Its April reports estimate cumulative AI spending of $7.6 trillion from 2026-2031, rising from $765 billion this year to $1.6 trillion by 2031.
James Covello said most enterprise deployments still show weak returns, citing surveys that found many AI pilots produced no value while risks, errors and IT costs kept rising.
Goldman said hyperscalers keep spending because of FOMO, with data-centre debt issuance reaching $182 billion in 2025, while Nvidia captures most compute economics and job losses appear modest.
If AI model costs are plummeting, why are enterprise financial losses from AI simultaneously skyrocketing?
As AI spending outpaces returns, what will be the catalyst for the inevitable market correction?
With Nvidia capturing nearly all AI profits, how can other tech giants escape this 'winner-take-all' dynamic?
The $700 Billion AI Infrastructure Surge: Unmatched Investment, Uncertain Returns, and Mounting Challenges in 2026
Overview
In 2026, major tech giants like Microsoft, Alphabet, Amazon, and Meta are investing between $635 billion and $665 billion to build the foundational compute capacity needed for advanced AI systems. This global race is driving massive spending on AI chips, servers, and data center infrastructure, with investors focusing on the physical and digital backbone powering the AI boom. The demand for specialized, AI-ready infrastructure is attracting significant capital, especially in regions experiencing strong growth from enterprise digitalization and cloud adoption. This unprecedented scale of investment highlights both the opportunities and challenges in the rapidly evolving AI landscape.