OpenAI has recently missed its own targets for new users and revenue, stumbles that have spooked some leaders about funding its massive data-centre spending.
OpenAI chief financial officer Sarah Friar has told other company leaders she is worried the firm might not be able to pay future computing contracts if revenue fails to grow fast enough.
Board directors have been digging harder into OpenAI’s data-centre deals in recent months and questioning OpenAI chief executive Sam Altman’s push to lock in even more computing power despite the slowdown.
That scrutiny is squeezing Altman’s once-boundless ambitions ahead of a potential initial public offering that could land by the end of the year.
The Friar and other executives are trying to clamp down on costs and inject some discipline, which is putting them at odds with their chief executive, people familiar with the issue said.
OpenAI chief executive Sam Altman and Friar issued a joint statement saying: “We are totally aligned on buying as much compute as we can and working hard on it together every day,” in a joint statement. Any suggestion that the pair are divided or pulling back on securing new computing resources is ridiculous.”
For years, Altman has tried to lock up as much data-centre capacity as possible, arguing that computing shortages were the biggest constraint on OpenAI’s growth.
He went on a dealmaking spree last year that put OpenAI on the hook for about $600 billion in future spending commitments, tying a chunky slice of the sector’s fortunes to OpenAI’s.
The “buy everything” computing strategy rode high on ChatGPT’s seemingly unstoppable run, backed by Friar and the board.
Then the chatbot’s growth slowed toward the end of last year, and the mood shifted inside the building.
OpenAI missed an internal goal of reaching 1 billion weekly active users for ChatGPT by the end of last year. It still has not announced that milestone, leaving some investors twitchy.
OpenAI missed its yearly revenue target for ChatGPT, after Google’s Gemini saw massive growth late last year and pinched market share, the people said.
The company has been wrestling with subscriber defection rates. Earlier this year, OpenAI missed multiple monthly revenue targets after losing ground to Anthropic in coding and enterprise markets, people familiar with its finances said.
OpenAI recently raised $122 billion in what was described as the largest funding round in Silicon Valley history, putting it on a sturdier footing.
But it has signed up for so much computing power that it expects to burn through that amount within three years, assuming it hits ambitious revenue targets.
Its coding tool Codex is growing quickly, and the firm is trimming costs by cutting other projects such as its video-generation app Sora.
OpenAI recently released GPT-5.5, a model it says topped several industry benchmarks.
A number of AI companies, including Anthropic, have hit a capacity crunch in recent weeks, with price increases for access to AI processors, outages and rationing.
Power users have been getting narked, particularly coders who are used to systems finishing jobs cleanly and are now seeing them stall.
OpenAI said in a recent memo to investors it has been able to secure more computing capacity than Anthropic, giving it an advantage in reaching users.
In recent months, Friar has expressed reservations about OpenAI’s plan to go public by the end of this year. She has been pressing executives and board directors to improve internal controls, warning OpenAI is not ready for the reporting grind expected of a public company.
