Companies building artificial intelligence infrastructure are increasingly facing an unexpected obstacle: connecting new data centers to power grids can take years. While waiting for new power plants, some operators have found a temporary solution: converting retired aircraft engines into natural gas generators.

As The New York Times reports, these units can provide electricity much sooner than a connection to the conventional power grid. They are far from an ideal replacement for full-scale power plants, but growing interest in them shows how quickly AI development has begun to outpace energy infrastructure, TechCrunch reported.

Years-long waits for electricity

The main advantage of converted aircraft engines is their speed of deployment. Companies that need to bring new computing capacity online as quickly as possible can no longer afford to wait for major energy projects to be completed.

Rystad Energy analyst Audun Martinsen told The New York Times that securing electricity as quickly as possible is the decisive factor in this situation. He estimates that a data center using these units can be brought online in roughly 24 months.

By comparison, connecting 100 MW of capacity to the grid in Northern Virginia takes about seven years, according to the International Energy Agency. The problem extends beyond grid connections: power equipment manufacturers are also struggling to keep up with demand. Mitsubishi Power, one of the world's largest manufacturers of power generation equipment, expects a substantial share of new capacity to become available only by 2031.

Delays are already affecting operators' business plans. In October, Microsoft withdrew as the anchor tenant for Nscale's data center in Loughton, England. The reason was a warning from the local electricity provider that it would not be able to supply the facility with enough power until the early to mid-2030s.

From retired aircraft to power plants for AI

Converting aircraft engines makes it possible to use existing equipment rather than wait for new gas turbines. The engines are removed from decommissioned aircraft, refurbished, and adapted to run on natural gas.

Elon Musk's xAI is among the companies already using this approach. According to The New York Times, the company installed eight aircraft-engine-based turbines in Mississippi. They help power the Colossus 2 data center in Memphis. The newspaper obtained information about the units from state permit documents and satellite imagery.

The equipment is supplied by ProEnergy, which converts GE engines that were widely used on Boeing aircraft.

In 2025, OpenAI also began using gas turbines to power its own computing facilities. The units were installed at the first Stargate project site in Abilene, Texas. The campus is being built by Crusoe, a Denver-based startup specializing in AI data centers. In April, the company ordered 13 converted engines from ProEnergy.

Crusoe is also exploring other energy options. In September, the startup pulled out of a $1.25 billion deal with Boom Supersonic to purchase 29 gas turbines. Boom is developing a supersonic passenger aircraft while also offering a version of its engine for use in data center power plants. After the deal was terminated, a Crusoe spokesperson told TechCrunch that the company still intended to use turbines.

A quick solution with environmental and technical trade-offs

Converted aircraft engines are attractive primarily because they eliminate the need to wait for new energy infrastructure to be built. However, they are less efficient than turbines originally designed for conventional gas-fired power plants.

These units also produce more pollution, and more of them are needed to generate the same amount of electricity. In other words, the time savings come with higher fuel consumption and a greater environmental burden.

Using aircraft engines may therefore prove to be more of a stopgap than a long-term strategy for powering data centers. As long as computing capacity needs to come online as quickly as possible, deployment speed outweighs the drawbacks. But some experts believe efficiency and environmental impacts could become much more important considerations for operators within the next year or two.

The situation is further complicated by the position of Donald Trump's administration, which is urging data center operators to supply their own electricity. This creates stronger incentives to seek alternatives to overloaded power grids.

As a result, AI development is changing not only the computing hardware market but also the approach to power supply itself: data centers are increasingly becoming facilities that must solve the challenge of generating electricity on their own. And while the conventional energy sector struggles to keep pace with the construction of these complexes, even engines that until recently powered passenger aircraft are being put to use.