$40 billion in August, $70 billion by late September — in barely a month, OpenAI's annualized revenue run rate nearly doubled. The figure surfaced at the company's DevDay conference in San Francisco, first reported by Axios, and it confirms what the industry has been whispering for a while: corporate demand for ChatGPT and OpenAI's API is outrunning anyone's ability to track it in real time.
A run rate simply takes the latest month of sales and multiplies it by twelve — a snapshot of current pace, not audited annual revenue. Still, the pace itself is striking. Since the quarter began, the metric has climbed more than 70%, and enterprise sales alone have more than doubled since July. Consumer revenue booked in the third quarter has already outpaced OpenAI's entire consumer haul for all of 2025.
For context, Anthropic — OpenAI's closest rival and the maker of Claude — reported an annualized run rate around $65 billion as of late July, a separate figure from the $65 billion funding round it closed a month earlier. It's a telling detail: a year ago the revenue gap between the two labs was measured in multiples, now it's down to single-digit billions.
The timing isn't accidental. OpenAI filed confidentially for a U.S. IPO back in June, and unconfirmed reports have already floated a valuation near $1.2 trillion, up from $852 billion this spring. A fast-growing run rate is exactly the kind of number that makes that case to investors who are increasingly asking about cash, not just capability.
But the money question doesn't end there. OpenAI still hasn't disclosed its compute spending or profitability, and those figures will ultimately decide whether it's a durable business or simply Silicon Valley's most expensive power customer. Anthropic's own IPO prospectus already showed operating losses above $8 billion and a decade-long infrastructure commitment of $518 billion — there's little reason to assume OpenAI's books look fundamentally different.
