AWS CEO Matt Garman: AI spending bubble fears overstated

Garman's remarks come amid wider debate about whether the AI buildout is overheated, and he framed the current moment by comparing it to venture capital: failed startups can be offset by successful investments, and he noted the internet's long-term importance survived an earlier technology bubble. He also contrasted today's spending with speculative buildouts by saying demand is rooted in production workloads — computing, storage and AI inference — rather than experimental projects.

Separately, AWS has been pushing to make AI more affordable. Earlier this month, AWS Chief AI and Technology Officer Matt Wood said AI costs could fall substantially, potentially expanding adoption beyond today's largest customers. Amazon is reportedly in talks to transfer about $8 billion worth of Nvidia's Grace Blackwell AI chips to an investor-backed special-purpose vehicle that could raise funds through debt, with Amazon leasing the chips back — a structure that would free up capital and shift assets off its balance sheet. That context sits alongside Amazon CEO Andy Jassy's July decision to raise the company's 2026 capital expenditure forecast by $20 billion to $220 billion, citing rising memory chip prices and strong demand for AI and AWS infrastructure.