Some AI data centers, stuck in years-long queues for grid connections and new turbines, have started making their own electricity using jet engines pulled from retired aircraft and rebuilt to burn natural gas, the New York Times reported on Tuesday. The improvisation is the latest sign of how far power supply has fallen behind the pace of the AI buildout — and how creative, and expedient, operators have become in closing the gap.
The engines are not anyone's idea of a permanent power plant. They are bought for one quality: speed. "It's all about getting power as fast as possible," Rystad Energy analyst Audun Martinsen told the Times. With jet-powered units, "they can have a data center running in 24 months," he said. For more context on this story, see our ongoing latest AI developments.
Twenty-Four Months Versus Seven Years
The comparison that explains the trend comes from grid arithmetic. Connecting 100 megawatts of new load to the grid in Northern Virginia — the world's largest data center market — takes about seven years, according to the International Energy Agency. And the backlog is not improving quickly: most new power capacity is expected to come online by 2031, an executive at Mitsubishi Power, one of the world's largest makers of power generation equipment, told Bloomberg last week.
Against those timelines, a rebuilt jet engine that can spin up within roughly two years looks less like a stopgap and more like the difference between capturing an AI market opportunity and missing it entirely. The trade-off is efficiency and fuel cost: aircraft-derived turbines burning natural gas are neither the cheapest nor the cleanest way to make bulk power, which is why utilities do not build them at data center scale. But for companies whose business models assume compute arrives on schedule, paying that premium beats waiting for the utility.
xAI's Mississippi Turbines and the Colossus Buildout
Elon Musk's xAI is among the companies already relying on the approach. The company 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, a Missouri-based company that specializes in rebuilding a General Electric engine model commonly found on Boeing commercial jets into stationary power generation units.
xAI's Memphis campus has been the most visible flashpoint for the practice. The company's use of on-site gas generation there has previously drawn air-quality litigation and regulatory scrutiny, and the Mississippi turbines extend the same playbook: when the interconnection queue is the bottleneck, generate next to the data center and accept the emissions trade-off.
OpenAI, Crusoe, and the Stargate Precedent
The approach predates the current wave. In 2025, OpenAI followed Musk's lead by deploying gas turbines at its first Stargate site in Abilene, Texas. Crusoe, the Denver-based company building that campus, ordered 13 of ProEnergy's converted units in April to power the site while larger, purpose-built capacity is developed.
Crusoe's procurement history also shows how volatile this market has become. In September, the company walked away from a $1.25 billion deal to buy 29 natural gas turbines from Boom Supersonic — the supersonic jet startup that began selling a variant of its Overture engine as a data center power plant. A Crusoe spokesperson said at the time that the company still plans to use turbines, but the scrapped order underlined how quickly sellers and buyers are recalibrating as everyone in the AI supply chain competes for the same scarce generation equipment.
When the Grid Says No, Projects Move or Die
The costs of waiting are no longer theoretical. Last month, Microsoft withdrew as the anchor customer for Nscale's planned data center in Loughton, England, after the local utility said it could not deliver enough power until the early to mid-2030s, TechCrunch reported. For a company racing to secure AI capacity, a project whose power arrives half a decade late is not a delayed asset — it is a dead one.
That dynamic is pushing the industry toward a structural shift. Data center operators that once treated power as a utility procurement problem now treat it as a core engineering discipline, building or contracting generation alongside compute. The Trump administration has encouraged the trend, pressing for data centers to generate their own electricity rather than wait in regional interconnection queues.
The Environmental Bill Comes Due
The jet-engine workaround solves the operator's scheduling problem while transferring costs elsewhere: local air quality, gas demand, and emissions that would otherwise require grid-scale planning. Communities near data center corridors — from Memphis to Northern Virginia to rural Texas — are increasingly organizing around exactly these impacts, and the machines being deployed are, by design, improvised rather than optimized.
The likely equilibrium is hybrid: converted turbines and other bridge generation to hit launch dates, followed by permanent capacity — grid upgrades, purpose-built gas plants, and whatever nuclear and renewable projects clear permitting — as the decade progresses. For now, though, the fastest path from GPU order to running AI cluster increasingly runs through a retired airliner's engines bolted to the local gas supply.
Follow our AI hardware and AI business coverage for more on the infrastructure race powering the AI boom.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
