Gartner said AI data centres need $8.5tn in sales by 2030. It sees $3.6tn.
AI providers are expected to spend nearly $1 trillion on AI infrastructure in 2026 and another $6 trillion from 2027 to 2030, according to Gartner analyst Will Sommer, reported by The Register on 14 October.
To cover that spending and its financing at a 12% return on invested capital, Gartner estimates data centre operators would need $8.5 trillion in revenue over the period. It forecasts no more than $3.6 trillion of enterprise and consumer AI revenue attributable to those data centres.
The report's title puts capital write-downs above $4 trillion by 2032. Gartner expects 80% of neoclouds, the firms that rent out GPUs for AI work, not to survive as independent companies beyond 2030. It expects the hyperscalers to fare better because they sell more than AI. "Not everyone is in trouble," Sommer told The Register.
The gap between the two forecasts is $4.9 trillion. Part of the $3.6 trillion is what companies pay for AI that runs in those data centres. A gap is not yet a loss. Operators could cut investment, lower costs or accept lower returns, and Gartner's figures do not say which.
If your AI provider is bought or folds before 2030, what happens to the capacity you have prepaid?
Image: AI-generated.
Sources
Our file on Gartner
- 24 Sept
Gartner raised its forecast for 2026 AI spending to 2.7 trillion dollars.
- 10 Sept
Gartner said vendors call AI agents employees for one reason. Headcount budgets are bigger.
- 10 Sept
Gartner says a third of AI job cuts will reverse. Rehiring costs more.
- 8 Sept
Server prices up 125%. Lead times reaching 18 months.
Every story here is open to read. The ERP LEADERS brief goes one step further.
The day in enterprise software, in one email: the stories that matter, the failures with their figures, one case taken apart. Read a sample or sign up for the brief.
Welcome back. · Issue 01
