More than $3.5tn of AI infrastructure spending could face impairment by the end of 2030 if providers keep investing at the current pace, Gartner said on 14 October. The firm estimates that data centre owners would need $8.5tn in revenue to cover outlays and financing costs, against no more than $3.6tn of related AI revenue. The gap matters because impaired assets must be written down when they cannot earn back their cost.
Gartner figures on spending and revenue gap
Gartner expects AI providers to spend nearly $1tn on AI infrastructure in 2026, with another $6tn following between 2027 and 2030, according to senior director analyst Will Sommer. To support that program through 2030, owners would need $8.5tn in revenue, including financing costs and a 12% return on invested capital used as the hurdle rate. The forecast for business and consumer AI revenue tied to AI data centres is capped at $3.6tn for the same period. The research note behind the figures was written by Sommer, Mark McDonald and John Lovelock.
Most AI-native providers operate asset-light businesses, so Gartner expects less than half of AI revenue to flow to companies that own significant AI infrastructure. Sommer summed up the position by saying enterprise and consumer AI revenue will be insufficient to support profitable returns on all planned capital expenditure in the period. The phrase quoted in the statement was direct: for every dollar spent on compute, only a fraction goes to the owners of compute. That structure leaves infrastructure owners dependent on a smaller slice of total spending.
Neoclouds, companies renting out computing power built for AI, sit at the center of the warning. Gartner predicts more than 80% of them will stop operating independently by 2030, citing cost pressure and the lack of a distinct business model. Spending on AI-optimised cloud infrastructure is expected to grow from $66bn in 2027 to $174bn in 2030. Sommer called that growth too low, arguing most compute providers will not recover investment before newer AI chips replace the current generation.
What the warning means for AI buyers
For companies buying AI capacity, the split between hyperscalers and specialists becomes a practical selection factor. Gartner said hyperscalers have large and varied revenue that could support AI spending for years, while neoclouds must soon earn returns on AI-optimised infrastructure to keep running. Bloomberg named CoreWeave and Nebius as specialist providers more exposed than Microsoft, Alphabet and Amazon, noting a steep fall in CoreWeave shares and bonds this week over AI debt. A small buyer chasing low prices may get attractive terms now, while a large buyer needs contract terms that survive a provider exit.
Ambitions to move up the stack add another risk to assess during procurement. Many neoclouds want to sell higher-margin software and services like cloud providers, Sommer told Bloomberg, adding that few may succeed. His comment to the outlet was blunt: all of them want to be the full stack, but they cannot all be. Buyers should therefore verify whether a vendor earns money from differentiated services or mainly resells raw computing power under margin pressure.
The marker to watch is whether spending discipline or revenue growth closes the gap before 2030. If the current path holds, Gartner sees more than $4tn in asset value at write-down risk through 2032, with the 12% return arriving no earlier than the end of 2032. Related signals include the Bain estimate of $6tn in annual AI revenue needed by 2031 and the IMF warning this week about circular AI deals. A slowdown in capital expenditure or consolidation among neoclouds would show the forecast is biting.
