The American power grid is entering a period of unprecedented stress. As demand from data centers, electric vehicles, and electrified manufacturing continues to accelerate, the generation fleet responsible for meeting that demand is simultaneously shrinking. The result is a reliability gap that most policymakers have yet to fully grasp.
According to the North American Electric Reliability Corporation, peak electricity demand in the United States is projected to grow by over 120 gigawatts in the next decade — roughly equivalent to adding the entire generating capacity of Texas to the national grid. Yet planned retirements of coal, natural gas, and nuclear plants continue to outpace new dispatchable generation coming online.
The baseload paradox
At the core of this challenge lies what might be called the baseload paradox. As intermittent renewable sources — primarily wind and solar — have grown to represent a larger share of the generation mix, the economic viability of the very plants that keep the grid stable during periods of low renewable output has deteriorated.
Natural gas peaker plants, which operate only during periods of peak demand, face increasingly compressed operating margins. Nuclear plants, despite their unmatched capacity factors and zero-emission profiles, struggle to compete in wholesale markets that undervalue their reliability contributions. The result is a generation portfolio that looks increasingly fragile precisely when it needs to be most robust.
We are systematically dismantling the infrastructure that keeps the lights on during extreme weather events — and replacing it with generation that cannot be dispatched on demand.
Mark P. Mills, NCEA Senior FellowCascading risk: beyond blackouts
The consequences of grid instability extend far beyond the inconvenience of rolling blackouts. When reserve margins tighten, electricity prices become volatile. Industrial consumers — from semiconductor fabs to pharmaceutical manufacturers — face operational uncertainty that undermines capital investment decisions. Data center operators, who require contractual guarantees of near-perfect uptime, may redirect billions in planned infrastructure spending to regions with more reliable power.
The economic modeling is stark. Research from NCEA estimates that each percentage point decline in grid reliability metrics correlates with measurable reductions in regional GDP growth, particularly in energy-intensive manufacturing corridors. For states actively competing for data center investment, grid reliability has become a decisive factor in corporate site selection.
The AI demand accelerant
The rapid scaling of artificial intelligence workloads has introduced a demand variable that most grid planners did not anticipate even three years ago. A single large-scale AI training cluster can consume as much electricity as a small city. The race among technology companies to build ever-larger computing infrastructure is creating concentrated load pockets that existing transmission networks were never designed to serve.
Unlike traditional data center loads, which can be somewhat flexible in their timing, AI training runs represent sustained, high-intensity demand that cannot easily be shifted or curtailed. This creates a baseload demand profile that paradoxically mirrors the very type of generation being retired from the fleet.
What policymakers should consider
Addressing the reliability gap will require coordinated action across multiple policy dimensions. Based on NCEA's research, several priority areas emerge:
- Capacity market reform — Current market designs in most regions fail to adequately compensate generators for the reliability services they provide. Reforms should ensure that dispatchable generation receives appropriate value for its contribution to grid stability.
- Permitting and licensing acceleration — New nuclear, natural gas, and advanced geothermal projects face regulatory timelines that are fundamentally incompatible with the pace of demand growth. Streamlined permitting processes are essential.
- Transmission infrastructure investment — Even where generation capacity exists, inadequate transmission often prevents it from reaching load centers. The current pace of transmission buildout needs to roughly triple.
- Technology-neutral clean energy standards — Policies that prescribe specific generation technologies rather than setting performance standards risk constraining the flexibility needed to maintain reliability while reducing emissions.
The path forward
The United States has navigated energy transitions before, but never at this pace and never with demand growth of this magnitude. The decisions made by federal and state policymakers over the next two to three years will determine whether the grid remains capable of supporting both the digital economy and the broader industrial base.
NCEA will continue to provide nonpartisan analysis and direct briefings to officials at all levels of government. Our goal is to ensure that energy policy decisions are grounded in rigorous data rather than ideological assumptions — because the stakes are too high for anything less.
1 North American Electric Reliability Corporation, 2025 Long-Term Reliability Assessment.
2 NCEA analysis based on EIA Annual Energy Outlook 2025 projections and state-level retirement filings.
3 Based on NCEA economic modeling using regional GDP data and NERC reliability metrics, 2020-2025.
