$3.5 Trillion AI Infrastructure Spending Exposed to Impairment by 2030

Approximately $3.5 trillion in artificial intelligence infrastructure expenditures could face potential impairment by the conclusion of 2030 if current capital spending trajectories persist, according to Gartner. Furthermore, the firm projects that 80% of neocloud providers will likely cease independent operations within the same timeframe.

“In 2026, AI providers are expected to allocate nearly $1 trillion toward infrastructure, with an additional $6 trillion in projected investments between 2027 and 2030,” stated Will Sommer, Senior Director Analyst at Gartner. “However, our analysis suggests that enterprise and consumer AI revenues will be insufficient to yield profitable returns on these anticipated capital outlays during this period.”

“To cover announced expenditures, financing, and a 12% return on invested capital (ROIC), AI data center owners would require cumulative revenues of $8.5 trillion by 2030,” Sommer explained. “Conversely, total revenues attributable to AI data centers are forecast to reach only $3.6 trillion. As most AI-native firms utilize asset-light models, we anticipate that less than half of this revenue will be captured by infrastructure owners.”

A failure to achieve a sustainable ROIC on these rapidly depreciating assets will likely necessitate significant financial write-downs. Gartner projections indicate that if current investment trends continue, more than $4 trillion in asset value could be at risk of write-down through 2032. Under these conditions, the firm does not anticipate reaching the 12% ROIC threshold until the end of 2032 at the earliest.

Neocloud providers are expected to be the most vulnerable to this discrepancy between revenue and capital spending. While hyperscalers possess diversified revenue streams capable of sustaining long-term AI investments, neoclouds must achieve immediate returns on AI-optimized IaaS to remain viable.

Gartner estimates that spending on AI-optimized IaaS will grow from $66 billion in 2027 to $174 billion by 2030. “This growth rate remains too low for the majority of compute providers to recover their investments before current semiconductor generations are superseded,” Sommer noted. Consequently, Gartner predicts that cost pressures and a lack of differentiated business models will force 80% of neoclouds into consolidation or closure.