OpenAI forecasts $856B in computing power spending by 2030

OpenAI expects to spend roughly $856 billion on computing power and infrastructure by the end of 2030. And it’s all going toward one thing: making AI models bigger, faster, and more capable. The figure represents the company’s single largest expense category and has been revised upward multiple times this year. Initial projections pegged the number at $600 billion back in February 2026. By July, it had climbed to approximately $750 billion. Now it sits at $856 billion, a 43% increase from the original estimate in less than a year. The math behind the megaspend OpenAI’s revenue forecasts are genuinely staggering. The company projects revenue will grow from $36 billion in 2026 to $350 billion by 2030, with cumulative revenue over that five-year stretch landing near $840 billion. The gap between what comes in and what goes out is expected to produce a negative free cash flow of $278 billion from 2026 to 2030. For 2026 alone, compute expenditures are projected at $50 billion, a figure OpenAI President Greg Brockman laid out during testimony in May 2026. The company raised $122 billion in a funding round in March 2026, at an estimated valuation of $852 billion. That funding is expected ...