Amazon, Microsoft, Google, Meta Pour $650B+ Into AI Infrastructure in 2026

A spending wave unlike any in private-sector tech history is underway: Amazon, Microsoft, Alphabet, and Meta are committing $650–700 billion of capital in 2026 almost entirely toward AI-related infrastructure. Amazon is leading with $200 billion aimed at AWS data centers for generative AI, while Google expects up to $185 billion in capex focused on Gemini and Vertex AI, and Microsoft and Meta are not far behind ^{[1][2][3][4][5]}.

What makes this historic is not only the size—exceeding even the Apollo moonshot in relative terms—but the fact that it's driven by private companies betting on AI's business impact. The bulk of the money is being directed toward physical assets: hyperscale data centers packed with GPUs and custom chips, industrial-scale energy projects, and networking fabric that requires upfront cash and entails considerable maintenance. For example, more than two-thirds of Microsoft’s data center costs are classified as 'short-lived' assets (chips, servers) that must be replaced every 2–3 years ^[2]. This has turned once asset-light cloud players into investment-heavy industrial giants almost overnight ^[6][7].

The consequences are broad: AI infrastructure capex accounted for up to 90% of US GDP growth in early 2025 ^[8]; utilities, construction, and chip supply chains are all stretching to keep up ^[6]. Analysts warn that this kind of build-up, while still backed by sky-high demand, now depends on turning those bets into sustainable revenue—and may mean a shake-out for those unable to match the pace or prove ROI ^[2][5][8].