Goldman Sachs Warns: Power Shortages May Become the Biggest Bottleneck for US AI Development

Goldman Sachs stated in a latest report that the biggest obstacle facing the United States in the field of artificial intelligence (AI) is not chips, rare earths, or talent, but electricity.

With the rapid development of AI and the surge in data centers, the US power grid is already overwhelmed. These large facilities now account for approximately 6% of total US electricity consumption.

Goldman Sachs analysts predict that the proportion of electricity consumption accounted for by data centers could nearly double to 11% by 2030, pushing parts of the US grid beyond critical load limits.

The analysts wrote: "As AI's demand for electricity surges, a reliable and adequate power supply could become the decisive factor in this competition, especially considering that bottlenecks in power infrastructure are often difficult to resolve quickly."

This issue could slow down the US lead in the AI technology race. The US currently leads in global AI infrastructure, possessing 44% of global data center capacity, roughly equivalent to the combined total of China, the EU, Japan, South Korea, and India.

However, the US power market is becoming increasingly tight. Due to increased demand from data centers, the US summer peak electricity reserve margin has dropped from 26% to 19% over the past five years.

Goldman Sachs stated that if AI continues to develop at its current pace, by 2030, the aforementioned reserve margin could fall below the 15% warning line, indicating "severe tightness."

Meanwhile, Goldman Sachs pointed out that while the US grid is under pressure, China is quietly stockpiling energy.