Breaking4h ago28 sources63 reactions

OpenAI annual recurring revenue approaches $70 billion in Q3 2026

24 DoomStory + reactionsStrong revenue milestone, framed as rapid growth

By Splitfeed EditorialUpdated Sep 29, 2026, 18:21 UTC

Split of the day · 28 sources

Boom: OpenAI's annualized revenue run rate neared $70 billion by late September 2026
Boom: Revenue grew roughly 70 percent since the start of Q3 2026
Boom: Enterprise sales more than doubled since July, driven partly by the Codex assistant
Boom: ChatGPT reached 1.2 billion weekly users, OpenAI stated
Neutral: Anthropic is reported to have a similar revenue run rate, narrowing OpenAI's lead
Boom: Oracle shares spiked 8 percent on the news amid ongoing IPO speculation

The story in full

OpenAI's annual recurring revenue run rate has reached nearly $70 billion as of late September 2026, according to Axios citing sources. The figure represents roughly 70 percent growth since the start of Q3 2026, with enterprise sales more than doubling since July. ChatGPT now reaches 1.2 billion users per week, OpenAI said. Oracle shares rose 8 percent on the news amid IPO speculation.

Enterprise sales, the Codex coding assistant, and price reductions are credited with accelerating the growth. Anthropic is reported to be closing the gap with a similar revenue run rate. The $70 billion figure is an annualized run rate, not booked annual revenue, a distinction that affects how the number is interpreted.

Analysis

434 words

By Splitfeed EditorialLast updated Sep 29, 2026, 18:21 UTC

On September 29, 2026, Axios reported, citing sources, that OpenAI's annualized revenue run rate had neared $70 billion by late September, representing roughly 70 percent growth since the start of Q3 2026. Enterprise sales more than doubled between July and late September, with the Codex coding assistant cited as a key driver. OpenAI separately stated that ChatGPT now reaches 1.2 billion weekly users. Oracle shares climbed 8 percent on the news amid ongoing IPO speculation. Anthropic is separately reported to be closing the gap with a comparable revenue run rate.

The figures carry important caveats. The $70 billion number is an annualized run rate, a projection based on recent monthly revenue, rather than actual booked annual revenue. That distinction matters because a run rate assumes current momentum holds, and several observers in the middle ground have pointed to structural fragilities. Lora Kolodny noted that Anthropic disclosed nearly a quarter of its revenue came from just two customers last year, with many large clients not locked into long-term contracts. The rapid rise of coding agents as a revenue category is also notable because it represents a shift from consumer subscriptions toward enterprise software spending, a model with different retention and margin dynamics.

The pro-AI camp reads the numbers as historic validation. Paul Graf pointed out a year-over-year revenue increase he described as 1,100 percent, calling it unprecedented for a company at this scale. Simon Willison argued that companies have begun spending millions on coding agent tools that barely existed a year ago, citing this as evidence of genuine product-market fit he said he anticipated in May. Critics are less impressed. Smack54 argued the business remains unsustainable because infrastructure costs exceed what customers are willing to pay, while other anti-AI voices used the moment to redirect attention toward safety concerns, with A51FR3D and Quartz both amplifying warnings from researchers about existential risk and the need for oversight. The middle ground is skeptical of the headline number itself. Blind Squirrel Macro questioned whether a revenue-multiple approach could realistically carry either company to a public listing, and Vignesh NV flagged that open-source models are narrowing the performance gap at far lower cost, which could erode the pricing power underlying the current run rate.

The argument will sharpen around any IPO filing or prospectus that would require OpenAI to disclose audited revenue, customer concentration, and actual contract structures rather than annualized run rates. How Anthropic's comparable figures hold up under similar scrutiny, and whether enterprise spending on coding agents sustains its pace into Q4 2026, will be the clearest test of whether the current trajectory is durable or front-loaded.

Pro-AI4

What Pro-AI voices are sayingRevenue growth is unprecedented and enterprise adoption is accelerating, with coding agents driving a product-market-fit moment that barely existed a year ago.

Quote 1 of 4

companies started blowing millions of dollars on coding agent tools that hardly existed a year ago

Simon Willisonvia Bluesky

Anti-AI50

What Anti-AI voices are sayingCritics argue the underlying economics are unsustainable and that OpenAI cannot be trusted to govern its own models safely, with some calling the valuation a bubble. A notable minority frames the doom rhetoric itself as a deliberate PR strategy to attract investment.

Quote 1 of 13

AI companies would need $725 billion in annual revenue for them to have both 10% operating margins and hyperscalers to have a 10% ROIC. We are not even close. None of the economics make sense for either side of the

Ed Zitronvia Bluesky

Middle Ground9

What Middle Ground voices are sayingObservers note that coding agents, not AGI, appear to be the actual business model, but concentration risk, open-source competition, and unproven valuations make the long-term picture uncertain.

Quote 1 of 9

Anthropic said nearly a quarter of its revenue came from two customers last year, and as part of its risk factors, warned that many of its largest clients were not locked into long-term contracts and could cut or stop spending.

Lora Kolodnyvia Bluesky

Add your take

0 reader votes

More Pro-AI reactions (3)

“OPENAI B2B REVENUE GROWS MORE THAN 100% IN Q3 - SOURCE”
FinTwitter, Bluesky · 15:36 UTC
“I've never seen a company with revenue in the billions with an 1100% increase in revenue year over year.”
Paul Graf, Bluesky · 16:17 UTC
“I wrote about why I think they've hit product market fit back in May”
Simon Willison, Bluesky · 13:58 UTC

More Anti-AI reactions (12)

“'we ought to be very concerned about the fact the AI companies we’re meant to trust to keep their models in check seem unable to do so. Worse than that, they don’t seem to know what their products are even”
Margot Finn, Bluesky · 06:04 UTC
“AI companies have shown themselves to be unqualified to keep track of their models - so why are we letting them be custodians of reporting incidents?”
Chris Stokel-Walker, Bluesky · 06:18 UTC
“Hope this goes on the history books as the 1st day the AI bubble started popping.”
Luis Alis・ルイス, Bluesky, skeptic · 12:04 UTC
“I'm genuinely wondering whether the current Stop Me Before I Train Again drama is because both OpenAI and Anthropic have been forced to the realisation that their costs are impossible and they desperately want it to look like "they made”
EssayWells, Bluesky, skeptic · 16:49 UTC
“AI companies like OpenAI and Anthropic will probably go out of business, and other companies like Google, Meta, Oracle, and Microsoft that went all in on the slop will take a huge hit they may never recover from.”
Sasquatch, Bluesky · 13:08 UTC
“OpenAI is doing live tool calls during long-horizon RL training runs?!? That might explain some of the external cyber attacks.”
Mark Riedl, Bluesky · 11:42 UTC
“if your product is shit gibbons throwing typewriters, just pretend it can end the world. The rubes will see your Phenomenal Cosmic power and FOMO will definitely make them give you more money.”
ninjamonkky.bsky.social, Bluesky, skeptic · 11:37 UTC
“"Move fast and break things"”
Ko van Huissteden, Bluesky · 15:30 UTC
“Wonder if all the people their products have killed already matter to them? Nah, of course not. They’re techbros. This is just PR for “how powerful our new stuff is!”.”
Ben Templesmith, Bluesky · 03:41 UTC
“First Thing: Anthropic warns of AI ‘existential risk’ as concerns emerge over Meta’s Muse and OpenAI’s model #TheGuardian”
#TuckFrump, Bluesky · 11:31 UTC
“They want a cartel. They want defense industry positions. They want their businesses to be protected by the state.”
animal from the international network, Bluesky · 04:21 UTC
“I think they blow a lot more hot air (which is saying a lot)”
Simon Willison, Bluesky · 05:11 UTC

More Middle Ground reactions (8)

“Ed's schtick of "well, 2025 losses should be indicative", are easily rebutted. I'd believe that spending has increased modestly for 2026, but revenue? you can look to the public statements about annualised revenue made by Anthropic execs under penalty of”
Liz Fong-Jones (方禮真), Bluesky, skeptic · 05:28 UTC
“every story about Anthropic or OpenAI makes me feel like I'm tripping on something”
Desarrayed, Bluesky · 17:00 UTC
“This year they found a business model, and it's not "AGI", it's coding agents”
Simon Willison, Bluesky · 05:06 UTC
“Things are all half-baked I'ts crazy to lock into OpenAI only, build on it, get stuck on it, then Anthropic or xAI releases machine god and I have to abandon my project to get access to it.”
Out From The Past, Bluesky, skeptic · 18:14 UTC
“not sure it comports with whatever the index below is meant to measure; token costs "collapsed"....to ruinously unprofitable late-2025 levels?”
YARDCG, Bluesky, skeptic · 05:37 UTC
“Their top models are the best but open source models can match them at significantly lower costs and better ownership.”
Vignesh NV, Bluesky · 12:28 UTC
“The 🐿️ does not believe that a PARMAP™ - ‘Pick a Revenue Multiple and Pray’ - approach to valuing the foundation model companies is going to get OpenAI and Anthropic public. So lets think through some scenarios 2/6”
Blind Squirrel Macro, Bluesky, skeptic · 16:32 UTC
“Cheaper models are nice, but the real story is the tradeoff”
papoo7.bsky.social, Bluesky · 18:12 UTC