The four largest hyperscalers are on pace to spend roughly $725 billion on AI capital expenditure in 2026 — up 77% from about $410 billion in 2025 — making this the largest coordinated technology buildout in history, and analysts already expect the number to top $1 trillion in 2027.
Who’s Spending What
According to capex tracking compiled by ValueAdd VC, Amazon leads at roughly $200 billion, followed by Microsoft at about $190 billion, Alphabet at $195-205 billion after raising its ceiling at Q2 earnings, and Meta at $125-145 billion after raising guidance twice. Including Oracle’s own buildout and the $500 billion Stargate project involving OpenAI and SoftBank, total sector AI infrastructure spending in 2026 exceeds $1 trillion already.
Where the Money Actually Goes
The overwhelming majority funds data centers, Nvidia GPUs, custom silicon, and — increasingly — power. Microsoft’s projected electricity demand for AI data centers is expected to surge over 600% by 2030, which explains why Google spent $4.75 billion acquiring a power company and Meta signed a major nuclear power purchase agreement this year alone.
Meta’s Framing Is Deliberately Different From Its Metaverse Era
CEO Mark Zuckerberg has explicitly framed this spending as a bet on “personal superintelligence” rather than the metaverse push that defined Meta’s prior capital cycle — a framing shift that’s coincided with the launch of Meta’s Muse AI agent, which represents the consumer-facing product this infrastructure spend is meant to eventually support.
Wall Street’s Growing Unease
Not everyone is convinced the spending pace is sustainable. Some analysts warn free cash flow could fall by up to 90% across the largest hyperscalers in 2026 as capex dramatically outpaces revenue growth, and the ratio of AI infrastructure spending to actual AI-generated revenue remains a genuine point of investor anxiety.
