Power Grid Bottlenecks Threaten U.S. AI Ambitions

A critical shortage of electrical equipment is slowing the construction of AI data centers across the United States, creating an unexpected obstacle for the administration's push to outpace China in artificial intelligence. Despite executive orders prioritizing rapid buildout, the very tariffs designed to bolster domestic manufacturing are contributing to project delays.

According to a recent Bloomberg analysis, nearly half of the data centers planned for this year face postponement or cancellation. The bottleneck isn't the advanced computing hardware, but the foundational power infrastructure: transformers, switchgear, and batteries. For decades, the U.S. has relied on China for these components. Delivery times, once under three years, have now stretched to as long as five.

This timing is significant. Industry assessments suggest China trails the U.S. in AI development by approximately the same five-year window now required for key electrical gear. While the policy aim is to reshore production, U.S. factories currently cannot meet the explosive demand. Market analysts at Sightline Climate note that only one-third of the major AI data centers slated for 2026 are actually being built.

The pressure is leading some developers to absorb tariff costs and potential security concerns to secure Chinese equipment on faster schedules. The administration's recent directive for tech firms to 'build, bring, or buy' power for their data centers did not address this core procurement challenge. As one industry observer bluntly put it, a data center's power source is irrelevant if you lack the equipment to connect it.

Source: Ars Technica