Adoption of small distributed gas turbines for data center on-site power generation is accelerating to meet surging AI infrastructure energy demands.
Demand for small gas turbines is surging as large data center developers increasingly seek behind-the-meter (BTM) generation to avoid lengthy grid connection delays. According to Enverus Intelligence Research, some 29.6 GW of BTM gas generation will be added in the U.S. through 2030, with data centers accounting for roughly 88% of demand. Developers are willing to pay higher power prices to prioritize faster deployment, and the shortage of manufacturing capacity for larger gas turbine units has led to rising costs and extended lead times.
Lead times vary significantly by technology. Large combined-cycle gas turbine (CCGT) plants can take up to six years to deliver, while small or medium-frame turbines require two years and aeroderivatives approximately 40 months. Aeroderivative turbines—derived from aircraft engines—have become the preference for many large-scale data center developers due to their rapid deployment and modular installation capabilities. Major suppliers include GE Vernova, Siemens Energy and Mitsubishi Power for large-frame units; GE Vernova, Siemens Energy and Baker Hughes for aeroderivatives; and Caterpillar, Wartsila and Innio Jenbacher for reciprocating engines.
Smaller turbines offer additional advantages: lower capital costs than larger units and the ability to install capacity incrementally in line with data center expansion. Crusoe, an AI infrastructure company, has exemplified this strategy by ordering 29 GE Vernova 35 MW aeroderivative gas turbines for U.S. data centers, including ten units for its Abilene, Texas campus. In June, Crusoe also agreed to procure around 750 MW of Bergen reciprocating gas engines for on-site generation at multiple sites.
"Some customers buy several smaller turbines for earlier availability, flexibility and redundancy," Matt Neal, President of Siemens Energy North America, told Reuters Events. "As a general rule, the larger a machine is, the longer it takes to manufacture and install."
Manufacturing constraints have driven substantial cost increases. Duke Indiana estimates its planned 1.5 GW Cayuga CCGT project will cost $3.3 billion—some $900 million more than assumed for new gas-fired capacity in 2024. However, while larger CCGTs carry higher upfront capital costs, they operate more efficiently over their lifetime. According to Bobby Noble, Senior Program Manager at the Electric Power Research Institute (EPRI), capital costs per unit of electricity can be up to 50% higher for smaller turbines.
Larger CCGTs also produce lower greenhouse gas emissions per MWh. Smaller open-cycle units have "lower firing temperatures and no heat recovery system," explained Dan O'Beirne, Interim Project Manager at Global Energy Monitor. This efficiency advantage must be weighed against the urgency developers face in deploying capacity quickly.
Manufacturing expansion is underway to address supply constraints. GE Vernova is investing $160 million to expand its Greenville, South Carolina factory by 2028, adding 650 workers, while Siemens Energy is restarting its Charlotte, North Carolina facility by the same year. These investments will help curb cost increases and shorten lead times, while also enabling developers to source domestically and avoid import tariffs and steel duties.
Some data center developers view on-site generation as temporary solutions—supporting operations until larger permanent power solutions or grid connections become available. Larger, newer gas-fired turbines are "more suited to eventually become a merchant grid asset" because their lower marginal costs preserve more value, according to Jack Painter, Vice President at Capstone. "Developers also care about the emissions profile, given that some of the final customers care about it," he noted.
Alternative technologies warrant consideration alongside conventional turbines. Virtual power plants, batteries, repowering with clean assets, hybrid resources, grid enhancements and energy efficiency improvements can be deployed incrementally. "Near-term alternatives could meet nearly all projected near-term U.S. load growth, while reducing exposure to turbine shortages and avoiding some of the reliability and affordability risks associated with waiting for new gas generation," said Mark Dyson, Managing Director at Electricity with the Rocky Mountain Institute (RMI).