OpenAI Nears $70B Revenue Run Rate as Enterprise Sales Double

OpenAI is now running at a pace that would make it one of the fastest-scaling companies in the history of the technology industry. According to financial figures reported by Axios on Tuesday, September 29, and relayed by Reuters, the ChatGPT maker's annualized revenue run rate is nearing $70 billion — a number it has reached with breathtaking speed, powered above all by a surge in enterprise sales.

Axios cited sources familiar with OpenAI's financials, and the core figures are stark: enterprise revenue more than doubled since July, the annualized run rate has grown more than 70% since the start of the third quarter, and the consumer business added more revenue in Q3 alone than it added in all of 2025. OpenAI did not respond to a Reuters request for comment on the numbers, so treat them as credible but unconfirmed company-sourced leaks — not audited results. With that caveat stated plainly, the scale of what is being claimed deserves a careful look.

Photo: OpenAI CEO Sam Altman speaking onstage at a TED event in April 2025. Credit: Steve Jurvetson, Wikimedia Commons (CC BY 2.0).

What Axios reported — and where the numbers come from

The headline figures, as reported by Reuters from Axios's Tuesday story, are threefold. First, OpenAI's annualized revenue run rate is approaching $70 billion. Second, business-to-business revenue has grown more than 100% since July — more than doubling in under three months. Third, consumer-side revenue added in the third quarter of 2026 exceeds what the consumer business added across all of 2025.

Axios described the $70 billion figure as annual recurring revenue, but Bloomberg Law's summary of the same report frames it as "annualized revenue based on its current performance." The difference matters, and we will get to it below. What is not in dispute is the direction: this is a dramatic acceleration from the more-than-$40-billion annualized pace that Bloomberg News reported in August, citing people familiar with the company's performance.

Put the trajectory in sequence and the slope is striking. OpenAI closed 2025 at a run rate above $20 billion, a figure confirmed by CFO Sarah Friar. The pace held near $25 billion through February and the spring, then broke upward: president Greg Brockman told staff internally that the run rate grew more than 20% month over month in July, Bloomberg reported a $40-billion-plus pace in mid-August, and now Axios is reporting near $70 billion at the end of September. That is roughly a tripling of the annualized pace in nine months.

A necessary word on what "run rate" actually means

Before anyone treats $70 billion as OpenAI's 2026 revenue, a brief exercise in precision — because the AI industry's favorite metric is also its most misunderstood.

A revenue run rate takes revenue from a recent period — typically the latest month — and extrapolates it across a full year, assuming the current pace holds. It is a measure of speed, not of distance traveled. Annual recurring revenue (ARR), properly defined, counts contracted, repeatable subscription revenue and excludes one-time charges. Actual booked revenue is what the company collected and recognized under accounting rules over a real period. These are three different numbers, and Axios's report does not provide a methodology explaining how much of its $70 billion figure is contractual recurring revenue versus usage-based API consumption or other variable income, as RuntimeWire pointed out in its analysis of the report.

That is the honest framing, and it applies equally to OpenAI's rivals. Every run-rate figure in this article — OpenAI's and Anthropic's alike — is an annualized snapshot of a recent pace, relayed by reporters from sources rather than disclosed in audited filings. Run rates can and do fall if growth stalls; they also understate where the business ends up if growth continues. As an analysis of the August numbers put it, a $70-billion run rate should be read as a current pace, not a forecast or a completed year's sales.

With the definitions on the table: even discounted for uncertainty, the pace of change here is extraordinary. For context on the comparison, remember that Anthropic's widely cited $65 billion figure from late July was corroborated by Bloomberg, CNBC, and Reuters — but was likewise a run rate, not booked revenue. The two companies are racing on a metric that both define loosely and that neither reports through official filings.

Enterprise is now the engine

The most important line in the Axios report is not the $70 billion figure. It is the enterprise number: B2B revenue more than doubled since July. OpenAI's story for years was a consumer phenomenon that monetized unevenly. The new numbers describe something else — a corporate-sales machine.

This was telegraphed. When OpenAI announced its March 31 financing round — $122 billion in committed capital at an $852 billion post-money valuation — the company said enterprise products already contributed more than 40% of revenue and were on track to reach parity with consumer revenue by the end of 2026. Axios's September figures suggest corporate sales are advancing ahead of that stated target, though OpenAI has not released an updated revenue mix to confirm the shift.

The drivers, per Bloomberg's August reporting, include the growth of OpenAI's AI coding software — the category where corporate IT budgets are currently loosening fastest — alongside subscription gains and an advertising business that Forge Global pegged at a $1 billion annualized run rate by the end of August. The backdrop, as Techstrong.ai's analysis put it, is a broader shift across corporate America: enterprises moving generative AI from pilot programs into fully budgeted, paid production deployments. Shorter procurement cycles, larger deal sizes — that is the pattern a source briefed on OpenAI's sales pipeline described to Axios.

The timing was not accidental either. The Axios report landed on the same day OpenAI held its annual DevDay developer conference at Fort Mason in San Francisco on Tuesday, September 29 — the company's flagship event for the developers building on its platform, headlined by CEO Sam Altman's keynote. A momentum number and a developer event on the same day is a classic piece of narrative choreography: give the ecosystem a growth story at the exact moment you are asking it to build on your stack.

The other side of the ledger: $278 billion in cash burn

There is a number in this story that is almost as dramatic as $70 billion, and it points the other way. To support its expansion, OpenAI anticipates roughly $278 billion in total cash burn between 2026 and 2030, according to figures reported by Reuters and summarized by Techstrong.ai. Data centers, specialized chips, and computing infrastructure are the price of the revenue curve, and the price is staggering: CFO Sarah Friar has indicated to investors that the company will not turn cash-flow positive before 2028 at the earliest, per a tech-insider.org analysis of investor communications.

This is the defining feature of the frontier-AI market: revenue doubling on a curve no software company has matched, while spending climbs just as fast. As one analysis of the summer numbers observed, $40 billion in run-rate revenue against hundreds of billions in infrastructure commitments is not a balanced ledger — it is a bet that revenue keeps compounding long enough to catch the spending. At $70 billion, the bet looks more defensible than it did at $40 billion. But it is still a bet, and it is funded by private capital rather than by operating cash flow.

That is what makes the IPO path so closely watched. Both OpenAI and Anthropic have filed confidential paperwork to go public, according to PYMNTS's reporting. Anthropic is expected to list first, possibly as soon as this fall, at a valuation that could approach $2 trillion — a number its own investors have framed against what data provider analysis called the fastest growth of any company in history. OpenAI, meanwhile, is in a different position: Altman told Fortune in a September 12 interview that the company will not go public in 2026, citing the industry's unresolved AI safety and alignment questions — while, separately, OpenAI has held early talks with investors about a pre-IPO funding round at roughly a $1.2 trillion valuation. The $70 billion run rate is the strongest card in that negotiating deck.

Photo: Sam Altman speaking onstage at TechCrunch Disrupt San Francisco 2019. Credit: TechCrunch, Wikimedia Commons (CC BY 2.0).

OpenAI vs. Anthropic: the race on the headline metric

The Axios leak arrived one day after months of Anthropic holding the run-rate crown. Anthropic's annualized revenue run rate crossed $65 billion by the end of July, up from $47 billion in May and roughly $9 billion at the close of 2025 — a more than sevenfold jump in under a year, corroborated by Bloomberg, CNBC, and Reuters. The company's 2025 booked revenue was $4.6 billion, up roughly twelvefold year over year.

If the two figures are measured on comparable bases — a real if, given different dates and undisclosed methodologies — the Axios number puts OpenAI back ahead on the headline metric. One analysis of the report framed it as "a rapid reversal in the enterprise race with Anthropic, which had recently reported a higher run rate." And the race is still accelerating on both sides: a more recent report citing the New York Times said Anthropic's run rate topped $100 billion in mid-September, which would put the September Axios figure and the newest Anthropic figure in a genuinely close contest.

What the headline duel does not tell you is profitability. Anthropic's leaked financials show operating losses exceeding $8 billion in 2025 and an estimated $518 billion in future cloud and computing commitments, per Techstrong.ai's summary. OpenAI's own cost structure is similarly unforgiving — that $278 billion burn figure looms over the entire industry. Both companies are spending far beyond current revenue on the thesis that the growth curve bends upward faster than the cost curve. Public investors will eventually demand the gross margins, not just the run rates. That is what the coming IPO filings are for.

What it means for markets, partners, and the industry

The market read the Axios numbers as validation of the AI-infrastructure trade. Oracle shares rose between 5% and 7% following the report, according to market coverage relayed by MoneyCheck and CoinCentral — a move that reflects how tightly Oracle's cloud and data-center fortunes are now coupled to OpenAI's compute spending. Microsoft, OpenAI's closest partner and largest backer, sits in the same orbit: every enterprise contract OpenAI signs deepens the value of the Azure-hosted ecosystem the two companies have built together. Treat the stock-move figures as relayed by secondary outlets, not as independently verified market data — but the direction of the signal is consistent with the quarter's broader AI trade.

Stepping back, the honest read of Tuesday's news has three layers. First, the numbers themselves: near-$70-billion annualized pace, enterprise doubled since July, consumer adding more in one quarter than in all of last year — all from sources familiar with OpenAI's financials, all relayed by Axios and confirmed by Reuters, Bloomberg Law, and others, none of it audited. Second, the trend: from a $20-billion-plus run rate at the end of 2025 to $40 billion in August to near $70 billion in September is the steepest revenue ramp any major software or cloud business has recorded over that span — faster, month for month, than the legendary cloud ramps of the 2010s. Third, the cost: a business burning toward $278 billion in cash through 2030 is not a normal company, and a $1.2 trillion pre-IPO conversation is not a normal valuation process.

The coming months will test whether the pace holds. Anthropic's IPO — expected before OpenAI's, possibly this fall at a valuation around $2 trillion — will give public markets their first audited look at a frontier lab's full economics. OpenAI's confidential filings, DevDay's developer momentum, and whatever the company discloses about its actual revenue mix will fill in the rest. Until then, $70 billion is a pace, not a result — and the most remarkable pace the industry has ever seen.

Sources: Axios reporting via Reuters, Bloomberg Law, and Techstrong.ai; Bloomberg News (August run-rate reporting); RuntimeWire analysis; MoneyCheck and CoinCentral market coverage; Techstrong.ai and PYMNTS on Anthropic financials and IPO filings; Fortune interview via secondary reporting for Altman's no-IPO-in-2026 comments; devday.openai.com and The Mac Observer for DevDay 2026 details. All run-rate figures are annualized estimates from company-sourced leaks, not audited filings.