US Power Grid Faces 55 GW Capacity Gap by 2030 Driven by AI Demand

The United States power sector is racing toward a significant capacity shortfall by 2030, fueled primarily by the explosive growth of AI datacenters. According to a report from McKinsey & Company, the risk of underinvesting in the grid now outweighs the danger of overbuilding, as new infrastructure will support broader electrification goals regardless of AI's long-term trajectory.

The scale of the demand is immense. Datacenters are projected to drive approximately 75% of all US power demand growth over the next decade. Current construction rates indicate that datacenters require nearly 30 GW of additional power every year, with 20 GW dedicated to IT loads and 10 GW required for cooling, distribution, and resilience.

This surge has created a precarious mathematical gap in the national energy balance. McKinsey estimates a projected capacity gap of 30 to 55 GW by 2030. This figure is derived from 120 GW of projected growth, offset by 40 GW of existing spare capacity and 100 GW of committed new capacity, while accounting for the loss of 50 to 75 GW from retiring coal and gas plants.

The Infrastructure Pivot

To manage this immediate pressure on an aging grid, the US is relying on a mix of legacy extensions and decentralized power. The Trump administration has authorized up to $500 million in funding to keep 13 coal-fired power plants operational to ensure energy security. Simultaneously, the industry is shifting toward self-sufficiency; nearly 60% of power sector leaders expect datacenters to maintain permanent on-site generation by 2030. Among those adopting on-site power, 64% intend to rely on natural gas.

Why the Investment is Future-Proof

While some critics warn of an "AI bubble," the strategic argument for aggressive investment is that the resulting grid upgrades are not wasted. Beyond AI, the US is facing increased demand from electric vehicles and the general electrification of industrial processes. By building out transmission and generation now, the US creates a foundation for a broader electrified economy, reducing the financial risk of the current building spree.

The Cost of Inaction

Failure to close the capacity gap could lead to severe grid instability and create a hard bottleneck for both technological innovation and climate goals. If the power sector cannot keep pace with the 30 GW annual requirement of the datacenter boom, the US risks stalling its lead in AI compute capabilities. The primary focus moving forward will be whether committed new capacity can be deployed fast enough to offset the retirement of traditional baseload plants.