OpenAI says it could have grown even faster if only it had more compute

OpenAI has released new business figures that the company claims show a direct link between compute capacity and revenue growth.

The company's compute capacity jumped from 0.2 gigawatts in 2023 to 0.6 gigawatts in 2024, with projections reaching 1.9 gigawatts in 2025 - roughly tripling each year. Revenue followed a similar curve: from 2 billion dollars in 2023 to 6 billion dollars in 2024 to more than 20 billion dollars expected in 2025. That's a tenfold increase in just two years.

OpenAI CFO Sarah Friar says the company "firmly believes" it could have grown faster with more compute and that there's a clear connection between the two: "more compute [...] would have led to faster customer adoption and monetization."

CEO Sam Altman has been even more direct, calling increased compute "the literal key to increasing revenue." Whether that formula holds up remains to be seen. OpenAI is currently in the middle of a massive funding round and needs to convince investors that its business model is sustainable.

Beyond compute, OpenAI wants to diversify its revenue streams by expanding into health, science, and enterprise markets. The company is also planning to run ads on ChatGPT to better monetize the 95 percent of ChatGPT users who don't pay for the service.

OpenAI's revenue targets are aggressive: the company is reportedly aiming for 145 billion dollars in revenue by 2029, with cash outflow expected to hit around 115 billion dollars by then. The biggest cost driver is compute. A recently announced partnership with Cerebras for more efficient inference—the process of an AI model generating responses—could help cut costs for running models at scale.

One interesting detail from the blog post: Microsoft isn't called out by name but lumped in as one of several "compute providers." The wording hints at where the partnership stands. Recent numbers show that Microsoft is now one of Anthropic's biggest customers. And Anthropic, alongside Google, is OpenAI's main competitor, particularly in the enterprise coding space.