OpenAI's revenue climbed roughly 50% quarter over quarter to $10 billion in the third quarter of 2026, but the company's operating loss widened to $10.9 billion, according to documents provided to investors and reported by The Wall Street Journal on Tuesday. Reuters, Bloomberg and Euronews carried the figures the same day, sealing one of the most closely watched private-company disclosures of the year.
The numbers describe a business scaling faster than almost any in technology history while spending even faster. OpenAI's quarterly revenue rose from roughly $6.7 billion in the second quarter to $10 billion in the third — a pace, by simple arithmetic on the reported totals, of about $109 million of revenue per day across July, August and September, up from roughly $74 million per day the quarter before. The operating loss, meanwhile, grew from $10.1 billion to $10.9 billion over the same period, reflecting the compute costs that underpin ChatGPT, the company's API business and its model training pipeline. For more context on this story, see our ongoing artificial intelligence updates.
The Margin Story: Better, Still Deeply Negative
There was progress in the ratio, if not in the absolute number. OpenAI's adjusted operating margin improved to -109% from -150% in the second quarter, a 41-point improvement that Forbes distilled into a stark framing: the company now loses about $1.09 for every dollar of revenue, down from roughly $1.50 a quarter earlier. The Journal characterized the results as mixed — revenue momentum is real, but so is the burn.
The word "adjusted" carries weight here. Because OpenAI is a private company, the figures come not from SEC filings but from documents shared with investors, which the Journal said it had reviewed; outside verification is limited, and the adjusted margin excludes some categories of expense, meaning the headline ratio and the actual cash burn are not the same thing. The disclosure nonetheless arrives at a delicate moment: OpenAI has ruled out an IPO for 2026, with CEO Sam Altman calling the idea ill-advised, even as its backers reprice their stakes around the huge private valuations of the past year.
Anthropic's Counterpoint
The most pointed comparison came from Anthropic's amended IPO filing earlier this month. The Claude maker, preparing a public listing that its bankers are reportedly sizing at $1.5 trillion or more, disclosed third-quarter revenue of $17 billion against an operating loss of $567 million — figures that put it ahead of OpenAI on the quarterly top line while losing a fraction as much money. In raw terms, Anthropic's quarterly operating loss is roughly what OpenAI reports losing in under two days of spending.
Barron's drew the connection explicitly in a piece titled "Microsoft Bet Big on OpenAI. Anthropic's Profit Lead Raises Questions" — a pointed headline for Microsoft, whose multibillion-dollar investment and cloud arrangement tie its AI fortunes to OpenAI's trajectory just as its rival heads toward a public listing. The comparison reframes a debate that has run since 2023: which of the two leading labs would convert the AI boom into a sustainable business first? On this quarter's numbers, the answer is no longer OpenAI by default.
How the Numbers Circulated
Because OpenAI does not report publicly, each quarterly disclosure moves through a characteristic channel: the Journal obtains or is shown the investor documents, wire services such as Reuters and Bloomberg summarize them within hours, and the figures then ricochet through markets commentary — this week alongside Bloomberg's framing that third-quarter revenue "soars nearly 50%" to $10 billion. The implied second-quarter figure of roughly $6.7 billion matches what coverage of OpenAI's prior quarter carried earlier this year, which lends the sequence some external consistency even without filings to check against.
What to Watch
Three threads will determine how this quarter is remembered. First, Anthropic's IPO pricing: a successful listing at a $1.5 trillion-class valuation would harden the comparison between the two labs' financial profiles. Second, OpenAI's fourth-quarter disclosure, where investors will look for whether the margin improvement from -150% to -109% continues or stalls as new infrastructure spending lands. Third, the response from Microsoft and other backers, whose returns on the largest AI investment ever made now depend on a company losing roughly a billion dollars — every ten days.
What the Numbers Mean
Two things can be true at once: OpenAI is growing extraordinarily fast, and its cost structure remains out of scale with its revenue. The gap between roughly 50% sequential revenue growth and a widening loss reflects an infrastructure bill measured in multi-gigawatt data-center commitments, custom silicon development and successive frontier-model training runs. Management has told investors that the margin trajectory, not any single quarter, is the metric to watch — and on that measure, at least, the company is moving in the right direction, from -150% to -109% in one quarter.
For the broader AI economy, the disclosure sets up an uncomfortable mirror. This week's headlines pair a private company losing $10.9 billion a quarter with a competitor preparing one of the largest IPOs ever on the strength of a $567 million quarterly loss. Public-market investors will get to price that divergence directly when Anthropic's shares begin trading — and OpenAI's investor documents will be read, once again, as the counterfactual. The next test of whether the burn is a growth strategy or a structural problem arrives with the fourth-quarter numbers.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
