OpenAI made $10B and lost $10.9B in one quarter

OpenAI made $10 billion in a quarter for the first time — and lost $10.9 billion in the same quarter. The loss is bigger than the revenue, and that is the story, not the milestone.

What was reported

OpenAI shared the numbers with investors on a call on Tuesday, November 10. The figures were first reported by The Information, then by the Wall Street Journal and Bloomberg. OpenAI declined to comment, and none of this comes from a filing — it is an investor-call disclosure relayed by anonymous sources, so treat every number below as reported, not confirmed.

The reported Q3 numbers: revenue of about $10 billion, up from $6.7 billion in Q2; an operating loss of $10.9 billion, up from $10.1 billion; an adjusted operating margin of -109%, improved from -150%; an annual revenue run rate of about $50 billion. CFO Sarah Friar called it a "blockbuster quarter." Enterprise revenue doubled quarter over quarter and is now the majority of revenue, according to president Greg Brockman.

The math, recomputed

I ran the reported numbers myself, because the headline version — "revenue up nearly 50%" — hides the more interesting arithmetic:

Three things fall out of that table.

The margin improved because revenue grew faster than the loss, not because the loss shrank. The loss went up by $800 million in absolute terms. At a -109% margin, every $1 of revenue still costs about $2.09 to produce. The improvement is real — 41 percentage points in one quarter is a genuine inflection in the slope — but "blockbuster" is a strange word for a quarter where you lose $900 million more than you take in.

The $50 billion run rate is not $50 billion of revenue. Annualizing Q3 gives $40 billion. A $50 billion run rate implies the exit rate — roughly, where September was running — sits 25% above the Q3 average. That is consistent with the story Bloomberg's sources told: the rebound came late in the quarter, driven by new models, aggressive price cuts, and agents. But a run rate is a snapshot of the last month extrapolated, not a forecast. The gap between $40 billion annualized and $50 billion run rate is a claim about momentum, and momentum is exactly what an investor call is designed to sell.

The year-over-year number is the loudest one in the room. At roughly 2.6x the year-earlier quarter, Q3 2025 implied revenue was about $3.85 billion. OpenAI roughly tripled in a year. That is the number that justifies a $1.25 trillion valuation round — the company is finishing a funding round at that valuation, with $11 billion already in the first close, ahead of a planned IPO next year.

The Anthropic comparison, and why it is not clean

The obvious comparison is Anthropic: about $17 billion in Q3 revenue, with growth slowing to roughly 47% quarter over quarter. On the headline, Anthropic is 1.7x OpenAI's quarterly revenue, and over the first nine months the gap is $33.2 billion to about $22.4 billion (OpenAI's $5.7B + $6.7B + $10B), or 1.48x.

But the two companies count revenue differently, and both sides say so. Anthropic counts sales made by its cloud partners; OpenAI reports only its share of some partner sales. The $17B vs $10B gap is real in direction and overstated in size, and nobody outside the two companies can say by how much. The cleaner read is the trend, not the level: both companies grew about 47–49% quarter over quarter, and both are still deeply unprofitable. Anthropic's Q3 operating loss was $567 million on $17 billion — a -3.3% margin. OpenAI's was $10.9 billion on $10 billion. Same growth rate, completely different cost structure.

The number that actually matters

Enterprise revenue doubling to a majority of the total is the one figure in this release that changes the competitive story. A week ago Brockman said enterprise had gone from about 30% of revenue to over half. Anthropic's entire lead was built on enterprise — it got there first, with Claude Code and the agent stack that enterprises actually buy. OpenAI catching up on enterprise share is the first sign that the gap is closing from OpenAI's side, not just widening from Anthropic's.

That is also why the IPO timing matters. Anthropic priced its IPO range at $70–$78 this week, targeting a $1.4 trillion valuation and listing November 19. OpenAI is raising at $1.25 trillion and aiming for 2027. Both are going public — or preparing to — while burning billions a quarter. The market is being asked to price growth first and profitability eventually, and this quarter is the test case for that pitch: revenue up 49%, losses up 8%, margin improving 41 points. If you believe the slope, the $10 billion quarter is the beginning of the story. If you believe the level, OpenAI just had its best quarter ever and still lost more money than it made.

Sources: Bloomberg Law, "OpenAI Third-Quarter Revenue Soars Nearly 50% to $10 Billion" (November 11, 2026); TechCrunch, "OpenAI's quarterly revenue reportedly crosses $10B for the first time" (November 11, 2026); The Wall Street Journal and The Information, via Bloomberg and Dealroom (November 10, 2026); Anthropic Q3 figures from its S-1/A (November 2, 2026) and my November 3 analysis of the same filing. All OpenAI figures are reported from the November 10 investor call, not from a public filing; OpenAI declined to comment. The table and ratios above are my own recomputation from the reported figures.