AI Data Centers Use Old Jet Engines to Bypass 7-Year Power Queues

Executive Overview

AI data centers are turning to retired jet engines repurposed to burn natural gas as a stopgap for on-site power, sidestepping grid interconnection queues that can stretch to roughly 7 years for a 100 MW connection in Northern Virginia. The jet-engine route cuts that timeline to about 24 months, according to Rystad Energy analyst Odom Martinsen speaking to The New York Times. xAI has installed 8 repurposed turbines in Mississippi to support its Colossus 2 data center in Memphis, while Crusoe ordered 13 ProEnergy units for the first Stargate site in Abilene, Texas, and withdrew from a $1.25 billion deal with Boom Supersonic for 29 natural gas turbines. The solution is temporary, less efficient, and more polluting than conventional gas turbines, but it underscores a widening electricity supply gap as AI infrastructure expands faster than grids can keep up.

Deep-Dive Features & Architecture

The engines are pulled from retired aircraft and rebuilt to burn natural gas. They are not designed to be permanent power plants, but they address the most pressing problem: time to power. "It's all about getting power as fast as possible," Rystad Energy analyst Odom Martinsen told The New York Times, noting that jet units can bring a data center online in 24 months. By contrast, connecting 100 MW to the grid in Northern Virginia takes about 7 years, according to the International Energy Agency, and industry sources expect most new capacity to arrive by 2031.

The delay is already costing operators real business. Last month, Microsoft withdrew from being the anchor customer for Nscale's data center in Loughton, England, after the local utility told it that sufficient power could not be supplied before the mid-2030s. In the United States, the Trump administration is pushing data centers to generate their own electricity, adding momentum to the shift.

On the deployment side, xAI installed eight repurposed jet-engine turbines in Mississippi to support Colossus 2 in Memphis, according to state permits and satellite imagery. The units are from ProEnergy and built on the GE engine common in Boeing aircraft. In 2025, OpenAI followed Musk's lead with gas turbines at the first Stargate site in Abilene, Texas, a site built by Crusoe, which ordered 13 ProEnergy units in April. In September, Crusoe withdrew from a $1.25 billion deal to buy 29 natural gas turbines from Boom Supersonic, though a spokesperson confirmed to TechCrunch that the company still plans to use turbines.

Benchmark & Competitive Performance

The core comparison here is not between AI models but between two power-provisioning paths: the conventional grid-connection route takes about 7 years for 100 MW in Northern Virginia, while the repurposed jet-engine route compresses the timeline to 24 months, a gap of nearly 5 years. The trade-off is efficiency and emissions: jet engines are less efficient and more polluting than turbines designed for conventional gas plants, and more units are needed to produce the same output. That makes them a short-term fix, with energy efficiency and environmental concerns expected to grow in importance within a year or two.

Industry Impact & Enterprise Adoption

The trend reflects a structural mismatch between AI infrastructure expansion and electricity supply. With grid queues stretching years, hyperscalers and data center developers are increasingly opting for self-generation, even at higher cost and emissions. The confirmed deployments—xAI's 8 turbines in Mississippi, Crusoe's 13 ProEnergy units in Abilene, and OpenAI's gas turbines at Stargate—show that major AI players are willing to accept interim, less efficient power solutions to avoid multi-year delays. The withdrawal from the $1.25 billion Boom Supersonic deal also signals that turbine supply chains and deal structures are still in flux, even as demand for fast power remains urgent.

Conclusion

Repurposed jet engines are an imperfect but pragmatic answer to the AI power crunch. They cut time to power from roughly 7 years to 24 months, enabling projects like xAI's Colossus 2 and OpenAI's Stargate to move forward. But their lower efficiency and higher emissions mean they are a bridge, not a destination. As grid capacity remains constrained and AI demand accelerates, the industry will need to balance speed of deployment with long-term energy efficiency and environmental impact.

Media Source: TechCrunch AI | Fact Verification & Analysis: AI Tools Oasis