Data centers in the United States are forecasted to consume four times more electricity by 2035 than they do today, fueled primarily by the rapid expansion of artificial intelligence workloads. This spike translates to nearly 200 gigawatts of capacity, placing unprecedented pressure on regional power systems, especially across the eastern and central regions.
- AI compute to drive data center electricity demand to nearly 200 GW by 2035
- PJM and ERCOT grids will face significant capacity pressures from new data centers
- Global data center electricity demand could rival India’s current annual consumption
What happened
A new BloombergNEF report projects that data centers in the US will use about 20% of the country’s electricity by 2035, which is four times their current consumption. This rise is primarily due to the exponential growth in AI compute requirements, with training and inference activities accounting for nearly half of the new capacity.
The US is expected to dominate AI chip power demand, hosting 64% of it by 2033. However, this dramatic expansion is putting substantial strain on regional power grids, especially in areas like the PJM Interconnection and ERCOT, where supply-demand imbalances are already pronounced.
Why it matters
The projected surge in electricity demand from data centers signals significant challenges for the US electrical grid infrastructure. Regions such as PJM face connection backlogs and rising energy costs, with prices jumping 76% in the past year due to tight capacity and competition between generators and large consumers like data centers.
As data centers increasingly monopolize grid capacity, balancing reliability, cost, and growth becomes more complex. These pressures may impact the pace and location of future data center development, potentially slowing down AI expansion or pushing it geographically.
What to watch next
Stakeholders should monitor how grid operators and utilities respond to the growing capacity demands from data centers, especially in strained regions like PJM and ERCOT. Infrastructure investments, regulatory policies, and innovations in energy efficiency will be critical to managing this growth sustainably.
Additionally, global trends in AI adoption could further escalate electricity demand worldwide, with forecasts indicating data center power needs might nearly match India’s entire annual electricity consumption by 2033. This emphasizes the need for wider coordination on energy planning and clean energy integration.