The four U.S. hyperscalers—Alphabet, Meta, Microsoft and Amazon—have amassed nearly $2.4 trillion in future obligations tied to AI infrastructure, from data center buildouts to long-term energy and leasing contracts. Alphabet disclosed $902 billion in purchase commitments and leases not yet commenced, roughly nine times last year’s level, while Meta’s near-$700 billion in obligations includes data center leases stretching up to 30 years. The surge underscores managements’ conviction that AI demand will justify front-loaded spending even as Alphabet and Amazon dip into negative free cash flow, with Meta likely to follow. Amazon CEO Andy Jassy likened today’s outlays to AWS’s early years, saying even roughly $220 billion of 2026 capex won’t satisfy cloud demand; AWS revenue rose 37% in the latest quarter. Disclosures vary by company—some include content and hardware—complicating comparisons, but the direction is clear: Wall Street must weigh massive, multi-decade commitments against the promise of AI-driven returns.
Related articles:
Amazon and Google are winning the AI capex race — but what’s the prize?
Why Cloud Giants Are Racing to Build AI Data Centers
Meta’s Reality Labs Spending and the Cost of Its AI Ambitions
