private company, is therefore significant not just for its own shareholders but for the broader technology investment thesis.

The discrepancy appears to have arisen from attempts by OpenAI’s own investors to produce a direct comparison with the annualised revenue figure reported by rival Anthropic. The two companies calculate the metric in different ways. Anthropic includes revenue generated through sales via cloud partners such as Amazon Web Services and Google Cloud, while OpenAI does not. When OpenAI’s investors sought to gross up the company’s revenue to make it comparable with Anthropic’s methodology, the resulting estimates produced the higher figures that were later reported.

Efforts to gross up OpenAI’s annualised revenue in this manner had earlier led to reports that the company had reached $40bn in August. OpenAI subsequently told investors that its revenues had grown by more than 70 per cent, and combining that growth rate with the grossed-up August figure produced the $70bn number that circulated last month.

The episode highlights a broader problem in the AI industry: the absence of standardised revenue reporting metrics. As long as private companies can define their own methodologies, investors and the media will be forced to rely on estimates and comparisons that may not be directly comparable. The difference between $50bn and $70bn is not a rounding error. It is a 40 per cent gap in the perceived size of the market leader’s revenue base, and it has real consequences for how analysts model the sector’s growth trajectory, how venture capital firms value their stakes, and how public market investors assess the durability of the AI trade.

The underlying demand story remains intact. OpenAI’s revenue is still growing at a remarkable pace, and the company’s own figures show a business that has scaled rapidly in a short period. But the reporting gap serves as a reminder that in a market where sentiment moves on headline numbers, precision matters. Until the industry adopts consistent definitions for its key performance indicators, investors would be well advised to treat any single annualised revenue figure with caution, regardless of which company is issuing it.