AI infrastructure spending is set to far exceed revenue through 2030
Gartner estimates that expected AI infrastructure spending will far exceed revenue from AI services through 2030, making profitable returns on all planned investment unlikely. The gap could lead to asset write-downs and consolidation among providers, with smaller specialist operators particularly exposed.
AI providers are expected to spend nearly $1 trillion on infrastructure in 2026 and a further $6 trillion from 2027 to 2030. Gartner estimates they would need $8.5 trillion in revenue to cover spending and financing costs while earning a 12 per cent return, but forecasts no more than $3.6 trillion in attributable AI revenue; it predicts 80 per cent of neocloud operators will not survive independently beyond 2030. Hyperscalers may fare better because they have businesses beyond AI, while other analyses cited in the article also warn of revenue gaps and broader financial risks.
- AI infrastructure investment is forecast to outpace revenue.
- Gartner projects a $4.9 trillion gap by 2030.
- It expects 80 per cent of neoclouds to lose independence.
New here? Start with this
AI services run on data centres filled with powerful computers, which require large amounts of electricity, specialist equipment and investment. The companies building this infrastructure include large technology firms with many other sources of income, as well as smaller providers focused on renting computing power for AI.
The expected costs are much higher than the revenue analysts think AI services may bring in. That matters because providers may struggle to earn back what they invest, and some could have to reduce the value of their assets or join larger companies. Smaller operators may be more exposed because they have fewer other businesses to support them.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
Large-scale AI infrastructure spending could be justified if it builds capacity for services whose adoption and revenue take time to emerge. The investment may also deliver wider benefits through more capable tools and lower costs, while established cloud firms can spread risk across their other businesses; forecasts of a revenue shortfall are uncertain and do not prove that every project will fail.
The case against
The projected gap between infrastructure costs and attributable AI revenue suggests that planned investment may greatly exceed what customers will pay for the resulting services. If demand or returns fall short, providers could face debt, asset write-downs and consolidation, with smaller specialist operators especially vulnerable; capital committed to underused facilities could have earned better returns elsewhere.
Read the full article at the source →
Originally published by The Register as “AI math isn’t mathing: infrastructure spending forecast to outstrip expected revenue”.