China’s Power-Grid Advantage Could Pressure U.S. AI Stocks

Summary

An OilPrice analysis argues that China’s AI industry may have a structural cost advantage because its models are reportedly achieving about 90% of U.S. performance at roughly 10% of the cost. Consultant Mehrdad Emadi attributes much of the gap to electricity, which he says can represent up to half of U.S. AI companies’ costs. The article links that advantage to China’s unified national grid and its extensive 800-1,100 kilovolt ultra-high-voltage direct-current network, while the United States operates three largely separate interconnections and relies more heavily on lower-voltage alternating-current transmission. The analysis says U.S. data-center capacity demand could rise from about 42 gigawatts to 118-134 gigawatts by 2030, with BloombergNEF projecting roughly 194 gigawatts by 2035. It argues that renewable generation is intermittent for always-on data centers, nuclear projects face long approval and construction timelines, and small modular reactors may not scale commercially until the late 2030s. Natural gas is presented as the more immediate power source, but gas-turbine suppliers GE Vernova, Siemens and Mitsubishi Heavy Industries reportedly have backlogs of up to seven years, with prices potentially tripling. The article then connects high costs and heavy borrowing, including private credit, to possible bankruptcies and a rapid selloff in U.S. AI valuations. Its quoted analysts offer highly severe scenarios, including a 35-50% decline in the sector’s total valuation, but these are forecasts and opinions rather than reported market outcomes.