Can utilities satisfy AI's growing appetite for energy?

Exactly how the AI revolution will shape our future is unclear, but there is little doubt the utilities sector will benefit.

Like cryptocurrency mining and cloud servers, the computing power necessary to run AI programs require a good deal of energy. In the U.S., clusters of servers known as datacenters currently make up 4% of U.S. electricity demand, and a McKinsey study projects that to increase by a 15% CAGR over 2023-2030 to around 60 additional GW needed to meet the increased power demand—that’s basically double from where datacenters are today. To put this in context, the country’s total nuclear power output currently is around 100 GWs. 

The coming load growth presents an opportunity for regulated utilities to expand to meet this need and ensure reliability of the grid. Those that can do so, while keeping rates and costs low, should attract new datacenters, especially by coveted “hyperscalers” such as Amazon, Apple, Google, Meta and Microsoft.

Investors also stand to benefit from the bid to increase this capacity. Regulated utilities have a long track record of consistent total returns from operating low-risk business models, due to the regulations which provide visibility into rate-base growth, capital structures, allowed returns and ultimately earnings and dividend growth. Usage rates drive rate base growth. The more built to handle the rising AI load, the larger the usage should become. Taken together, this should in theory translate into increased earnings and dividends per share. The old and the new have a golden opportunity to create a prosperous future.