Burning Bits, Bulldozing Plains: How AI’s Carbon Footprint Is Being Built on Fracked Gas
The breathtaking advances in artificial intelligence come with a less glamorous companion: fossil fuels. The latest revelation is stark — much of the AI infrastructure powering your tools and services is being built atop fracked gas and leveled earth. The environmental toll is staring us in the face.
The AI–Fossil Fuel Nexus: What’s Going On
While AI is often framed as a forward-looking, even green technology, beneath the surface lies a much darker story. As reported by TechCrunch, major AI companies — from OpenAI to Meta to rising players like Poolside — are increasingly constructing data center complexes in regions rich in natural gas, especially in Texas and Louisiana. These complexes often generate their own power by tapping directly into fracked gas, bypassing the electrical grid.([TechCrunch][1])
One striking example: Poolside, in partnership with CoreWeave, is building the Horizon project on over 500 acres of West Texas land. It plans to generate 2 gigawatts of computing power — roughly equivalent to the Hoover Dam — by burning fracked gas.([TechCrunch][1]) Meanwhile, OpenAI’s Stargate facility in Abilene is reportedly tied to a dedicated gas-fired power plant (although the company claims it’s just backup power). Residents in the surrounding communities contend with demolished landscapes and mounting disruption to land and water.([TechCrunch][1])
Meta, too, is pursuing a similar model. In Richland Parish, Louisiana, it plans a colossal data center (1,700 football fields in size) powered by multiple natural gas plants — directly feeding on fossil fuel infrastructure.([TechCrunch][1])
Motivations, Consequences, and Trade-offs
Why this push toward gas? AI firms argue that the scale of their computing needs demands fast, reliable energy sources that renewables currently can’t guarantee. They also explain it as a geopolitical imperative — infrastructure built domestically on gas gives the U.S. an edge in the AI arms race.([TechCrunch][1])
However, the consequences are heavy:
- Environmental and public health risks: Fracking has long been linked to water contamination, air pollution, and induced seismic activity. Bulldozing wildlands erases habitats and worsens local ecological degradation.([TechCrunch][1])
- Water stress: West Texas already faces drought. New data centers require massive initial fills of water for cooling, plus ongoing electric demand that increases further indirect water use at power plants.([TechCrunch][1])
- Long-term liabilities: Many of these projects are backed by decade-spanning contracts. When contracts expire, local communities could be left with fossil fuel plants and high utility costs.([TechCrunch][1])
- Overcapacity and inefficiency: A study from Duke University suggests many utility infrastructures are underutilized — roughly 47% of capacity sits idle. That means existing systems may be able to absorb additional demand without new gas plants.([TechCrunch][1])
There is also a policy angle. In July 2025, the U.S. government fast-tracked gas-powered AI data center projects, streamlining permits and incentivizing fossil-based infrastructure — while explicitly excluding support for renewables.([TechCrunch][1])
Humans, Power, and the AI Mirage
The disconnect between AI’s ethereal promise and its grounded costs is growing harder to ignore. Users care about model performance, interface, novelty — not the kilowatt-hours or emissions behind them. But the communities where these operations land invariably bear the burden.
As one West Texas resident put it, she moved seeking peace and light, only to watch bulldozers and all-night glow take over her view.([TechCrunch][1])
While clean energy substitutions like modular reactors, solar, or fusion remain hopeful investments, they’re not yet mature at this scale. In the interim, the AI industry is doubling down on fossil fuel infrastructure — with long-term implications for climate justice, public health, and energy sovereignty.
Glossary
- Fracking (Hydraulic Fracturing): A method of extracting oil or gas by injecting high-pressure fluid into underground rock formations to fracture them and release hydrocarbons.
- Data Center: Large facilities housing computing systems and associated hardware (servers, storage, networking) supporting digital services and AI workloads.
- Backup Power vs. Primary Power: Backup power refers to reserve energy sources used when primary sources fail; primary power is the main consistent supply.
- Closed-Loop Cooling Systems: Cooling systems designed to reuse water, reducing the need to continuously draw new water. But they often require more electricity, which increases indirect water use at power plants.
- Gigawatt (GW): A unit of power equal to one billion watts. For reference, a large hydroelectric dam or nuclear reactor might produce 1 GW.
What This Means Going Forward
The AI-industrial complex is being literally built on fossil fuel foundations. As this momentum continues, pressure must grow on stakeholders to demand accountability and pathways toward cleaner alternatives. Whether through regulation, community resistance, or energy innovation, the current paradigm cannot hold if climate goals and equity are more than buzzwords.
Source: TechCrunch article: “Your AI tools run on fracked gas and bulldozed Texas land” https://techcrunch.com/2025/10/17/your-ai-tools-run-on-fracked-gas-and-bulldozed-texas-land/
| [1]: https://techcrunch.com/2025/10/17/your-ai-tools-run-on-fracked-gas-and-bulldozed-texas-land/ “Your AI tools run on fracked gas and bulldozed Texas land | TechCrunch” |
