Nearly 50% of U.S. AI Data Center Projects Face Delays Amid Power and Supply-Chain Constraints

Generative AI has driven a surge in U.S. demand for data centers and high-performance computing, prompting technology giants to plan more than $650 billion in infrastructure investment. But insufficient grid capacity, tight supplies of critical equipment such as transformers, permitting reviews and shortages of skilled technical workers have made power and project delivery—not chips—a major bottleneck for AI expansion.

A recent survey found that nearly 50% of U.S. data center projects could be delayed or canceled. About 60% of projects scheduled for completion next year have yet to begin construction. Large campuses face lengthy waits for grid-connection approvals, transmission upgrades and transformer procurement. Even with more than $650 billion in planned investment, funding alone cannot quickly resolve power-supply and supply-chain constraints.