OpenAI Tells Staff Revenue Growth Is Accelerating as IPO and Anthropic Pressure Mount
CFO Sarah Friar said July's annualized revenue alone beat all of the second quarter, as the company works to calm nerves about competition and its enormous spending plans.
OpenAI's finance chief has a message for employees worried about the competitive heat coming from Anthropic and a wave of cheap Chinese AI models: the numbers are still moving in the right direction, fast.
In an internal all-hands meeting on Wednesday, CFO Sarah Friar told staff that OpenAI's annualized recurring revenue in July alone surpassed the company's entire second-quarter total, according to a partial transcript of the meeting reviewed by CNBC. Friar didn't undersell the comparison, either. "And Q2 was no slouch," she said, according to the transcript.
The meeting, which also featured board chair Bret Taylor, amounted to a pep talk aimed at reassuring staff that the underlying business is healthy even as OpenAI fights on multiple fronts: an emboldened Anthropic in the enterprise market, Google's expanding AI push, and a growing crop of low-cost open-weight models out of China. Friar and Taylor pointed to three things driving the recent surge — the rollout of OpenAI's GPT-5.6 model series, the debut of an enterprise-focused agent called ChatGPT Work, and rising adoption of Codex, the company's AI coding tool.
Why OpenAI needs the growth story
The stakes behind that internal messaging are hard to overstate. OpenAI is preparing for what could be a massive initial public offering, and it needs to justify an $852 billion valuation in a market that's grown considerably more skeptical of AI spending. Both OpenAI and Anthropic confidentially filed paperwork for IPOs with the SEC in June, though neither has said publicly when it expects to actually go public.
That valuation is tied directly to OpenAI's infrastructure ambitions, which are enormous by any standard. In February, the company told investors it plans to spend roughly $600 billion on compute by 2030. It's currently negotiating with Nvidia over a backstop worth up to $250 billion to help finance a massive new AI data center OpenAI plans to lease in Ohio, CNBC reported earlier this week. Money like that doesn't come free, and outside investors funding those infrastructure bets want to see revenue that can eventually support the bill.
That's where the competitive squeeze gets uncomfortable. Anthropic, once seen as OpenAI's smaller sibling in the AI race, has been growing at a startling clip. The company said in May that its revenue run rate had hit $47 billion, up from roughly $10 billion for all of 2025, powered largely by Claude Code, its developer-focused coding tool that's become widely popular. Anthropic also passed OpenAI in valuation earlier this year — a symbolic but significant shift in a market that had assumed OpenAI's lead was untouchable.
For context on scale, The Information reported in March that OpenAI had recently crossed $25 billion in annualized revenue, citing a person familiar with the matter. Friar's comment that July's run rate alone beat the whole second quarter suggests the growth curve has steepened considerably since then, though CNBC's reporting doesn't include a specific dollar figure for July.
The coding market fight
Taylor didn't sugarcoat the competitive picture, telling employees that Anthropic had a strong start to the year and that OpenAI has had to play catch-up specifically in AI coding tools — a category Anthropic's Claude Code has come to dominate. But he expressed optimism about Codex closing the gap, framing some of that growth as coming directly at Anthropic's expense.
"You're seeing people who went deep on Claude Code, ended up with a very high bill, and started looking for an alternative," Taylor said, according to the transcript.
That's a notable dig, suggesting OpenAI sees an opening in customers who found Anthropic's coding tool effective but expensive. Whether that dynamic actually shows up in a meaningful way in OpenAI's revenue numbers isn't something Taylor's comments confirm on their own, but it signals where OpenAI thinks it can win back ground.
The cheaper-model threat from China
Beyond Anthropic, OpenAI faces pressure from an entirely different direction: open-weight models coming out of China that are far cheaper to run. Earlier this month, Chinese AI company Moonshot AI released Kimi K3, which the company claims narrows the performance gap with top U.S. models and even beats OpenAI's and Anthropic's most capable systems on certain benchmarks. Those models don't need to match GPT-5.6 or Claude on every metric to be disruptive — if developers can get most of the performance at a fraction of the cost, that reshapes the economics of the entire industry.
That's the backdrop against which OpenAI is racing to sign up enterprises and developers: it needs revenue growing fast enough to keep pace with its own infrastructure spending commitments, while fending off a cheaper alternative on one side and a fast-rising, well-funded rival on the other.
Neither OpenAI nor Anthropic has said when its IPO might actually happen. But the internal messaging this week makes clear that OpenAI's leadership wants employees — and, by extension, the market — to see July's numbers as proof that the growth story is still intact, even as the competitive landscape shifts month to month.
