The United States leads the world in artificial intelligence, but its dominance may soon face a critical challenge — its aging and overstretched power grid. According to a new report from Goldman Sachs, the real bottleneck in America’s AI ambitions isn’t just chips, rare earth minerals, or talent — it’s electricity. Data centers powering AI models are already consuming about 6% of total U.S. electricity demand, and that figure is projected to rise to nearly 11% by 2030, according to Goldman’s analysts. This surge threatens to push parts of the nation’s grid beyond safe operating limits, creating a potential chokepoint for AI development. “The U.S. power market has been tightening,” the report notes, with peak summer spare power generation capacity dropping from 26% five years ago to 19% today. If current trends continue, that spare capacity could fall below the “critically tight” threshold of 15% by the end of the decade. Despite its lead in AI infrastructure — with 44% of global data center capacity, equal to the combined total of China, the EU, Japan, Korea, and India — the U.S. is struggling to keep up with energy demand. The country is retiring coal plants faster than it’s adding new natural gas or renewable energy capacity, and data center projects often face long approval timelines and a global shortage of gas-powered turbines. Meanwhile, China is making strategic moves to secure its energy future. Goldman projects that by 2030, China will have around 400 gigawatts of effective spare power capacity — more than three times the world’s expected data center power demand. This surplus means China can support rapid growth in AI infrastructure while still meeting needs in manufacturing, transportation, and other sectors. China’s energy buildup followed a major power shortage in 2021, which prompted Beijing to accelerate investments in renewables, natural gas, nuclear, and coal to ensure energy security. In contrast, U.S. states are introducing more regulations on data center expansion, with some imposing strict limits on energy use. Nvidia CEO Jensen Huang recently echoed these concerns, warning that an overabundance of regulations could stifle innovation in the U.S. He noted that China’s government-backed energy subsidies make it significantly cheaper for domestic tech firms to power AI chips — effectively making power “free” in some cases. Goldman’s analysts conclude that reliable and ample power supply will be a decisive factor in the global AI race. “Power infrastructure bottlenecks can be slow to solve,” they write, underscoring the urgency of addressing energy constraints if the U.S. wants to maintain its lead in artificial intelligence.