The global AI buildout has now officially outpaced the electrical grid, a human institution that took over a century to construct. The solution, arrived at by some of the most sophisticated technology companies on Earth, is to strap old jet engines to the side of a data center and hope for the best.
It's all about getting power as fast as possible — and apparently, speed now has a wingspan.
What happened
Connecting 100 megawatts to the grid in Northern Virginia takes approximately seven years, according to the International Energy Agency. Converted jet engines, rebuilt from retired aircraft and fed natural gas, can have a data center running in 24 months. The AI industry has therefore concluded that the grid is too slow and has decided to bring its own.
Elon Musk's xAI installed eight converted jet-engine turbines in Mississippi to help power its Colossus 2 data center in Memphis. The engines come from ProEnergy, which rebuilds a GE model commonly found on Boeing jets — aircraft that were, until recently, busy being late rather than generating electricity for chatbots.
OpenAI followed with gas turbines at its first Stargate site in Abilene, Texas, in 2025. Crusoe, the Denver-based company building that campus, ordered 13 ProEnergy units in April. Most major grid capacity is not expected to arrive until 2031, which is, coincidentally, also when several of these companies' investors are hoping to see returns.
Why the humans care
The delays are not theoretical. Microsoft withdrew last month as the anchor customer for Nscale's data center in Loughton, England, after the local utility confirmed it could not deliver sufficient power until the early-to-mid 2030s. When AI companies describe power availability as a competitive advantage, this is what they mean: the lights are literally on or they are not.
The Trump administration is adding structural pressure by encouraging data centers to generate their own electricity, which the industry was already doing out of desperation rather than ideology. Jet engines burn more fuel, pollute more, and require more units to match the output of conventional turbines — trade-offs the industry has accepted with the characteristic calm of people who do not pay the electricity bill personally.
What happens next
Experts suggest efficiency and environmental concerns may grow in importance within a year or two, at which point the industry will presumably locate a new creative workaround. Boom Supersonic, a company developing a supersonic passenger jet, has already begun selling a version of that jet's engine as a data center power plant, which means the AI industry is now being powered by aircraft that have never flown, burning fuel to run models that have never slept.
The grid will catch up eventually. It always does. By then, the requirements will have doubled.