Aircraft engines instead of the power grid: xAI uses eight turbines for data center
Refitted units can deliver up to 50 MW each. The solution helps provide computing power with electricity but draws criticism from environmental organizations
Elon Musk's xAI is using eight refitted aircraft turbines in Mississippi to power its computing infrastructure. This is reported by media outlets citing permitting documentation and satellite imagery.
The units are supplied by ProEnergy. The company uses restored components from General Electric aircraft engines, including those from the family used on the Boeing 747, and adapts them for stationary electricity generation.
The capacity of one unit reaches 50 MW. For eight turbines, this amounts to up to 400 MW of total nominal capacity—though this estimate does not mean they all operate at maximum load continuously.
Self-generation allows data center operators to be less dependent on grid connection timelines. The construction of computing complexes can proceed faster than the expansion of the energy infrastructure required to power them.
However, gas turbines become a source of emissions. The Southern Environmental Law Center is challenging xAI's power plant permit in Mississippi. The organization claims that the planned measures do not adequately protect residents from air pollution, including formaldehyde.
A separate legal dispute concerns temporary gas units, which, according to plaintiffs, were operated without the necessary permits. These complaints pertain to different parts of the company's energy infrastructure: having a permit for one power plant does not automatically resolve all issues surrounding other units.
On October 30, a federal court refused to dismiss the NAACP's lawsuit. This allowed the proceedings to continue but does not represent a final decision on all alleged violations.
xAI's story shows that scaling AI requires addressing not only the supply of accelerators. Available electricity, permits, and the impact of generation on neighboring communities also become constraints for data center growth.
