OpenAI posts $25B revenue, $14B loss
- OpenAI’s latest financial debate turned on two clashing signals: Sarah Friar said demand is a “vertical wall,” but recent reports said growth missed plan. - The number making this feel real is simple — OpenAI says it now generates $2 billion a month, yet 2026 losses are still projected near $14 billion. - That gap matters because OpenAI just raised $122 billion at an $852 billion valuation, so the next fight is durability.
OpenAI is now in the weirdest possible position for a private tech company — huge, fast-growing, and still burning money at a scale that makes normal startup language useless. The business says it is generating $2 billion a month in revenue. But the argument around the company this week is not whether demand exists. It clearly does. The argument is whether demand is outrunning costs fast enough to justify the spending machine OpenAI has built around chips, data centers, and model training. ### What actually happened this week? The immediate trigger was a clash between two narratives. On April 28, reports said OpenAI had missed some internal revenue and user-growth targets and that finance chief Sarah Friar was worried about how to fund future compute agreements if growth slowed. Then on May 1, Friar pushed back publicly, saying OpenAI was “beating” its plan for products. ### Why does $25 billion sound bigger than it is? Because that figure is annualized revenue, not booked full-year revenue. OpenAI said on March 31 that it is now generating $2 billion in revenue per month. That run rate lines up with the chatter about roughly $25 billion annualized revenue by late February or early spring 2026. It means the company is proof that margins have caught up. ### So where is the money going? Mostly into compute. OpenAI’s whole strategy depends on securing massive amounts of data-center capacity to train frontier models and serve inference at global scale. Recent reporting tied OpenAI to giant commitments with Oracle, Amazon, Nvidia, and others, including a $300 billion five-year computing deal with Oracle, an expanded $138 billion. Globally, OpenAI is trying to lock in the industrial base for AI before competitors do. ### Why can losses stay huge if revenue is exploding? Because AI is not normal software. In classic SaaS, more users can mean better economics as the product scales. In frontier AI, more users also mean more inference bills, and better models usually mean even more expensive training runs. That is why a company can be adding billions in revenue even as its bill is growing with it. ### Didn’t OpenAI just raise a mountain of cash? Yes — and that is part of why
