Surging AI infrastructure energy consumption is raising the risk of electricity blackouts and grid strain across the United States.

Artificial intelligence's growing energy demand could trigger blackouts across the US as the country's electricity grid struggles to keep up. The North American Electric Reliability Corporation predicts that electricity consumption will rise 15% in the next decade. While many of the country's coal plants are shutting down, renewable energy sources are being linked to the grid slowly, placing some areas of the US at risk of energy shortfalls in 2025. Big Tech is scrambling to find different ways of powering its data centers, including through nuclear power, but NERC warned that the surge in demand will almost certainly outpace any new supply.

Expanding the grid to accommodate AI data centers has become a top national security priority for the outgoing Biden administration. In September, the White House hosted AI CEOs and energy company executives to discuss how to manage data center development. "The one thing that makes me nervous," US National Security Adviser Jake Sullivan said, "is the need for us to deploy clean energy rapidly enough to power the computing power necessary to stay at the cutting edge of artificial intelligence." Meanwhile, the incoming Trump administration is expected to slash regulation on fossil fuels, and the president-elect has repeatedly talked about "unleashing American energy," leading some industry insiders to predict a surge in new data centers.

If the US is unable to provide the energy needed for AI domestically, tech firms could turn to the Middle East, where laxer regulations and cheaper energy are turning the UAE and Saudi Arabia into data-center powerhouses. The Gulf states are investing heavily in AI infrastructure, although cooling the heat-producing facilities in the desert presents a challenge. The US has approved the sale of cutting-edge AI chips to an Emirati company after it cut ties to China, while Saudi officials have been lobbying Washington and tech executives to secure access to advanced AI chips.

The surging energy demand could lead to a rise in green investments, speeding up the development of renewable energy technology. Some AI companies are betting on small modular nuclear reactors, while geothermal energy companies have also signed major deals with certain tech companies. Another tactic could be to make AI development more energy efficient, by changing how chips are fabricated, building models with smaller and more tailored data, or scheduling training runs when there is a surge of renewable energy.