On September 29, 2026, the same day OpenAI hosted its annual DevDay developer conference in San Francisco, Axios reported a number that reset the conversation about the AI business: OpenAI's annualized revenue run-rate is approaching $70 billion, up more than 70% since the start of the third quarter. The striking detail was not the total but the source of the acceleration — business-to-business revenue had more than doubled over the same period, and the company had added more consumer revenue in the quarter than in all of 2025.
This was not a funding announcement or a product launch. It was evidence that the consumer phenomenon of ChatGPT is being converted into something much harder to build: a durable enterprise revenue engine. That conversion is one of the most instructive business stories of 2026.
Background
OpenAI's growth through 2024 and 2025 was famously consumer-led. ChatGPT became one of the fastest-adopted products in history, and revenue followed the familiar playbook of subscriptions, API usage, and developer tooling. But consumer momentum alone was not the question analysts were asking. The question was whether hundreds of millions of users experimenting with a chatbot could translate into the kind of contracted, recurring corporate revenue that justifies the enormous infrastructure spending behind the models.
In mid-August 2026, Bloomberg reported the run-rate had topped $40 billion — roughly double its level at the end of 2025. That was already a remarkable trajectory. The Axios reporting six weeks later suggested something had shifted gears entirely: enterprise sales moving at more than twice their July pace.
The challenge
Converting consumer enthusiasm into enterprise contracts is a classic valley of death for technology companies. Corporate buyers do not purchase on excitement; they purchase on procurement checklists — security, compliance, data handling, uptime, support, integration with existing systems. A product that spreads bottom-up through employees can stall at the IT department for years.
OpenAI also faced a structural competitor problem. Anthropic, its closest rival, had reported a revenue run-rate crossing $65 billion in July 2026, with 2025 booked revenue up twelvefold to $4.6 billion. The enterprise AI market was becoming a genuine two-horse race, and both companies were preparing for public listings — Anthropic potentially as soon as the fall of 2026, OpenAI later — which would turn every quarterly metric into a valuation argument. Annualized revenue, it is worth noting, is a projection extrapolated from a shorter period, not an audited annual figure — and neither company had disclosed the expense side of the growth, which Axios itself flagged as crucial context.
The approach
OpenAI's enterprise push has been systematic rather than theatrical. The DevDay timing of the Axios disclosure was no accident: the conference is the company's annual pitch to the developer ecosystem, the channel through which corporate adoption actually spreads. Developers build on the API, prototypes become pilots, pilots become procurement — and procurement becomes the kind of committed revenue that doubles in a quarter.
The company has also leaned into partnerships that de-risk enterprise adoption. Microsoft's close partnership embeds OpenAI's models inside the software environments corporate buyers already trust, while Oracle's massive compute buildout for OpenAI — the ChatGPT maker reportedly represents around half of Oracle's compute backlog, according to D.A. Davidson analyst Gil Luria — signals to the market that the capacity behind the contracts is real. "OpenAI accelerating growth is a very good sign for Oracle," Luria noted, "as it reinforces OpenAI's ability to live up to its expectations for compute capacity."
Behind the scenes, the financing strategy has matched the commercial one. Bloomberg reported on September 29 that OpenAI is looking to raise at least $30 billion at a valuation of about $1.4 trillion — roughly 20 times the annualized run-rate. That multiple only makes sense if investors believe the enterprise revenue is recurring, expanding, and durable. The company's public stance that it will not go public this year — citing safety concerns around its technology — buys it time to prove exactly that before public markets start asking quarterly questions.
Results as reported
The September 29 reporting, sourced to people familiar with OpenAI's finances, painted this picture:
- Annualized revenue run-rate approaching $70 billion, up more than 70% since the start of the third quarter.
- Business-to-business revenue more than doubled over the same period — the engine of the acceleration.
- More consumer revenue added in Q3 than in all of 2025, showing the consumer flywheel is still compounding underneath the enterprise story.
- Market reaction was immediate: Oracle shares rose between roughly 4% and 7% on the news, as investors read faster OpenAI revenue growth as validation of the compute buildout.
Important caveats apply. These are company-financial figures relayed by sources, not audited results or a published revenue breakdown. The mix between recurring contracts, API consumption, and one-off deals is unknown. And the expense picture — the cost of training and serving models at this scale — remains undisclosed. Any serious reading of the $70 billion number has to hold it alongside the $30 billion capital raise: this is a company growing revenue at historic speed while spending at historic speed too.
Key lessons
Distribution converts to enterprise if you build the on-ramp. ChatGPT's consumer reach became a corporate salesforce by another name — employees brought the tool to work, developers built on the API, and DevDay-style engagement turned experiments into contracts. The lesson for founders: bottom-up distribution is only half the job; the other half is procurement-ready packaging.
Time announcements to credibility moments. The revenue disclosure landing on DevDay, in front of the developer ecosystem that drives adoption, was choreography — and it worked. Narrative timing is a legitimate business discipline.
Run-rate numbers are claims, not proof. The most disciplined takeaway from this story is methodological: annualized revenue extrapolates the present into the future. It says nothing about margins, churn, or the cost of growth. The companies heading for IPOs know that public markets will demand the full picture — which is precisely why the enterprise mix matters so much.
Partnerships are enterprise infrastructure. Microsoft's distribution and Oracle's compute capacity do work that OpenAI cannot do alone: they turn a fast-moving startup into a vendor a CIO can defend buying. Growth this fast usually requires someone else's balance sheet somewhere in the stack.
Sources
