Gartner projects an AI data center revenue gap of at least $4.9 trillion through 2030, while Elon Musk warns that some computing capacity could be left stranded.
Gartner’s figures frame a financial question for the AI buildout: whether revenue from services using data centers will match the costs of new capacity and the returns owners seek. Gartner estimates cumulative revenue needs for data center owners at $8.5 trillion through 2030, enough to pay for announced infrastructure spending and financing costs and deliver a 12% return on invested capital. For enterprise and consumer applications, Gartner puts AI-linked revenue associated with data centers at a maximum of $3.6 trillion over that horizon. The difference between those estimates is at least $4.9 trillion; it is a forecast comparison, not a reported industry loss.
Gartner’s Oct. 14 forecast says independent operation will cease for more than 80% of neocloud providers by 2030. These firms provide the computing power used to run AI workloads. Gartner cited cost pressure and similar business models as factors behind the projection, and said hyperscalers have a broader range of revenue sources. The projection addresses independence; Gartner is not predicting that every one of these firms will close.
Separately, Tesla CEO Elon Musk wrote on X Saturday, Oct. 24: “I think we’re going to see a lot of stranded compute in partially complete data centers.” His post gave no project or capacity details and made no link to neoclouds or Gartner’s independence forecast.
Gartner’s spending projection puts AI infrastructure outlays near $1 trillion in 2026, with a further $6 trillion across 2027 through 2030. CoreWeave’s June 30, 2026, filing listed $35.068 billion in debt obligations after discounts and issuance costs, plus $16.319 billion in operating lease liabilities. CoreWeave reported quarterly revenue of $2.575 billion and a net loss of $626 million. The filing shows one company’s financial position, not that of the neocloud sector as a whole.
