AI Data Centers Seek Alternative Power Sources Amid Grid Delays
With lengthy waiting periods for grid connections and new turbines, some AI data centers are opting for a creative solution: generating their own electricity using retrofitted jet engines. According to a report by the New York Times, these engines, salvaged from decommissioned aircraft and refurbished to operate on natural gas, are becoming increasingly attractive in the race for efficient power supply.
The rapid deployment capabilities of these converted jet engines are a major draw. “It’s all about getting power as fast as possible,” remarked Audun Martinsen, an analyst at Rystad Energy, highlighting that data centers can be operational within 24 months using these gas-powered units.
In stark contrast, connecting to the grid for a 100-megawatt power source in Northern Virginia can take around seven years, as reported by the International Energy Agency. Most additional power capacity is not expected until 2031, according to recent statements from an executive at Mitsubishi Power, a leading manufacturer in the energy sector.
Some companies are facing dire consequences due to these delays. In October, Microsoft withdrew from a significant data center project with Nscale in Loughton, England, after the local utility indicated that it would not be able to deliver sufficient power until the early 2030s.
xAI, founded by Elon Musk, has already installed eight of these jet-engine turbines in Mississippi to support its Colossus 2 data center in Memphis, as confirmed through state permits and satellite data. These units have been refurbished by ProEnergy, utilizing engines commonly found on Boeing aircraft.
In a similar vein, OpenAI followed suit in 2025 by incorporating gas turbines at its inaugural Stargate site in Abilene, Texas. The construction of this facility is managed by Crusoe, a startup focused on developing AI-centric data centers, which ordered 13 of ProEnergy’s jet-engine turbines in April.
However, in September, Crusoe decided against a $1.25 billion agreement to purchase 29 natural gas turbines from Boom Supersonic, a firm aiming to create a supersonic passenger jet. A company spokesperson assured that Crusoe remains committed to utilizing turbines for their operations.
The pressure on data centers to produce their own electricity has been further amplified by actions from the Trump administration, which is advocating for self-sufficiency in energy generation.
Despite their appealing quick setup, converted jet engines come with several drawbacks. They tend to be less efficient and result in higher pollution levels than traditional gas turbines, necessitating a greater number of units to achieve the same power output. Some experts caution that efficiency and environmental considerations may grow increasingly relevant in the near future.
Editor’s Take
This shift toward using retrofitted jet engines reflects a critical issue in the power infrastructure for growing AI ventures. Companies must innovate to meet their immediate energy needs, but the long-term implications may pose challenges concerning efficiency and sustainability. If this trend gains traction, we could see a temporary solution overshadowing the vital need for a robust and greener power grid for the AI sector.
Source: techcrunch.com
