OpenAI’s revenue topped $10 billion in a single quarter for the first time, but its losses widened to $10.9 billion, underlining how expensive the race to build artificial intelligence has become.
The figures, for the third quarter of 2026, were disclosed to investors at a meeting and reported by The Information on 10 November and by Bloomberg. Revenue was up about 50 per cent from $6.7 billion in the second quarter, and nearly triple the level of a year earlier.
The company’s annualised revenue run rate reached about $50 billion by the end of September, according to chief financial officer Sarah Friar, a measure of how fast the business is growing. The adjusted operating loss widened to $10.9 billion in the quarter, from $10.1 billion in the second quarter, although the adjusted operating margin improved to minus 109 per cent from minus 150 per cent.
The enterprise business doubled quarter on quarter, and Codex, the company’s coding product, became its largest source of revenue, overtaking the consumer business, according to co-founder Greg Brockman, speaking on 5 November.
The contrast with its main rival is striking. Anthropic, which launched its IPO roadshow on the same day OpenAI’s figures were disclosed, reported third-quarter revenue of about $17 billion with a much smaller operating loss of $567 million, according to its IPO prospectus.
OpenAI’s cloud infrastructure spending commitments through 2030 now total about $750 billion, and its cumulative negative free cash flow for 2026 to 2030 is forecast at $278 billion, figures that show the scale of the bet the company is making on computing power.
The disclosure landed on the same day Anthropic launched its IPO roadshow, sharpening the debate about the economics of artificial intelligence. Investors are asking whether the huge spending on data centres, chips and energy will ever turn into the profits that the valuations of the leading AI companies imply.
OpenAI, which is used by about 1.2 billion people a week through ChatGPT, has been expanding into new areas, including business tools, a travel agent and commerce, as it tries to turn its enormous audience into revenue beyond subscriptions.
For now, the company is growing faster than almost any in history, but it is also losing money faster than almost any in history. The question for investors, and for the industry, is whether the growth can outrun the losses before the spending has to slow.
The figures will be watched closely by the rest of the AI industry, where every leading laboratory is facing the same question: how to pay for the computing power that the next generation of models will need, and how long the money will keep flowing.
