OpenAI's annualized recurring revenue in July surpassed the company's entire second quarter, CFO Sarah Friar told employees during an internal meeting. The growth came from three products: the GPT-5.6 model series, the enterprise agent ChatGPT Work, and the AI coding tool Codex.
Friar's remark that "Q2 was no slouch" signals confidence heading into a possible IPO. OpenAI has filed documents confidentially with the SEC, though no timeline has been set. With an $852 billion valuation on the line, the company needs to prove its revenue engine can keep pace with infrastructure costs it pegged at $600 billion through 2030.
What's driving the revenue jump?
Board chair Bret Taylor pointed to three pillars. GPT-5.6, the latest reasoning model, brought in new enterprise contracts. ChatGPT Work, the company's AI agent for business workflows, expanded adoption among existing customers. And Codex, OpenAI's coding assistant, is winning back users who tried competitors first.
Taylor acknowledged that Anthropic's Claude Code had taken an early lead in the AI coding market. But he claimed momentum is shifting. "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 CNBC.
That framing matters. Enterprise buyers treat AI coding tools as variable costs, and a usage-based model that balloons unpredictably pushes procurement teams to shop around. OpenAI is betting that Codex can capture that churn.
Anthropic's counter-narrative
Anthropic is not standing still. The company said in May that its revenue run rate exceeded the $47 billion it generated during all of 2025, with Claude Code as a major driver. That number, if accurate, suggests Anthropic's enterprise motion is scaling faster than most observers expected.
The two companies now compete on overlapping terrain: enterprise subscriptions, coding assistants, and AI agents. Neither has disclosed precise ARR figures, so the scoreboard remains opaque. What's clear is that both are burning capital on compute while racing to convert usage into durable contracts.
The $600 billion infrastructure bet
OpenAI told investors in February it plans to spend roughly $600 billion on compute through 2030. To fund part of that, it's negotiating with Nvidia for up to $250 billion in financing support tied to a large AI data center lease in Ohio.
That capital structure raises a question SaaS founders will recognize: can recurring revenue grow fast enough to justify a valuation built on future infrastructure? OpenAI's July ARR update is one data point. Investors will want many more before pricing an IPO.
Chinese competition heats up
Beyond Anthropic, OpenAI faces pressure from Chinese developers. Moonshot AI released Kimi K3 earlier this month, claiming it narrows the gap with leading US models and outperforms OpenAI and Anthropic on selected benchmarks. Whether those benchmarks translate to enterprise adoption remains to be seen, but the pricing advantage of open-weight Chinese models could squeeze margins industry-wide.
Logicity's Take
For SaaS operators watching the AI infrastructure race, the real signal here isn't ARR growth. It's margin. OpenAI is spending hundreds of billions on compute while fighting price wars against Anthropic, open-weight alternatives, and now Chinese labs. Revenue can climb while unit economics erode. Until OpenAI discloses gross margins on its enterprise products, the July ARR headline is encouraging noise, not proof of a sustainable business.
The internal meeting, CNBC reported, aimed to reassure employees about the company's trajectory. With an IPO potentially on the horizon and competition intensifying from multiple directions, that reassurance may need to become public soon.
Another example of AI agents entering enterprise workflows, the same market OpenAI's ChatGPT Work targets
Need Help Implementing This?
Logicity helps SaaS teams evaluate AI tooling decisions and build internal adoption playbooks. Reach out if you're comparing enterprise AI assistants or planning your own AI infrastructure spend.
Source: Tech-Economic Times / ET
Huma Shazia
Senior AI & Tech Writer
Produced with AI assistance and reviewed by the Logicity editorial team. Learn more in our Editorial Policy.
