Artificial intelligence companies are increasingly turning to retired aircraft jet engines to generate electricity for data centres as delays in obtaining grid connections threaten the expansion of their computing infrastructure. Across the United States, companies are removing engines from decommissioned aircraft, rebuilding them to operate on natural gas and connecting them to electricity generators. These converted engines, known as aeroderivative gas turbines, offer a way to secure power faster than waiting for conventional grid infrastructure or newly manufactured industrial turbines.
The trend highlights a growing challenge for the AI industry: computing chips and data centre construction are advancing faster than electricity infrastructure. Some data centre operators face several years of waiting for grid connections, while converted aircraft engines can potentially provide electricity within approximately 24 months. Companies including xAI and Crusoe are already using or ordering these systems. However, the technology comes with trade-offs, including lower efficiency and higher pollution compared with purpose-built power generation equipment.
Why AI Data Centers Are Turning to Retired Jet Engines
The rapid expansion of artificial intelligence has dramatically increased electricity requirements for data centres.
Modern AI facilities contain thousands of specialised processors that consume substantial amounts of electricity. They also require cooling systems, networking equipment and other infrastructure that operate continuously.
As companies such as Microsoft, Google, Meta and OpenAI expand their AI computing capacity, securing reliable electricity has become a major infrastructure challenge.
Traditional electricity grids were not designed to accommodate the sudden arrival of enormous data centre loads in certain locations. Building new transmission lines, upgrading substations and obtaining regulatory approvals can take several years.
In some major US data centre markets, connecting a facility requiring 100 megawatts of electricity can take approximately seven years, according to figures cited by the International Energy Agency.
For AI companies operating under aggressive infrastructure expansion schedules, waiting that long can delay projects and increase costs.
Retired aircraft engines offer an alternative because they can be converted into stationary electricity generators and deployed closer to data centre facilities.
How Old Aircraft Engines Are Converted Into Electricity Generators
The technology behind these systems is not entirely new. Aeroderivative turbines have been used in electricity generation for decades, but growing demand from AI infrastructure developers is creating a new market for engines recovered from retired aircraft.
The conversion process involves modifying the engine so that it generates mechanical power for an electricity generator rather than producing thrust for an aircraft.
The process generally involves several steps:
- Recovering retired engines: Companies purchase engines from decommissioned aircraft stored at aircraft storage and reclamation facilities.
- Rebuilding the engine: Engineers inspect, repair and modify the engine for stationary power generation.
- Changing the fuel system: The combustion system is adapted to operate on natural gas.
- Connecting a generator: The engine’s rotating shaft is connected to electrical generation equipment.
- Installing the system: The turbine and associated equipment are packaged for deployment at or near a data centre.
The resulting system can generate electricity without depending entirely on the local utility grid.
The engines are particularly attractive because aircraft turbines are designed to deliver substantial power while maintaining relatively compact dimensions.
xAI and Crusoe Are Among the Companies Using the Technology
Several companies involved in AI infrastructure are already exploring or deploying converted aircraft engines.
Elon Musk’s xAI has installed eight converted jet-engine turbines in Mississippi to help power its Colossus 2 data centre in Memphis, according to permits and satellite imagery reviewed by The New York Times.
The units are supplied by Missouri-based ProEnergy, which converts General Electric engines commonly used in Boeing aircraft into stationary electricity generators.
Each converted turbine can produce approximately 50 megawatts of electricity, according to ProEnergy executives. That makes multiple units useful for supplying substantial power to large computing facilities.
Crusoe, a Denver-based data centre developer, is another major participant.
The company ordered 13 converted turbines from ProEnergy in April 2026. Crusoe has also previously worked with GE Vernova on aeroderivative turbine packages for its AI infrastructure projects.
In 2025, GE Vernova announced a 29-unit agreement involving its LM2500XPRESS turbine packages for Crusoe’s AI data centres. The units were expected to provide nearly one gigawatt of electricity.
Crusoe subsequently withdrew from a separate $1.25 billion agreement to purchase 29 natural gas turbines from Boom Supersonic in September 2026. The company indicated that it still intended to use turbine-based power generation.
These developments demonstrate that AI infrastructure developers are considering multiple approaches to securing electricity rather than relying exclusively on conventional utility connections.
GE Vernova Reports a Sharp Increase in Aeroderivative Turbine Orders
The growing demand for alternative power generation is also visible in equipment manufacturers’ financial disclosures.
GE Vernova reported a substantial increase in orders for aeroderivative gas turbines during the third quarter of 2026.
| GE Vernova aeroderivative turbines | Q3 2025 | Q3 2026 |
|---|---|---|
| Orders | 9 | 57 |
| Units sold | 6 | 13 |
The company also reported 127 aeroderivative orders during the first nine months of 2026, compared with 45 during the same period in 2025.
The figures indicate a significant increase in demand for this category of power-generation equipment.
However, GE Vernova’s filing does not identify all customers or specify how many orders are intended for AI data centres. The increase therefore reflects broader demand for aeroderivative turbines and should not be interpreted as an exact measure of AI data centre purchases.
Why Data Centers Cannot Simply Wait for Grid Power
Electricity availability has become one of the biggest constraints on the construction of new AI infrastructure.
Building a data centre requires more than securing land and purchasing processors. Developers must also arrange electricity supply, cooling infrastructure, backup systems and transmission capacity.
Grid connections can be delayed by several factors:
- Transmission infrastructure: New power lines and substations require significant investment and construction.
- Regulatory approvals: Energy projects often require permits from multiple authorities.
- Growing electricity demand: Utilities are handling requests from data centres and other large industrial customers.
- Equipment shortages: Transformers, turbines and other power-generation components can have long delivery schedules.
- Local opposition: Some communities oppose new transmission infrastructure and large industrial developments.
For AI companies, these delays can affect the timing of expensive computing investments.
A data centre filled with advanced GPUs cannot generate revenue if it does not have sufficient electricity to operate.
This explains why developers are increasingly considering on-site power generation, even when it is more expensive than conventional electricity supply.
Old Jet Engines Offer Speed but Create Environmental Concerns
Although converted aircraft engines can help solve electricity shortages, they are not an ideal long-term solution.
Older engines generally operate less efficiently than purpose-built stationary power turbines. They can also produce higher emissions per unit of electricity generated, depending on the engine design, fuel and operating conditions.
The use of natural gas also means these facilities continue to produce carbon dioxide emissions.
There are additional concerns about local air pollution. Permits reviewed by The New York Times showed that xAI’s converted generators in Mississippi were allowed higher limits for certain pollutants, including carbon monoxide and volatile organic compounds, than other turbines at the site.
Environmental groups have raised concerns about using older, less efficient equipment to power an industry that is increasingly associated with technological progress.
Another challenge involves maintenance. Aircraft engines require specialised servicing, replacement components and regular inspections. Operating several converted engines at one facility can create additional maintenance and fuel-management requirements.
The technology may therefore work best as a temporary power source while data centre operators wait for more efficient and reliable electricity infrastructure.
Companies Are Exploring Other Ways to Power AI Data Centers
Repurposed aircraft engines are only one part of a broader shift towards on-site electricity generation.
AI companies are also exploring several alternatives:
| Power source | Potential advantage | Main challenge |
|---|---|---|
| Converted jet engines | Faster deployment using available engines | Efficiency and pollution |
| Conventional gas turbines | Large-scale electricity generation | Equipment availability and delivery times |
| Solar and battery storage | Lower operational emissions | Intermittent generation and storage costs |
| Nuclear energy | Reliable, low-carbon electricity | Long development timelines |
| Fuel cells | Modular on-site generation | Cost and fuel availability |
| Grid connections | Access to broader electricity networks | Transmission and interconnection delays |
Some companies are also signing long-term electricity agreements with power producers and exploring dedicated generation facilities.
The choice depends on how quickly electricity is needed, the location of the data centre, fuel availability, environmental regulations and the total cost of operating the facility.
The Bigger Picture
The growing use of retired jet engines reflects a structural challenge in the AI industry. The expansion of computing capacity is increasingly dependent on physical infrastructure, particularly electricity generation and transmission.
AI companies can purchase advanced processors and construct data centres, but those investments cannot deliver their full value without reliable power. The willingness to repurpose aircraft engines demonstrates how urgently developers are trying to close this gap.
However, the approach also highlights a tension between rapid AI expansion and environmental sustainability. If older turbines become a significant source of electricity for data centres, regulators and local communities may demand stronger pollution controls and clearer long-term energy plans.
Looking Ahead
The demand for aeroderivative turbines is likely to remain connected to the speed at which utilities can provide electricity to new data centre facilities. Companies such as ProEnergy, GE Vernova and other equipment suppliers could benefit from the need for faster power deployment. At the same time, conventional gas turbines, nuclear power, renewable energy and battery storage will remain important parts of the industry’s longer-term energy strategy.
For AI developers, the priority is securing enough reliable electricity to operate increasingly powerful computing systems. Converted aircraft engines offer one way to address immediate shortages, but their efficiency, maintenance requirements and environmental impact could limit their long-term role. The larger challenge remains building electricity infrastructure quickly enough to support the next generation of AI data centres.
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
