The fundamental tension in AI infrastructure economics is directional. Building and operating the systems that train and serve large models is capital-intensive and scales with ambition, while the commercial price of AI services has come under sustained pressure. OpenAI has projected it expects to burn $280 billion by 2030, tying the figure to deeply negative cash flows driven by infrastructure investment and the pricing environment it faces.
The constraint sits in the spread between those two forces. As OpenAI continues to commit capital to infrastructure, its cost base grows. The pricing pressure the company has identified as a concurrent factor works against recovering those costs from revenue, which is why the cash flow projection runs so deeply negative.
The $280 billion represents what OpenAI projects it will burn through 2030. The company has named infrastructure investment and price pressures as the two drivers, and by its own account, both are already weighing on cash flow.