The aircraft graveyards of the American Southwest are becoming an unlikely staging ground for the AI boom. Retired jet engines, once left idle in desert boneyards, are being stripped from old planes and rebuilt to burn natural gas for data centers desperate to secure electricity.¹

The shift follows a widening mismatch between the speed of AI construction and the pace of power infrastructure. Developers can wait years for grid connections and new generation equipment: connecting 100 megawatts in Northern Virginia can take about seven years, while converted jet-engine units can help get a data center operating in roughly 24 months, according to Rystad Energy analyst Audun Martinsen. “It’s all about getting power as fast as possible,” he said.²

That pressure is already reshaping projects. Microsoft withdrew as the anchor customer for Nscale’s proposed Loughton, England, data center after the utility said adequate power would not arrive until the early or mid-2030s. In the US, xAI installed eight converted jet-engine turbines to support its Colossus 2 site in Memphis, while OpenAI’s first Stargate site in Abilene, Texas, followed with gas turbines in 2025.²

The political push is reinforcing the commercial scramble. President Donald Trump has urged data-center operators to provide their own power rather than draw from the grid, declaring in August: “Data centers now are mandated by me that they build their own electricity, so they’re not taking it from the grid.”¹

For developers, the appeal is bluntly transactional: speed. “How fast can you get me energy?” said Bill Kleyman, chief executive of Apolo, describing an industry now obsessed with “speed to power.”¹ Yet the stopgap carries a clear downside. Converted engines are less efficient, require more units for comparable output and pollute more than purpose-built gas turbines—suggesting the shortcut may become harder to defend once immediate capacity pressures ease.²