Goldman Sachs highlighted in a research report that significant disparities in power supply are becoming a critical variable that could slow the expansion of AI infrastructure in the U.S. In contrast, China's ample and cost-controllable electricity supply may emerge as a new magnet for data center investments. This trend could not only reshape the global geographical distribution of data centers but also profoundly impact upstream and downstream industries, including energy equipment and high-power-consumption sectors like aluminum production.

According to Goldman Sachs' estimates, under rapid data center demand growth scenarios by 2030, the U.S. could see its effective reserve power capacity fall below the industry-recognized "critical threshold" of 15%, posing severe challenges to grid reliability. Meanwhile, China presents a starkly different picture. The report forecasts that, thanks to sustained large-scale investments in renewable energy, coal power, and nuclear energy, China will possess approximately 400 GW of reserve power capacity by 2030—far exceeding domestic demand and more than triple the projected global data center power requirement (around 120 GW).

The report warns that power infrastructure bottlenecks may temporarily hinder U.S. AI development progress at least until 2030. For businesses and investors reliant on large-scale data centers, certain U.S. regions could face higher electricity costs, prolonged project approval timelines, and even power shortages in the future.