For most of 2026 the story about OpenAI’s money was that it was growing, just not as fast as the company next door. That story may have just changed. According to a Bloomberg report citing people familiar with the numbers, OpenAI’s revenue for the third quarter crossed $10 billion, the first time the company has ever booked that much in a single quarter.
If the figure holds up, it means OpenAI grew roughly 50% in three months, after a first half of the year that insiders and analysts described as sluggish. It still trails Anthropic on revenue, and it is still losing a lot of money. But the gap in growth speed, which had been the most uncomfortable chart for OpenAI’s investors, looks a lot smaller this quarter.
The short version
- OpenAI’s Q3 revenue reportedly topped $10 billion, up from $6.7 billion in Q2
- That is growth of about 49% quarter over quarter, compared with roughly 18% from Q1 to Q2
- Adjusted operating margin improved from minus 150% to minus 109%
- Annualized revenue run rate is now about $50 billion, per Bloomberg
- Anthropic reported $17 billion in Q3 revenue and an $81 billion run rate at the end of October
The numbers, quarter by quarter
OpenAI is a private company, so everything here comes from reporting rather than audited filings. Bloomberg’s sources put second quarter sales at $6.7 billion, and earlier reports pegged the first quarter at about $5.7 billion. Line those up and the shape of the year becomes clear: a slow climb, then a jump.
The jump from 18% growth to roughly 49% growth is the real headline. Plenty of software companies would be thrilled with 18% in a year. For a company that has raised money at valuations built on hypergrowth, though, 18% in a quarter while a rival was compounding faster was a problem. A near 50% quarter buys OpenAI a much better narrative heading into 2027.
The Bloomberg figure also lines up with what OpenAI president Greg Brockman said publicly earlier this month. Speaking at a Wall Street Journal event on November 4, he said annualized revenue grew about 70% during the third quarter and that the enterprise business roughly doubled. He did not give dollar amounts, but a run rate that rises 70% inside a quarter is exactly what would produce a quarterly total jumping by half.
What turned it around
Bloomberg’s sources and TechCrunch’s reporting point to a mix of four things rather than one big launch:
| Driver | What changed | Why it matters for revenue |
|---|---|---|
| Newer models | Stronger demand for the latest GPT releases | Better models pull developers back to the API |
| Pricing | Tighter pricing plus targeted discounts | Wins volume deals without giving away the whole catalog |
| Enterprise push | More products aimed at companies, not consumers | Business contracts are larger and stickier than $20 subscriptions |
| AI agents | Tools that complete multi step tasks | Agents burn far more tokens per job than a chat reply |
The pricing piece deserves a closer look. OpenAI spent much of the fall cutting the cost of its models, most visibly with GPT-6.1 Terra at $1 per million tokens. Cheaper tokens sound like they should shrink revenue. In practice, lower prices tend to unlock workloads that were too expensive to run before, especially agent workflows that chain dozens of calls together. If usage grows faster than the price falls, revenue goes up.
The enterprise push matters even more. OpenAI says it has more than a billion users, but most of them pay nothing or pay for the cheapest tiers. Anthropic built its lead by selling to companies and developers, where contracts are measured in millions. OpenAI’s Q3 suggests it is finally getting real traction in that same market.
The losses are shrinking as a percentage, not in dollars
Bloomberg reported that OpenAI’s adjusted operating margin went from minus 150% in the second quarter to minus 109% in the third. That is a genuine improvement. It is also easy to misread.
A margin of minus 150% means that for every dollar OpenAI brought in, it spent about $2.50 on operations. At minus 109%, it spent about $2.09. Apply those ratios to the reported revenue and something interesting falls out.
In rough terms, OpenAI’s adjusted operating loss was about $10 billion in Q2 and about $11 billion in Q3. Revenue grew by more than $3 billion, and the loss stayed in the same neighborhood. That is the classic pattern of a company growing into its costs: spending on compute and talent is still enormous, but each new dollar of sales costs less to produce than the one before it. Whether that curve ever crosses zero is the question that will define OpenAI’s eventual IPO.
Keep in mind: “Adjusted” operating margin typically strips out items such as stock based compensation. OpenAI’s true loss on a standard accounting basis is likely larger than these figures suggest.
$10 billion a quarter, $50 billion a year?
Bloomberg also reported that OpenAI’s annualized revenue run rate has reached about $50 billion. That can look inconsistent with a $10 billion quarter, since four quarters of $10 billion only adds up to $40 billion. It is not.
A run rate takes the most recent month and multiplies it by twelve. A $50 billion run rate implies OpenAI is now bringing in a bit more than $4 billion a month. A quarter that totals just over $10 billion, but ends at that pace, means revenue was climbing steadily through July, August and September. In other words, the $10 billion quarter is a snapshot of an average, and the run rate shows where the line was pointing when the quarter closed.
Run rate math, simplified
--------------------------------------
Annualized run rate $50.0B
Implied latest month $50.0B / 12 = ~$4.2B
Q3 total (reported) ~$10B+
Average month in Q3 ~$3.4B
Exit month vs. average ~+23%How OpenAI stacks up against Anthropic
None of this erases Anthropic’s lead. The Claude maker reported $17 billion in third quarter revenue, and on November 10 it said its run rate had jumped by $10 billion in a single month to $81 billion at the end of October. Anthropic is preparing for an IPO that could value it at around $1.4 trillion, and its S-1 filing laid out both enormous revenue and enormous losses.
| Metric | OpenAI | Anthropic |
|---|---|---|
| Q3 2026 revenue | $10B+ (reported) | $17B |
| Q2 2026 revenue | $6.7B (reported) | About $11.5B (reported) |
| Latest run rate | ~$50B | $81B (end of October) |
| Core strength | Consumer reach, 1B+ users | Enterprise and developer spend |
| How partner sales are counted | Only OpenAI’s share of some partner sales | Includes sales made by cloud partners |
| Public listing | Not yet filed | IPO expected soon |
That last accounting row matters more than it looks. When a business buys Claude through Amazon Bedrock or Google Cloud, Anthropic counts that sale in its revenue. When a company buys OpenAI models through a partner, OpenAI only reports its own cut in some cases. So the headline comparison flatters Anthropic somewhat. The real gap is narrower than $17 billion versus $10 billion, though no outsider can say exactly by how much.
Growth rates tell a more even story. Anthropic’s Q2 to Q3 jump, from roughly $11.5 billion to $17 billion, works out to about 48%. OpenAI’s reported jump was about 49%. For the first time in a while, the two companies appear to be growing at roughly the same speed, with Anthropic simply starting from a bigger base.
What to watch next
Three things will tell us whether this quarter was a turning point or a one off. First, whether the enterprise doubling Brockman described keeps going as companies finalize 2027 budgets. Second, whether cheap models like Terra keep pulling in more usage than they give up in price. And third, whether the margin line keeps climbing toward zero, or stalls once OpenAI’s next wave of data center spending hits the books.
Competition is not getting easier either. Google, Meta, Amazon and Microsoft all ship their own frontier models, and capable open models from Chinese labs keep driving down what customers expect to pay. A $10 billion quarter puts OpenAI back in the conversation. Staying there will take a few more of them.
About this article: GeekBlog covers U.S. technology news, AI, phones, smartwatches and gaming. Every story is written and checked under our Editorial Policy. Spotted a mistake or have a story tip? Contact our editors.
