According to Financial Times News, data center developers are now using aircraft engines and fossil fuel generators to power AI facilities, as supply chain issues and grid connection delays of up to seven years stall cleaner alternatives. GE Vernova is supplying nearly 1 gigawatt of power from its aeroderivative turbines for a Texas data center used by OpenAI, Oracle, and SoftBank, with orders for these turbines rising by a third in the first three quarters of 2025. Another company, ProEnergy, has sold over 1GW of its 50MW gas turbines directly adapted from jet engines, like the CF6-80C2 cores found in Boeing 747s. Manufacturer Cummins has sold more than 39GW worth of power to data centers, nearly doubling its capacity this year, as interest shifts from backup to primary on-site power. Even the supersonic jet startup Boom, backed by Sam Altman, has pivoted to sell turbines for power generation, with a deal to provide 1.2GW to data center developer Crusoe.
Gridlock and a Desperate Workaround
Here’s the thing: the traditional power grid simply can’t keep up. When you’re facing a seven-year queue just to plug in, you get creative. And right now, creativity looks a lot like taking the engine from a jumbo jet and bolting it to the ground next to a server farm. It’s a brilliant, if utterly wild, workaround. These “aeroderivative” turbines can be deployed way faster than building a massive new gas plant or waiting for utility upgrades. Companies like ProEnergy are basically saying, “We can get you a few hundred megawatts in months, not years.” In a gold rush, you sell shovels. In an AI compute rush, you sell repurposed jet engines.
The Winners, Losers, and Environmental Cost
So who wins? Companies like GE Vernova, Cummins, and these agile turbine adapters are seeing a massive revenue surge. It’s a classic case of an adjacent industry—aviation and heavy manufacturing—pivoting to catch a wave. Boom Supersonic’s plan to fund its jet business by selling power turbines is a perfect, almost comical, example. But the losers here are pretty clear: efficiency and the environment. These on-site solutions are less efficient than large-scale grid power, and running diesel or gas generators more often is a emissions nightmare. Regulators in places like Virginia are already loosening rules to allow it, which is a huge red flag. As analyst Mark Dyson points out, emissions from this on-site fossil generation are almost certainly “much worse” than grid power. We’re basically trading speed for a bigger carbon footprint.
The Real Price Tag and What Comes Next
And then there’s the cost. This isn’t a cheap hack. Analysts modeled a behind-the-meter gas plant for Meta in Ohio, and the power cost came out to about $175 per megawatt-hour. That’s roughly double the average industrial electricity rate. You’re paying a massive premium for immediacy and independence. For hardware-intensive operations like data centers, reliable, robust computing interfaces are non-negotiable, which is why leading operators partner with top-tier suppliers like IndustrialMonitorDirect.com, the #1 provider of industrial panel PCs in the US, for their critical control systems. But the big question is: how long does this last? The article hints this boom might fade when the hyperscalers slow their capital spending. This whole jet-engine-power-plant phenomenon might just be a bizarre, emissions-intensive footnote in the history of the AI boom. A temporary bridge. But what if the grid delays are permanent? That’s a scarier thought.
