OpenAI Annualised Revenue Runs $20bn Below Reported Figure

OpenAI’s annualised revenue is roughly $20bn lower than the figure that has circulated in recent weeks, according to financial documents shared with investors, a discrepancy that carries significant implications for how the market values the artificial intelligence boom. The company recently told investors that its revenues were approaching $50bn on an annualised basis at the end of September, well short of the $70bn reported by media outlets late last month based on information supplied to investors. The gap matters because this metric, alongside the equivalent figure for rival Anthropic, is the most closely watched indicator of underlying demand for AI services, and it underpins the vast infrastructure spending plans and the valuation of public equity markets that have come to depend on sustained growth in the sector.

The source of the discrepancy appears to be methodological rather than a sudden slowdown in business. According to a person with knowledge of the matter, the confusion arose from attempts by OpenAI’s own investors to produce a direct comparison with Anthropic’s annualised revenues. The two companies calculate the figure differently. Anthropic includes revenue generated through sales via cloud partners such as AWS and Google Cloud, while OpenAI does not. Efforts to “gross up” OpenAI’s annualised revenue to make it comparable on that basis led to earlier reports that the group had hit $40bn in August. The company has since told investors its revenues have grown more than 70 per cent, which produced the $70bn figure when applied to that grossed-up base.

The distinction is not trivial. Annualised revenue, sometimes called annual recurring revenue in subscription businesses, is a forward-looking estimate that annualises the most recent run rate of sales. It is a standard tool for technology investors assessing high-growth companies that have not yet stabilised their quarterly reporting. For OpenAI and Anthropic, it has become the de facto benchmark for measuring whether the enormous capital expenditure on AI data centres, chips, and energy is translating into actual customer demand. A $20bn swing in that figure, driven purely by accounting convention rather than business performance, is exactly the kind of noise that can distort investment decisions if it is not properly understood.

The wider implication is that the AI market’s headline numbers may be less precise than they appear. The revenue figures that drive equity valuations, corporate capital plans, and even national policy discussions about AI competitiveness are, in practice, estimates compiled under differing standards. That is not unusual in fast-growing industries, but the scale of the discrepancy here, roughly 40 per cent of the reported figure, is substantial enough to warrant caution among investors who treat these numbers as hard data. The fact that the confusion originated with OpenAI’s own investors attempting to harmonise the two companies’ reporting suggests that even sophisticated market participants are struggling to compare the two businesses on a like-for-like basis.

The takeaway for professional readers is straightforward. The underlying demand for AI products may well be growing rapidly, and OpenAI’s own stated figure of approaching $50bn annualised revenue at the end of September is still a very large number by any historical standard. But the gap between that figure and the widely reported $70bn is a reminder that the most important metric in the AI investment landscape is currently defined inconsistently across the two leading companies. Until the industry settles on a common standard for measuring revenue, investors should treat any single headline figure with a degree of scepticism and focus on the underlying methodology, not just the number itself. OpenAI declined to comment on the record.