A year ago, the U.S. faced a chip shortage. Now, that crisis has faded, but a new one looms: a critical shortage of transformers and other power infrastructure. Elon Musk warned on a podcast that within two years, the U.S.—and even the globe—might lack sufficient power to support the growing demand for AI chips. His prediction, once dismissed as alarmist, is now materializing.

The U.S. power grid faces unprecedented strain. According to the National Renewable Energy Laboratory (NREL), electricity demand could surge by 16% by 2030 due to advanced computing and electrification, requiring millions more transformers. By 2050, demand may jump 260% compared to 2021 levels. Supply chain disruptions from COVID-19, aging infrastructure (55% of residential transformers are near end-of-life), and extreme weather events like hurricanes and wildfires have exacerbated the crisis. For example, Duke Energy needed 16,000 transformers after Hurricanes Helene and Milton—more than some utilities use annually.

Transformer lead times now stretch to three years, up from 4–6 weeks pre-pandemic. Bottlenecks include scarce specialty electrical steel and over 80,000 transformer designs complicating production. Manufacturers like Schneider Electric and Hitachi Energy are investing billions to expand capacity, but experts warn these efforts may fall short. Solutions like standardized designs, federal policy support, and "storm stockpiles" are proposed, yet challenges remain. As NREL’s Killian McKenna notes, “Replacing aging infrastructure could be a golden opportunity—or a missed one.”