Data centers are turning to old jet engines as the wait for power grows

Speed is the main attraction. Rystad Energy analyst Audun Martinsen told the Times that jet-powered units let operators get a data center running in about 24 months. By contrast, the International Energy Agency estimates connecting 100 megawatts to the grid in Northern Virginia takes roughly seven years, with most capacity expected online by 2031, a Mitsubishi Power executive told Bloomberg.

Delays are already costing operators contracts. Microsoft withdrew as anchor customer for Nscale's data center in Loughton, England, after the local utility said sufficient power would not be available until the early-to-mid 2030s. Meanwhile, the Trump administration is encouraging data centers to generate their own electricity, adding further pressure toward self-supply.

Adoption is spreading quickly. xAI has installed eight converted jet-engine turbines in Mississippi to help power its Colossus 2 data center in Memphis, the Times reported, citing state permits and satellite imagery; the units come from ProEnergy, which rebuilds a GE engine commonly used on Boeing jets. OpenAI followed with gas turbines at its first Stargate site in Abilene, Texas, in 2025, and Crusoe, the Denver startup building that campus, ordered 13 ProEnergy units in April. In September, Crusoe abandoned a $1.25 billion deal to buy 29 natural gas turbines from Boom Supersonic, though it said it still plans to use turbines.

The trade-offs are real. Converted engines are less efficient, pollute more, and require more units for equivalent output than turbines designed for conventional gas plants, the Times noted. Some experts believe efficiency and environmental concerns could grow in importance within a year or two, potentially making jet-engine power a short-lived solution.

What to watch: whether the 24-month advantage survives as emissions scrutiny and fuel costs tighten, how many operators follow xAI and Crusoe before turbine supply and grid queues ease, and whether policy support for on-site generation persists — or whether stranded, underused converted units become the next write-down for AI infrastructure builders.