Hyperscalers are set to nearly double AI spending to $690 billion by 2026, after spending roughly $380 billion on AI-related capital expenditures in 2025. Microsoft is leading the charge with AI-related expenditures expected to exceed $120 billion in fiscal 2026, an $80 billion Azure order backlog, and data-center lease commitments exceeding $300 billion. Nvidia captures an estimated 90% of AI accelerator spending, translating to approximately $180 billion in annual GPU purchases, but faces risks from customer concentration and a shift toward custom silicon. Analysts refer to a trillion-dollar capex gap, as current AI revenues across the entire industry remain in the low tens of billions annually. Microsoft occupies an asymmetric position by building both the infrastructure and selling the services that run on top of it, allowing it to monetize its massive capital investments directly.
Source: Read the original article
