OpenAI CFO Sarah Friar tells employees that annualized revenue in July topped all of Q2
CNBC ● Covered by 4 sources
OpenAI's CFO told staff that July's annualized revenue beat the whole of Q2. She says growth is strong even as rivals close in.
Based on reporting by CNBC — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
OpenAI wants its own people to relax. In a Wednesday staff meeting, CFO Sarah Friar and board chair Bret Taylor laid out numbers meant to calm nerves about competition, according to a partial transcript reviewed by CNBC. Friar said the company's annualized recurring revenue in July topped the entirety of the second quarter, adding, with a bit of pride, that "Q2 was no slouch." The pitch: momentum is real, and it's coming from the GPT-5.6 model series, a new enterprise agent called ChatGPT Work, and rising use of the coding tool Codex.
The timing isn't random. OpenAI is trying to justify an $852 billion valuation while prepping for a possible IPO, all while the ground keeps shifting under it. Anthropic and Google remain steady rivals, but the newer headache is a wave of cheaper open-weight models coming out of China. Moonshot AI's Kimi K3, released earlier this month, claims to beat OpenAI and Anthropic's top systems on some benchmarks — a claim that, true or not, is exactly the kind of noise that makes investors twitchy.
Money is the whole game here. OpenAI told investors back in February it's aiming for roughly $600 billion in total compute spending by 2030, an outside-capital-fueled bet that requires matching revenue growth to keep the story credible. The company is reportedly in talks with Nvidia over a backstop worth up to $250 billion to help fund a massive new data center in Ohio. Big numbers demand big justification, and internal reassurance meetings like this one are part of how that justification gets delivered to the people actually building the product.
Anthropic's shadow looms large in all this. It overtook OpenAI in valuation earlier this year, and in May said its own revenue run rate hit $47 billion, up sharply from roughly $10 billion for all of 2025, largely on the strength of its Claude Code tool. Taylor didn't dodge it — he told employees Anthropic had a strong start to the year and admitted OpenAI had to catch up in coding. But he pointed to Codex's growth as evidence of a turn, noting that developers who ran up big bills on Claude Code have started shopping for alternatives.
The Information had reported in March that OpenAI's annualized revenue had recently crossed $25 billion. Both OpenAI and Anthropic quietly filed for IPOs with the SEC back in June, and neither has said when it actually plans to go public. For now, this is a company managing perception as much as product — trying to convince its own staff, and eventually the public markets, that the growth curve is real and not just a very expensive story.
My take — AI-written commentary, not fact-checked reporting
Internal pep talks about revenue always deserve a raised eyebrow, especially when they arrive right as a rival's numbers and a wave of cheap Chinese open-weight models start rattling confidence. Beating your own Q2 in a single month sounds great until you remember OpenAI is also on the hook for roughly $600 billion in compute spending — that kind of commitment needs more than a good month, it needs years of them. Codex catching up to Claude Code is a real signal worth watching, but comparing a monthly annualized run rate to a rival's independently disclosed figure isn't apples to apples, and everyone in that room knew it.
Read more about this at: CNBC