The United States hosts 5,427 data centers, more than ten times the count of any other country, and consumes more energy for data center operations than any other nation, according to data compiled in the 2026 Stanford AI Index Report. The concentration reflects decades of investment by US-based hyperscale operators including Amazon, Google, Microsoft, and Meta, whose combined capital expenditure exceeded $400 billion in 2025.
Statista market analysis shows global data center capacity demand continues to accelerate as AI model training and inference workloads require substantially more compute per task than traditional cloud workloads. The number of AI-focused data center projects under development in the US reached record levels in 2025, with hyperscale campuses announced or under construction in Virginia, Texas, Georgia, Arizona, and Iowa accounting for the majority of new capacity.
A significant supply chain risk identified by researchers is the concentration of AI chip fabrication at a single foundry. Taiwanese Semiconductor Manufacturing Company fabricates almost every leading AI chip used in US data centers, making the global AI hardware supply chain dependent on one facility in Taiwan. TSMC began operations at a new US facility in 2025 as part of a partial onshoring strategy, though the Arizona facility's output capacity remains a fraction of the Taiwan-based production that supplies the bulk of current AI chip demand.