Business

EPRI Report: US Data Center Grid Strain Casts Cloud Over AI Race

EPRI Report: US Data Center Grid Strain Casts Cloud Over AI Race

TechTarget and Informa Tech’s Digital Business Combine.TechTarget and Informa

Together, we power an unparalleled network of 220+ online properties covering 10,000+ granular topics, serving an audience of 50+ million professionals with original, objective content from trusted sources. We help you gain critical insights and make more informed decisions across your business priorities.

As the world wrestles with the AI arms race and power needs, the picture for the US is becoming clear: grid demands won’t be slowing down anytime soon.

The rapid expansion of data centers, fueled primarily by the AI boom, will dramatically reshape US electricity demand by the end of this decade.

According to the Electric Power Research Institute’s (EPRI) latest analysis, “Powering Intelligence 2026,” data centers could account for 9% to 17% of total US electricity consumption by 2030 – more than doubling their current share of 4–5%.

That’s a 60% increase over EPRI’s 2024 estimates, driven by an unprecedented surge in announced and under-construction data center projects over the past 18 months.

The fast acceleration is fueled by massive investments in large-scale AI-focused facilities, which require significantly more computational power than traditional cloud, streaming, or enterprise workloads.

In 2024, U.S. data centers consumed an estimated 177–192 terawatt-hours (TWh) of electricity, according to EPRI. By 2030, that figure could climb to roughly 380–790 TWh. To put this in perspective, a single large data center (100–1,000 MW) can draw power comparable to 80,000-800,000 average US homes, equivalent to a mid-sized to large city.

Related:AFCOM: Rack Density and Build-Outs Surge as AI Overhauls Data Center Design

Today’s aggregate data center load already rivals the combined demand of dozens of large cities; by 2030, additional growth could equal the power needs of multiple major metro areas.