Cornell Researchers Propose Power Credit Market to Ease AI Data Center Strain on Electric Grids

As the rapid expansion of artificial intelligence accelerates the construction of gigawatt-scale data centers across the United States, mounting pressure on regional electrical grids has raised concerns about reliability and infrastructure bottlenecks. To help mitigate the strain, researchers at Cornell University have proposed a market-based mechanism that would permit data centers facing mandatory power reductions to pay flexible energy users to take on those obligations during peak periods, as reported by TechXplore (https://techxplore.com/news/2026-10-power-reductions-center-strain-energy.html).

The findings were published on Sept. 22 in the journal Applied Energy by lead author and doctoral student Richard Mahuze, alongside Max Zhang, the Irving Porter Church professor of engineering in the Cornell Duffield College of Engineering. Their proposed framework, designated the Curtailment Credit Market (CCM), seeks to resolve the tension between strict grid curtailment rules and the operational demands of digital infrastructure.

To circumvent multi-year interconnection queues, utilities and regional transmission operators increasingly offer "non-firm" grid connections. Under these contracts, data centers can connect faster on the condition that they curtail power consumption when transmission lines face congestion. However, data centers risk substantial revenue losses and financial penalties under customer service-level agreements (SLAs) if they throttle compute capacity or halt operations.

Under the CCM framework, when a grid operator issues a curtailment signal, data centers could purchase curtailment credits from an ecosystem of flexible electricity consumers. Sellers could include industrial manufacturing facilities, commercial businesses, and virtual power plants (VPPs) aggregating distributed energy storage, electric vehicles, rooftop solar, and building control systems. The market would clear only transactions that the local transmission network can physically accommodate.

The researchers compare the mechanism to airline ticket vouchers, where carriers offer rising financial incentives to passengers willing to voluntarily give up seats on overbooked flights. By shifting reductions to parties capable of adjusting demand without losing critical service, the market could lower reliance on backup diesel generators and optimize existing transmission capacity.

In simulations across three test grids—including a simplified model of New York's electrical grid—the researchers found that trading curtailment obligations avoided at least one-third of the economic damage typically caused by power reductions. Supported by the Cornell Atkinson Center for Sustainability, the research team is currently collaborating with the Environmental Defense Fund to design an operational implementation of the market mechanism.