Gartner puts more than $3.5 trillion of AI infrastructure spending at risk

Gartner put more than $3.5 trillion of AI infrastructure spending at risk of impairment by 2030 in an analysis reported on 15 October, against a forecast of no more than $3.6 trillion in attributable enterprise and consumer AI revenue through that year, Bloomberg reported.

Gartner sets an $8.5 trillion revenue hurdle

AI data centre owners would need cumulative revenue of $8.5 trillion through 2030 to meet capital spending and financing costs while earning a 12% return on invested capital, Tech Wire Asia reported, citing Gartner analyst Will Sommer. Gartner forecasts no more than $3.6 trillion in enterprise and consumer AI revenue attributable to those data centres over the period.

Gartner puts the projected gap at $4.9 trillion against its revenue requirement, which includes the assumed return. The forecast accounts for depreciation of AI computing infrastructure. Under a continuation of the current investment pace, Gartner estimates that more than $4 trillion in AI infrastructure assets could face write-down risk through 2032 and does not expect the sector to reach its assumed 12% return before the end of that year.

The revenue forecast differs in another account of Gartner’s analysis: TechTarget reported projected AI revenue of $3.5 trillion through 2030. Bloomberg and Tech Wire Asia reported the figure as no more than $3.6 trillion.

AI providers face a $6 trillion build-out

Gartner projects infrastructure outlays by AI providers of nearly $1 trillion in 2026 and a further $6 trillion across 2027–2030. Its impairment forecast applies if providers maintain their current investment plans. The spending creates costs for infrastructure owners before the forecast revenue is earned, including financing expenses and depreciation on computing equipment.

In a separate worldwide spending forecast, Gartner puts total AI expenditure at $2.7 trillion in 2026, up 49.5% from 2025. Its infrastructure category accounts for approximately $1.48 trillion, or 56% of that total, alongside $576.5 billion for AI services and $461.6 billion for software. Gartner projects infrastructure spending in that category to reach $1.98 trillion in 2027.

CoreWeave’s costs offer a company-level example. Its annual filing put technology and infrastructure expenses at $2.93 billion in 2025, compared with $961 million in 2024, Tech Wire Asia reported. Depreciation and amortisation accounted for approximately $2.3 billion of the 2025 figure, up from $843 million a year earlier. CoreWeave estimates a six-year useful life for its data centre computing hardware, having extended it from five years effective January 2023.

Neoclouds confront the 2030 revenue test

Gartner expects neoclouds, specialist providers of AI computing services, to face greater financial pressure than established hyperscalers. Neoclouds depend more heavily on revenue from AI compute and related platforms, while hyperscalers have more diversified sources of income to support infrastructure investment. Gartner also expects infrastructure owners to receive less than half the revenue generated by AI services because many AI-native providers do not own the computing capacity they use.

Spending on AI-optimised infrastructure-as-a-service is forecast to rise from $66 billion in 2027 to $174 billion in 2030. Gartner nevertheless forecasts difficulty recouping investments for most AI compute providers before the next generation of AI semiconductors supplants today’s chips.

Gartner predicts that more than 80% of neocloud providers will cease independent operations by 2030 because of cost pressure and difficulty distinguishing their business models. Sommer put the figure at 80% in comments reported by TechTarget. He said ceasing independent operations could mean an acquisition, a partnership or failure.