AI's Hunger for Power Is Reviving Natural Gas
AI needs electricity faster than clean sources can supply it, and natural gas is filling the gap. Data-center demand is driving a wave of new gas power plants in Texas and Pennsylvania, and the rush is so intense that prices for the turbines they need are up nearly 200 percent over 2019, with waitlists stretching into the 2030s.
AI needs electricity faster than clean sources can supply it, and natural gas is filling the gap. Data-center demand is driving a wave of new gas power plants in Texas and Pennsylvania, and the rush is so intense that prices for the turbines they need are up nearly 200 percent over 2019, with waitlists stretching into the 2030s. The AI build-out is becoming an energy build-out.
The scale of the demand is hard to overstate. Global electricity use from data centers is set to more than double over the next five years, consuming about as much power by 2030 as all of Japan does today. AI workloads run around the clock and cannot wait for the wind to blow, so developers want power that is always on, and gas is the fastest way to get large amounts of it built.
That is why gas is gaining share. Natural gas as a portion of planned new capacity has jumped from about 11 percent in 2024 to over 18 percent in 2026, and data-center gas consumption is forecast to keep climbing through the decade. After years of being framed as a fuel on the way out, gas is suddenly central to the most important growth story in tech.
The bottleneck has shifted to hardware. The constraint is no longer just generating power but building the physical plants, with gas-turbine prices up nearly 200 percent and order books full for years. Costs for new gas plants have surged, and that feeds straight into the price of electricity and the economics of every data center being planned. The shortage is in the machines, not the gas.
The honest tension is environmental and cyclical. Leaning on gas to power AI raises emissions just as the world targets cuts, and it ties the AI boom to a fossil fuel and its price swings. If AI demand ever cools, some of these long-lived plants and contracts could look overbuilt. Building for a boom always carries the risk the boom slows.
So the cleanest-sounding industry in the world is running on one of the oldest fuels, because power is the real constraint on AI. A surge in gas plants, turbine prices up nearly 200 percent, and demand set to double. Electricity, not algorithms, may decide how fast AI can grow. Watch turbine supply and power prices for the next squeeze.