The AI Power Crisis: Why Gas Turbines Are the Ultimate Bottleneck

While Goldman Sachs projects U.S. data center power demand will soar from 31 gigawatts in 2025 to 66 GW by 2027, the heavy-duty power equipment industry is operating on a much slower clock. The "Big Three"—GE Vernova, Siemens Energy, and Mitsubishi Heavy Industries—are effectively sold out for years, with a combined backlog and reservation list reaching 220 GW. Global manufacturing capacity remains stuck between 60 and 70 GW annually, making the industry's total output insufficient to meet even one year of the projected AI-driven demand spike.

This supply chain crisis is rooted in specialized manufacturing shortages, specifically a lack of skilled welders and foundries capable of producing high-temperature turbine components. Consequently, lead times for new power plants have extended from three and a half years to roughly five. Prices have reflected this scarcity; average combined-cycle project costs have climbed significantly, with turbine prices alone projected to rise nearly 200% compared to 2019.

Utilities are already feeling the strain. PJM, the largest wholesale power market in the U.S., recently faced a 6.8 GW shortfall in its capacity auction, despite paying record-high prices. In response, developers are being forced to place massive down payments on reservation slots—a strategy manufacturers use to offload project risk. As states like Texas and Illinois begin to audit and scale back ambitious data center interconnection queues, it is increasingly clear that the power bottleneck will not be solved on factory floors, but through regulatory intervention and the inevitable culling of speculative energy projects.