US Hyperscalers Set to Spend $5 Trillion on AI Compute by 2030 as Moats Shrink
Updated
Updated · Financial Times · Jun 25
US Hyperscalers Set to Spend $5 Trillion on AI Compute by 2030 as Moats Shrink
3 articles · Updated · Financial Times · Jun 25
Summary
$5 trillion in projected spending by 2030 would lock Meta, Microsoft, Alphabet and Amazon into one of tech’s biggest capital bets, with some already trimming buybacks and raising debt or equity.
3.3 times annual growth in AI-chip computing capacity since 2022 has improved large language models, but bigger data centers and more GPUs have not eliminated hallucinations or reasoning errors.
That scaling race is also eroding differentiation: rivals are building similar models, price wars are intensifying, and high operating costs are squeezing margins while some companies turn to cheaper open-source Chinese models.
Depreciating chips and the risk that more efficient AI systems displace today’s LLM-heavy approach could leave the industry with excess data centers and weaker returns than investors expect.
A failed compute boom could spread beyond Big Tech to banks, pension funds and the wider economy, reviving bailout fears after OpenAI briefly floated government loan guarantees for data-center construction.
Will new, efficient AI models make today's multi-trillion dollar data centers obsolete before they pay off?
As Chinese open-source AI surges, are US tech giants spending trillions only to lose the global innovation race?
Is AI's $5 trillion infrastructure bet creating the next 'too big to fail' financial crisis?
Powering the AI Boom: Data Center Capacity, Environmental Impact, and Market Dynamics Through 2030
Overview
The report highlights a global surge in AI compute investment, driven by the escalating demands of artificial intelligence and high-performance computing. This unprecedented wave is causing a dramatic increase in data center capacity, with facilities now expected to operate at the gigawatt scale. The rapid evolution and widespread adoption of generative AI are placing extraordinary power requirements on data centers, fundamentally shifting infrastructure needs. As a result, the industry is experiencing a supercycle of growth, with annual data center demand projected to rise sharply through 2030, reshaping both technology and energy landscapes worldwide.