Gartner: AI infrastructure needs $8.5T in revenue by 2030 to pay back

On 14 October 2026, Gartner published an analysis from Stamford, Connecticut, examining whether the revenue generated by AI infrastructure can cover what it costs. The firm put the requirement at a cumulative $8.5 trillion in revenue through 2030, enough to cover announced capital spending, financing costs and a 12% return on invested capital. Gartner also warned that if AI providers keep spending at their current pace, more than $3.5 trillion of AI infrastructure investment could be exposed to impairment by the end of 2030.

Around 19 October, Gartner chief forecaster John Lovelock restated the $3.5 trillion figure in an interview, saying current returns do not yet support the full scale of investment. He noted that compute is already oversupplied, with large amounts of tokens and processing capacity being given to users through freemium models. In his view, emerging AI cloud companies will not collapse en masse, but roughly 80% may face consolidation through acquisitions and mergers. He added that most of the markets Gartner tracks around the AI race could shrink into a duopoly, with Anthropic and OpenAI as the two players, and that the compute assets neoclouds build to serve those two companies require cash flow that has not yet grown in proportion to the investment.

The two reports differ slightly on the 80% figure: one says more than 80% of neocloud providers will stop operating independently by 2030, while the other says about 80% of new cloud companies may face consolidation and M&A. The story currently rests at Gartner's published analysis and the chief forecaster's follow-up remarks.