AI Data Centers Turn to Converted Jet Engines | Zeniteq
Loading
Loading article
Zeniteq is now available on mobile for iPhone, iPad and Apple Watch.Download nowLearn more
AI Data Centers Turn to Converted Jet Engines
Years-long waits for grid connections and new turbines are pushing AI operators toward aircraft engines rebuilt for stationary power, with efficiency and pollution costs attached.
AI data center operators are turning to converted jet engines to solve a power-availability crisis, with xAI installing eight units in Mississippi and Crusoe ordering 13 for OpenAI's Stargate campus in Abilene, Texas, according to a November 3, 2026 TechCrunch report.
Missouri-based ProEnergy rebuilds General Electric aircraft engines to burn natural gas, with each unit rated at up to 50 megawatts. The appeal is speed — roughly 24 months to deployment versus seven years for a Northern Virginia grid connection — but TechCrunch reports the engines are less efficient, more polluting, and require more units than conventional gas-plant turbines.
xAI has installed eight converted jet-engine turbines in Mississippi, and Crusoe has ordered 13 units for OpenAI’s Stargate campus in Abilene, Texas, according to TechCrunch’s November 3, 2026 report. The draw is availability: aircraft engines rebuilt to burn natural gas can generate electricity while conventional power supplies remain years away.
The report, drawing on New York Times reporting, describes a power-procurement problem with real consequences for the AI industry. Companies can build data centers and acquire computing equipment faster than they can secure the electricity needed to run them. Converted engines offer a bridge, but the coverage describes them as less efficient and more polluting than turbines designed for conventional gas plants.
That makes this a scheduling decision with a long operating tail. Bringing servers online sooner may be commercially valuable, while every subsequent hour of generation adds fuel consumption and emissions.
The available evidence is secondary reporting, not independently inspected permits or equipment tests. The examples show that operators are adopting this approach; they do not show its prevalence across the industry or a universal cost advantage.
Power Delivery Has Become a Deployment Constraint
TechCrunch reports that converted jet-powered units can bring a data center online in about 24 months, citing Rystad Energy analyst Audun Martinsen speaking to the Times. In Northern Virginia, connecting a 100-megawatt load to the grid takes about seven years, according to an International Energy Agency estimate cited in the same coverage.
Those figures explain the appeal, but they need their conditions attached:
Power route
Published timing
What the figure describes
Converted jet-engine generation
About 24 months
An analyst’s estimate for bringing a data center online using jet-powered units
These are not matched bids for the same site. The 24-month estimate cannot be subtracted from the Northern Virginia figure to promise every developer a five-year saving. Location, project scale, fuel availability and approvals still matter.
The comparison nevertheless reveals why operators are considering less-efficient equipment. If electricity is the constraint holding up a completed campus, a faster power source changes when the computing investment can start producing useful work.
Power delays have already affected customer commitments. TechCrunch reports that Microsoft withdrew as the anchor customer for Nscale’s Loughton data center in England in October, after the local utility said sufficient power would not arrive until the early to mid-2030s.
My reading is that the strongest case for converted turbines is schedule insurance. Their value comes from making a deployment possible sooner, not from evidence that they produce the cheapest electricity. The published reporting does not provide enough pricing or operating data to calculate that premium.
How Aircraft Engines Become Stationary Generators
Missouri-based ProEnergy supplies the converted units described in the reports. TechCrunch says the company rebuilds General Electric engines commonly found on Boeing aircraft, adapting retired aviation equipment to burn natural gas.
At a basic level, the conversion uses a familiar turbine mechanism: combustion produces hot gas that turns rotating machinery. In a stationary generating installation, that mechanical energy drives an electrical generator instead of serving an aircraft’s propulsion needs.
The opened coverage does not provide the exact engine model, conversion specification or generator configuration. It supports the broad description of aircraft engines repurposed for electricity production, not a detailed engineering account of the modifications.
Tom’s Hardware reports that each ProEnergy unit can generate up to 50 megawatts. A megawatt measures power at a given moment; megawatt-hours measure energy delivered over time.
Using that published maximum, eight units would add up to 400 MW of potential generating capacity. If all eight delivered that output continuously for 24 hours, they would generate 9,600 MWh.
That is arithmetic based on the reported rating, not measured production. It does not account for downtime, operating conditions or electricity consumed by the generating installation itself. Even with that distinction, the scale is significant: these are substantial generating installations, not a handful of small emergency generators.
xAI and Crusoe Show Two Stages of Adoption
The two projects in TechCrunch’s report have different deployment statuses.
Project
Reported equipment
Status described
xAI’s Colossus 2 project
Eight ProEnergy converted jet-engine turbines in Mississippi
Installed
OpenAI’s Stargate campus in Abilene, built by Crusoe
Thirteen ProEnergy converted units
Ordered in April 2026
TechCrunch attributes the xAI installation details to the Times’ examination of state permits and satellite imagery. For Crusoe, an order establishes a purchasing commitment, not that every unit has been installed or is producing electricity.
Crusoe’s other turbine plans also show why different technologies should not be grouped together too loosely. In September, the company walked away from a $1.25 billion agreement to buy 29 natural-gas turbines from Boom Supersonic, according to TechCrunch. Boom is developing a supersonic passenger aircraft and selling a version of its jet engine for power generation. That is a different supply proposition from rebuilding retired aircraft engines.
A Crusoe spokesperson told TechCrunch at the time that the company still intended to use turbines. The canceled agreement therefore does not establish that Crusoe abandoned self-generation, nor does it explain why that particular purchase fell through.
For readers following AI competition, the distinction matters. An equipment order, a completed installation and an operating power supply are separate milestones. Counting all three as available capacity would make the infrastructure pipeline look more certain than the evidence supports.
Faster Deployment Comes With an Efficiency Bill
TechCrunch, citing the Times, reports three disadvantages of converted aircraft engines compared with conventional gas-plant turbines:
They are less efficient.
They produce more pollution.
More units are needed to deliver the same amount of power.
The coverage does not publish efficiency percentages, fuel-consumption figures or emissions per megawatt-hour. It therefore supports the direction of the tradeoff, but not a numerical comparison of operating costs.
The mechanism behind the efficiency penalty is straightforward. Lower electrical efficiency means more fuel is needed to produce a given amount of electricity. With the same natural-gas fuel, greater fuel consumption also means more combustion-related carbon dioxide for that electricity output, all else being equal.
That creates a distinction between getting power quickly and operating economically over time. Earlier availability may justify a higher generating cost during a constrained period. As operating hours accumulate, fuel efficiency becomes increasingly consequential.
I would treat any claim that repurposed engines are a cheaper overall solution cautiously. The reporting supplies neither purchase prices for the ProEnergy installations nor the fuel prices, maintenance costs and expected operating hours needed for a lifecycle comparison. Rapid availability is documented as the attraction; lifetime savings are not.
The reported need for more units also matters operationally. Multiple turbines provide generating capacity in increments, but each unit brings equipment that must be installed, operated and maintained. The sources do not provide enough detail to compare site footprints or staffing requirements numerically.
Mississippi’s Permit Dispute Makes Pollution Concrete
The environmental issue extends beyond carbon emissions. Tom’s Hardware reports that the Mississippi permits allow a 25% higher limit for volatile organic compounds, while also describing concerns about carbon monoxide.
That figure needs careful handling. A higher permitted limit is not a measurement of what the turbines actually emit. The coverage also does not make the comparison baseline clear enough to translate the number into a precise statement about additional neighborhood exposure.
The Southern Environmental Law Center has challenged Mississippi’s approval, according to Tom’s Hardware’s account of the Times reporting. The group argues that xAI failed to guarantee compliance with emissions standards for pollutants including formaldehyde, and alleges that projected emissions were undercounted.
Those are the organization’s arguments, not findings established here. The account also says the generators have state-level permissions and permits. Approval and a challenge to that approval can both be true: the dispute concerns whether the authorized installation adequately meets the applicable requirements.
The distinction between pollutants is useful. Carbon dioxide bears on climate impact. Carbon monoxide, volatile organic compounds and formaldehyde raise separate air-quality questions. A discussion limited to the turbines’ carbon footprint would leave out much of the local dispute.
My view is that the permit terms and actual operating hours are more informative than the novelty of using an aircraft engine. The practical environmental consequences depend on how much equipment runs, for how long, under what controls and with what verified emissions.
A Stopgap Needs a Defined Exit
Converted engines change the power-supply problem; they do not remove it. A developer substitutes a stationary generating project for some portion of its dependence on a future grid connection. Fuel supply, maintenance and environmental compliance then become part of delivering computing capacity.
The unresolved question is how long that arrangement lasts. Calling equipment temporary says little about its cumulative cost or emissions unless there is a credible replacement schedule.
For evaluating these projects, four details would be especially useful:
Expected annual operating hours: Occasional support and sustained generation have very different fuel and emissions consequences.
Measured fuel use per MWh: This would allow a meaningful efficiency comparison with other power options.
Pollutant limits and verified emissions: Permitted ceilings and actual releases answer different questions.
A replacement milestone: A grid-connection date or permanent-generation plan would define the intended bridge period.
The current reporting does not supply those details for both projects. It does, however, show a consequential shift in procurement priorities: xAI and Crusoe are sourcing aircraft-derived generating equipment because the normal power timetable conflicts with their deployment plans.
That strategy can be understandable without being an attractive permanent arrangement. The longer the bridge remains in service, the more its fuel consumption, maintenance needs and local emissions become part of the data center’s operating model. A stopgap’s real test is whether the project has a workable route beyond it.
Frequently Asked Questions
3 questions
1
Are These Engines Still Burning Jet Fuel?
The reported installations use aircraft engines rebuilt to burn natural gas for stationary electricity generation.
2
Are All 21 Reported Units Already Operating?
No. TechCrunch reports eight units installed for xAI and 13 ordered by Crusoe. An order does not confirm installation or electricity production.
3
Does a Higher Pollution Limit Mean Higher Measured Emissions?
No. A permit limit sets an authorized ceiling. Actual emissions require operating data or measurements, which the opened coverage does not provide.
