As the world races to train smarter AI, a new controversy is bubbling beneath the surface—AI data centers and fracked gas are becoming deeply intertwined.
The rapid rise of artificial intelligence is quietly giving the U.S. fracking industry a second life. Tech giants and startups alike are building colossal data centers in gas-rich regions, tapping directly into fossil fuels to power their energy-hungry AI systems.
Fracking Fuels the AI Revolution
According to a new Wall Street Journal report, AI coding startup Poolside is developing a sprawling, 500-acre data center complex in West Texas—dubbed Horizon. The facility, nearly two-thirds the size of Central Park, will produce two gigawatts of computing power, the equivalent of the Hoover Dam’s capacity, by burning fracked gas from the Permian Basin.
The project is backed by CoreWeave, a major cloud provider renting out access to Nvidia GPUs, and signals a growing “energy Wild West” where AI firms generate their own power to meet skyrocketing computational demand.
AI Giants Join the Gas Rush
Poolside isn’t alone. OpenAI, Meta, and xAI are all investing heavily in similar gas-powered infrastructure. OpenAI’s Stargate data center in Abilene, Texas, for example, consumes around 900 megawatts across eight massive buildings—some powered by on-site gas turbines.
Locals like Arlene Mendler say the projects have “completely changed” their communities, with bulldozed shrublands, relentless construction, and nighttime light pollution replacing once-quiet landscapes.
Meanwhile, Meta is constructing a $10 billion data center in Louisiana’s Richland Parish, fueled by three new natural-gas plants. The company says it will later match its operations with clean energy, but environmental advocates remain skeptical.
Communities Bear the Hidden Costs
Beyond the environmental toll, residents worry about water scarcity. Drought-prone West Texas is already struggling to maintain water reservoirs, while AI facilities require millions of gallons for cooling.
Experts like Shaolei Ren of UC Riverside warn that even so-called “closed-loop” systems indirectly consume more water due to the extra electricity needed to operate them.
The Geopolitical Justification
Tech leaders, including OpenAI’s VP of Global Affairs Chris Lehane, argue that fossil-fueled expansion is a matter of national security. In a recent TechCrunch interview, Lehane pointed to China’s rapid energy buildout and said the U.S. must “generate a gigawatt of energy a week” to stay competitive in AI.
The Trump administration’s July 2025 executive order further accelerates this trend, fast-tracking permits for gas, coal, and nuclear-powered AI centers—while excluding renewables from incentives.
The Bigger Question: Is It Even Necessary?
Despite the frenzy, a Duke University study found U.S. utilities typically use only 53% of available capacity, suggesting data centers could expand without new fossil-fuel plants.
If operators reduced power demand by half for just a few hours annually, the grid could handle an extra 76 gigawatts—enough to meet AI’s projected 2029 demand.
For now, however, the AI data centers and fracked gas boom continue to expand, reshaping local economies, straining natural resources, and reigniting debates about technology’s environmental footprint.