The AI era is providing the fracking industry with a surprising second act. This is a new chapter for a sector that, even during its early 2010s boom, faced criticism from climate advocates for issues like poisoned water tables, man-made earthquakes, and prolonging the use of fossil fuels. AI companies are now constructing massive data centers near major gas-production sites, often generating their own power by tapping directly into fossil fuels. This trend has been largely overshadowed by news about AI in healthcare and climate solutions, but it is one that could reshape the communities hosting these facilities and raise difficult questions for them.
Consider a recent example. The Wall Street Journal reported that AI startup Poolside is building a data center complex on more than 500 acres in West Texas. This facility, located about 300 miles west of Dallas, has a footprint two-thirds the size of Central Park. It will generate its own electricity by using natural gas from the Permian Basin, the nation’s most productive oil and gas field, where hydraulic fracturing is the standard method of extraction.
This project, named Horizon, will produce two gigawatts of computing power. That is equivalent to the entire electric capacity of the Hoover Dam, but instead of harnessing a river, it burns fracked gas. Poolside is developing the facility with CoreWeave, a cloud computing company that rents access to more than 40,000 Nvidia AI chips. The situation has been aptly described as an energy Wild West.
Yet Poolside is far from alone in this strategy. Nearly all major AI players are pursuing similar paths. Last month, OpenAI CEO Sam Altman toured his company’s flagship Stargate data center in Abilene, Texas. He was candid about the power source, stating they are burning gas to run the data center. According to the Associated Press, the complex requires about 900 megawatts of electricity across eight buildings and includes a new gas-fired power plant using turbines similar to those on warships. The companies state this plant is for backup power, with most electricity coming from the local grid, which draws from a mix of natural gas and West Texas wind and solar farms.
The people living near these projects are not comforted. Arlene Mendler lives across the street from the Stargate site. She told the AP she wishes someone had asked for her opinion before bulldozers cleared a huge tract of mesquite shrubland. She moved to the area 33 years ago seeking peace, quiet, and tranquility. Now, construction noise is the background soundtrack, and bright lights have spoiled her nighttime views.
Water is another major concern. In drought-prone West Texas, locals are nervous about how new data centers will impact the water supply. The city’s reservoirs were at roughly half capacity during Altman’s visit, with residents on a twice-weekly outdoor watering schedule. Oracle claims each building will need only 12,000 gallons of water per year after an initial million-gallon fill for its cooling systems. However, a professor who studies AI’s environmental footprint notes this is misleading, as these systems require more electricity, leading to more indirect water consumption at the power plants generating that electricity.
Meta is also pursuing a similar strategy. In the poorest region of Louisiana, Richland Parish, the company plans to build a ten billion dollar data center the size of 1,700 football fields. It will require two gigawatts of power for computation alone. The utility company Entergy will spend over three billion dollars to build three large natural gas power plants to feed the facility by burning gas extracted through fracking in the nearby Haynesville Shale. Residents in Louisiana, like those in Abilene, are not thrilled to be encircled by constant bulldozing.
Even Elon Musk’s xAI has fracking connections for its controversial Memphis facility. The local utility, which sells power to xAI, purchases natural gas on the spot market. This gas is piped to Memphis via two companies whose pipelines carry natural gas from major hydraulically fractured shale formations.
When asked why AI companies are pursuing this path, the argument often extends beyond electricity to beating China. Chris Lehane, a veteran political operative who is now a vice president at OpenAI, made this case recently. He stated that the United States will need to generate roughly a gigawatt of energy a week and pointed to China’s massive energy buildout of 450 gigawatts and 33 nuclear facilities in just the last year. When questioned about building in economically challenged areas, Lehane returned to geopolitics, suggesting this is an opportunity to re-industrialize countries and modernize energy systems.
The current administration is supportive. A recent executive order fast-tracks gas-powered AI data centers by streamlining environmental permits, offering financial incentives, and opening federal lands for projects using natural gas, coal, or nuclear power, while explicitly excluding renewables from support.
For now, most AI users remain largely unaware of the carbon footprint behind their new tools. They are more focused on capabilities than on where the electricity comes from. The companies have positioned natural gas as the pragmatic answer to AI’s exploding power demands. However, the speed and scale of this fossil fuel buildout deserves more attention.
If this is a bubble, the fallout could be severe. The AI sector has become a circular loop of dependencies, with companies all buying from and selling to each other. If the foundation cracks, a great deal of expensive infrastructure, both digital and gas-burning, could be left standing. One key question is whether all this new capacity is even necessary. A Duke University study found that utilities typically use only 53 percent of their available capacity throughout the year, suggesting significant room to accommodate new demand without constructing new power plants.
The researchers estimate that if data centers reduced electricity consumption by roughly half for just a few hours during annual peak demand periods, utilities could handle an additional 76 gigawatts of new load. This would absorb the 65 gigawatts data centers are projected to need by 2029. This kind of flexibility could allow for faster deployment of AI data centers and provide a reprieve from the rush to build natural gas infrastructure, allowing time to develop cleaner alternatives.
However, the industry argues that such a delay would mean losing ground to an autocratic regime. Consequently, the natural gas building spree appears likely to continue, potentially saddling regions with more fossil-fuel plants and leaving residents with soaring electricity bills to finance today’s investments long after the tech companies’ contracts expire. For instance, Meta has guaranteed it will cover the utility’s costs for new generation in Louisiana for 15 years, and Poolside’s lease runs for 15 years. What happens to customers after these contracts end remains an open question.
Things may eventually change. A lot of private money is being funneled into small modular reactors and solar installations with the expectation that these cleaner alternatives will become central energy sources for data centers. Fusion startups have also raised substantial funding from key AI players. This optimism has spilled into public markets, where energy companies with no current revenue have high market valuations based on the expectation they will one day fuel these data centers.
In the meantime, which could still last for decades, the most pressing concern is that the people who will be left holding the bag, both financially and environmentally, never asked for any of this in the first place.