U.S. data centers projected to consume one-fifth of national electricity by 2035 amid AI surge

A new report from BloombergNEF projects that U.S. data centers will consume one-fifth of the nation’s electricity by 2035, quadrupling from current levels as AI-driven compute demand accelerates. The consultancy estimates total U.S. data center capacity will reach nearly 200 gigawatts by 2035, with 49% of that capacity allocated to AI training and inference workloads.

Nearly two-thirds of global AI chip power demand could be concentrated in the U.S. by 2033, BloombergNEF forecasts, reflecting the country’s dominance in AI infrastructure despite growing capacity elsewhere. The report also notes that its 2035 electricity demand estimate is 83% higher than BloombergNEF’s prior forecast from December 2024, underscoring the speed of AI deployment.

Grid operators are already feeling the strain. PJM Interconnection, which serves regions from Virginia to Illinois, may see 34% of its electricity devoted to data centers, while ERCOT in Texas could dedicate 22% of its generating capacity to the same purpose. PJM temporarily paused new grid connection applications for four years due to congestion, and even after reopening the queue, data centers represented 38% of charges in the grid manager’s most recent capacity auction.

Electricity prices in constrained regions have surged 76% over the past year amid supply-demand imbalances. The report highlights that while the U.S. leads in AI compute, international demand is also rising: by 2033, global data centers could add 1,935 terawatt-hours of new electricity demand—nearly equivalent to India’s current annual consumption—if AI adoption follows an aggressive growth trajectory.