Used Turbines and Jet Engines Become AI Power Rush's Stopgap Solution

Used turbines and converted jet engines are in demand as AI data centre power needs outstrip supply, with new equipment lead times stretching to 2031.

News source: Google /  https://cnalifestyle.channelnewsasia.com/

Utilities and data centre developers racing to power AI facilities are scouring the world for used turbines and repurposed jet engines, as lead times for new gas turbines stretch into the next decade. The secondary market for power generation equipment has become, in the words of one industry executive, "very, very hot right now."  The surge reflects a structural mismatch: global gas turbine orders now exceed 100 gigawatts, while manufacturing capacity sits between 60 and 70 gigawatts.

In the United States, delivery of a new gas turbine averages six years, with conversations firmly focused on the 2030 to 2031 window. That timeline is incompatible with the pace at which AI infrastructure is being built. Data centres and their operators have become major customers for gas turbines, a market that until a few years ago served almost exclusively traditional utilities.

Jet Engines Pulled from Boneyards

The most striking manifestation of the shortage is the revival of retired aircraft engines. Across the Sonoran desert in Arizona, thousands of decommissioned aircraft sit in boneyards, many with engines that have not fired in years. Several companies are now stripping those engines and converting them into electrical generators for data centres. ProEnergy, a Missouri-based manufacturer rebuilds GE engines commonly found on Boeing 747 aircraft into gas-fired turbines capable of producing roughly 50 megawatts each.

Elon Musk's xAI has installed eight converted jet-engine turbines in Mississippi to power its Colossus 2 data centre, according to state permits and satellite imagery. Crusoe, a Denver-based AI infrastructure firm, ordered 13 aeroderivative turbines from ProEnergy in April, with a combined capacity of 650 megawatts, for its US data centre projects. FTAI Aviation, a jet-engine leasing company, launched a division converting retired CFM56 engines into 25-megawatt turbines and signed a $1.5 billion supply agreement with an unnamed cloud service provider in July.

"The chips are going to be piling up and not be able to be turned on," Musk said earlier this year, identifying electricity rather than semiconductors as the primary bottleneck to AI development.

If the fastest path to powering AI runs through 1970s aircraft engines, what does that say about the planning horizon of the industry building it?

The Cost Equation

The economics favour the secondary market. Well-maintained used turbines sell for approximately $1.5 million to $1.7 million per megawatt, and refurbished units for about $1.3 million, according to Origin Partners. New turbines cost between $2 million and $2.5 million per megawatt. For smaller units, the discount can be steeper: two 14.5-megawatt Solar Turbines recently sold for $12.5 million, or roughly $430,000 per megawatt, compared with about $35 million for new equivalents with a five-year delivery window.

The trade-offs are real. Overhauled turbines carry unclear maintenance histories and limited remaining life. Converted jet engines are less efficient and more polluting than turbines designed for continuous power generation, and more units are required to produce the same output. Environmental groups have challenged permits for xAI's jet-powered generators, arguing they emit carbon monoxide at twice the rate of conventional turbines at the site.

A Partial Answer to a Structural Problem

Analysts caution that the secondary market cannot meaningfully offset the broader supply gap. "The market is already constrained, with limited excess supply, so it is unlikely to offset a meaningful share of the delays," said Noor Qureshi of Enverus Intelligence Research. Turbines are also just one constraint: air permits, grid access, and other factors compound the delays.

The policy environment has added pressure. President Trump has said data centres should generate their own electricity rather than draw from the grid, a stance that accelerated the turn to on-site generation. Finland's Wartsila has booked roughly 3 gigawatts of orders from US data centres for gas-engine power units since spring last year.

For now, the used turbines market serves as an improvisation, a way to bridge the years between AI's electricity demands and the infrastructure built to meet them. Whether it remains a stopgap or becomes a permanent feature of the power landscape depends on how quickly new turbine capacity and grid connections come online. The answer, at current pace, will not arrive before 2030.

FAQ:

Q: Why are data centres buying used turbines?
A: New gas turbine delivery times average six years in the US, incompatible with AI buildout schedules.

Q: What are converted jet engines used for?
A: Retired aircraft engines are rebuilt as gas-fired generators producing up to 50 megawatts each.

Q: Which companies are using jet-engine power?
A: xAI has eight units in Mississippi; Crusoe ordered 13 ProEnergy turbines.

Q: How much cheaper are used turbines?
A: Roughly $1.3M to $1.7M per MW versus $2M to $2.5M for new units.

Q: Are converted jet engines cleaner?
A: No, they are less efficient and emit more pollutants than conventional gas turbines.

Q: Can the secondary market solve the supply gap?
A: No, analysts say limited excess supply means it can only partially offset delays.