OpenAI’s forecast turns $840bn of revenue into a $278bn financing problem

An investor presentation reportedly projects $278bn of cumulative negative free cash flow from 2026 through 2030. Revenue is forecast to rise from roughly $36bn this year to $350bn in 2030—about $840bn over the five years—while compute and infrastructure spending alone reaches approximately $856bn.

These are management forecasts, not audited outcomes. They assume both extraordinary demand and the ability to obtain chips, power, data centres and financing at a scale no software company has previously attempted. OpenAI is said to have about $122bn of available funds and is discussing another round around a $1.2tn valuation; Sam Altman has ruled out a 2026 IPO. Even if the revenue forecast proves accurate, the presentation implies that capital availability, rather than customer demand alone, remains the binding constraint.

The consequences travel through the supply chain. Nvidia, Oracle, SoftBank, cloud providers and data-centre developers are booking future revenue against OpenAI-related demand. If fundraising, construction or model economics disappoint, the risk is not confined to one balance sheet. Conversely, falling inference costs could help margins while simultaneously making model access more competitive and reducing pricing power. The revealing point is that enormous growth does not resolve the financing question: under OpenAI’s own reported scenario, each new dollar of demand brings enough capacity obligations to keep cash flow deeply negative for years.