OpenAI has scaled back its compute spending plans to $600 billion over the next four years, it has been reported.

The ChatGPT maker, which is in the midst of raising money for a long-talked about $100bn funding round, expects its revenue to grow rapidly to fund this heavy spending.

It made $13bn in 2025, but expects annual revenue to reach $280bn by 2030, according to sources close to the company that spilled the beans to CNBC, Reuters, and anyone else willing to listen.

Quite how it reached the $280bn projection is open to question, but the company has shown repeatedly in the past that it is not afraid to talk about big numbers. CEO Sam Altman had previously stated it would be spending $1.4 trillion on compute by 2033, though it would seem this has now been revised downwards.

Sources told CNBC that the $13bn revenue figure was higher than its anticipated $10bn annual income, while its costs were $8bn, lower than its $9bn target. As a private company, OpenAI does not have to report its financial position.

OpenAI ended 2025 with 1.9GW of compute capacity, according to CFO Sarah Friar. This apparently marked a 9.5× increase from 2023, with an annual increase of 3× – scaling from 200MW in 2023, to 600MW in 2024, then on to 1.9GW.

The conveniently timed leaks about higher-than-expected revenue come as OpenAI is poised to announce a blockbuster funding round to help finance its plans to scale up. It hopes to raise $100bn at a valuation of between $700-$800bn. As reported by DCD earlier this month, GPU giant Nvidia is expected to invest $30bn in the funding round, much of which it will likely receive back as payment for chips. Nvidia had previously said it was planning to invest up to $100bn in OpenAI on its own, but CEO Jensen Huang has poured cold water on this in recent weeks.

The company is in the midst of a major data center buildout in the US and around the world through its Stargate project, being undertaken in partnership with Oracle and SoftBank. Announced to great fanfare in January 2025, the project aims to develop an eventual 10GW of compute capacity for OpenAI, though progress has reportedly been patchy.

DCD has contacted the firm for comment on the reports.