Gas turbine shortage and AI data center power gap

The global gas turbine shortage is intensifying as a critical bottleneck for artificial intelligence infrastructure. While global manufacturing capacity is estimated at 60 to 70 GW per year, total industry orders have surpassed 110 GW, causing capital costs for combined-cycle power plants to roughly double.

In the United States, the energy gap is particularly acute. Goldman Sachs projects U.S. data center power demand will rise from 31 GW in 2025 to approximately 66 GW by 2027. Morgan Stanley reports that between 2026 and 2028, U.S. data centers will require roughly 68 GW of new capacity, yet an estimated 38 GW of that demand remains uncovered by existing construction and contracted grid capacity.

This shortfall is exacerbated by grid interconnection delays, which can take five to seven years in some regions, whereas hyperscale operators require capacity within two to three years. Consequently, tech firms are increasingly pursuing “behind-the-meter” solutions to bypass grid congestion. Natural gas turbines are identified as a potential provider for 15 to 20 GW of this capacity.

Major manufacturers continue to face massive backlogs. GE Vernova reported a gas power equipment backlog of 116 GW as of Q2 2026, with delivery slots extending to 2031. Siemens Energy noted that approximately 60% of recent orders are linked to data center projects, while Mitsubishi Heavy Industries also maintains a significant large-frame turbine backlog. Other infrastructure providers, such as Eaton and Vertiv, are also seeing increased demand for electrical components and data center-related services.