The AI revolution is revitalizing fracking, a surprising development for an industry that faced criticism from climate activists during its boom in the early 2010s for contaminating water sources, triggering earthquakes, and maintaining reliance on fossil fuels.

Companies in the AI sector are establishing large data centers near significant gas extraction locations, frequently producing their own energy by directly utilizing fossil fuels. This trend, often eclipsed by discussions linking AI to healthcare and climate change interventions, could transform and pose challenging dilemmas for the communities that accommodate these operations.

Consider the most recent instance. This week, the Wall Street Journal revealed that Poolside, an AI coding assistant startup, is developing a data center complex spanning over 500 acres in West Texas — situated about 300 miles from Dallas — which is approximately two-thirds the size of Central Park. The center will generate its own power by extracting natural gas from the Permian Basin, the most productive oil and gas field in the US, where hydraulic fracturing is not only prevalent but essentially the sole option available.

The initiative, referred to as Horizon, is set to generate two gigawatts of computational power. This is comparable to the total electrical capacity of the Hoover Dam, except instead of harnessing the Colorado River, it is combusting fracked gas. Poolside is collaborating with CoreWeave, a cloud computing firm that leases access to Nvidia AI chips and is supplying more than 40,000 of them. The Journal describes it as an “energy Wild West,” which seems fitting.

However, Poolside is not alone in this approach. Almost all major players in the AI industry are implementing comparable strategies. Last month, OpenAI CEO Sam Altman visited his company’s main Stargate data center in Abilene, Texas — located roughly 200 miles from the Permian Basin — where he was transparent, stating, “We’re burning gas to run this data center.”

The complex necessitates about 900 megawatts of power across eight buildings and features a new gas-fired power plant utilizing turbines similar to those employed in naval vessels, as reported by the Associated Press. The companies assert that the plant provides only backup power, with the majority of electricity sourced from the local grid. For reference, that grid combines energy from natural gas along with extensive wind and solar farms in West Texas.

Yet, residents living near these developments are not exactly reassured. Arlene Mendler resides directly across from Stargate. In an interview with the AP, she expressed regret that no one consulted her before bulldozers cleared a large area of mesquite shrubland for the construction underway.

“It has completely altered our way of life,” Mendler told the AP. Relocating to the area 33 years ago in search of “peace, quiet, tranquility,” she now finds that construction noise dominates the background, and bright lights have marred her nighttime vistas.

Then there’s the issue of water. In drought-prone West Texas, local residents are particularly worried about the effects of new data centers on water availability. At the time of Altman’s visit, the city’s reservoirs were at about half capacity, and residents were under a twice-weekly outdoor watering schedule. Oracle asserts that each of the eight buildings will use only 12,000 gallons per year after an initial million-gallon reservoir for closed-loop cooling systems. However, Shaolei Ren, a professor at the University of California, Riverside, who studies the environmental impact of AI, informed the AP that this claim is misleading. These systems consume more electricity, implying increased indirect water usage at the power plants supplying that electricity.

Meta is following a similar path. In Richland Parish, the poorest area of Louisiana, the company intends to construct a $10 billion data center equivalent to 1,700 football fields that will demand two gigawatts of power for computation alone. The utility company Entergy will invest $3.2 billion in three large natural-gas plants with 2.3 gigawatts of capacity to supply the facility by combusting gas extracted through fracking in the nearby Haynesville Shale. Louisiana residents, like those in Abilene, are not pleased to be surrounded by continuous construction activity.

(Meta is also expanding in Texas, but in a different part of the state. This week, the company announced a $1.5 billion data center in El Paso, close to the New Mexico border, with one gigawatt of capacity anticipated to be operational by 2028. El Paso is not adjacent to the Permian Basin, and Meta claims the facility will be powered by 100% clean and renewable energy. One point for Meta.)

Even Elon Musk’s xAI, whose facility in Memphis has stirred significant controversy this year, has links to fracking. Memphis Light, Gas and Water — which currently supplies power to xAI but will ultimately own the substations xAI is establishing — procures natural gas on the spot market and transports it to Memphis through two companies: Texas Gas Transmission Corp. and Trunkline Gas Company.

Texas Gas Transmission operates a bidirectional pipeline transporting natural gas from Gulf Coast supply zones and several major hydraulically fractured shale formations through Arkansas, Mississippi, Kentucky, and Tennessee. Trunkline Gas Company, the additional supplier for Memphis, also moves natural gas from fracked sources.

If you’re questioning why AI companies are taking this approach, they’ll tell you it’s not solely about electricity; it’s also about outpacing China.

This argument was articulated by Chris Lehane last week. Lehane, an experienced political operative who joined OpenAI as vice president of global affairs in 2024, presented the case during an interview on stage with TechCrunch.

“We believe that in the near future, at least in the U.S., and indeed globally, we will need to generate around a gigawatt of energy each week,” Lehane stated. He highlighted China’s expansive energy development: 450 gigawatts and 33 nuclear facilities erected in the previous year alone.

When TechCrunch inquired about Stargate’s choice to build in economically challenged locations like Abilene or Lordstown, Ohio, where further gas-powered plants are slated, Lehane returned to the issue of geopolitics. “If we [as a nation] manage this properly, we have an opportunity to re-industrialize nations, bring manufacturing back, and also transition our energy systems to ensure the necessary modernization occurs.”

The Trump administration is undoubtedly supportive of this agenda. The executive order issued in July 2025 accelerates gas-powered AI data centers by simplifying environmental permits, providing financial incentives, and opening federal lands for projects relying on natural gas, coal, or nuclear power — explicitly excluding renewables from support.

At present, most AI users remain largely unaware of the carbon footprint associated with their impressive new tools and resources. They focus more on functionalities like Sora 2 — OpenAI’s highly realistic video generation product that consumes significantly more energy than a simple chatbot — rather than on the sources of the electricity involved.

The companies are banking on this lack of awareness. They’ve presented natural gas as the practical, necessary solution to the increasing energy demands of AI. However, the rapid and extensive expansion of fossil fuel resources requires more scrutiny than it is currently receiving.

If this situation constitutes a bubble, it will not be pretty. The AI industry has transformed into a cyclical reliance on dependencies: OpenAI relies on Microsoft which depends on Nvidia which needs Broadcom which requires Oracle which depends on data center operators who, in turn, need OpenAI. They are all engaged in mutual buying and selling in a self-reinforcing cycle. The Financial Times pointed out this week that if this foundation falters, a significant amount of costly infrastructure, both digital and gas-burning, will be left behind.

OpenAI’s ability to fulfill its obligations is “increasingly a concern for the broader economy,” according to the outlet.

A crucial question that has largely been overlooked is whether all this newly generated capacity is even required. A study from Duke University found that utilities typically only utilize 53% of their available capacity year-round. This indicates substantial capacity for accommodating new demand without needing to construct additional power plants, as noted earlier this year by MIT Technology Review.

The Duke researchers estimate that if data centers cut their electricity use by roughly half for just a few hours during peak demand periods each year, utilities could accommodate an additional 76 gigawatts of new load. This would effectively satisfy the projected 65 gigawatts of demand from data centers by 2029.

Such flexibility would enable companies to establish AI data centers more swiftly. More importantly, it could offer a respite from the rush to construct natural gas facilities, allowing utilities the time to develop more sustainable alternatives.

But again, that would imply conceding ground to an authoritarian regime, according to Lehane and many others in the field, so it appears the current fracking boom is likely to burden regions with additional fossil-fuel plants and leave residents facing escalating electricity bills to cover today’s expenditures, even long after the tech companies’ agreements have lapsed.

For example, Meta has committed to covering Entergy’s expenses for the new Louisiana power sources for 15 years. Similarly, Poolside’s agreement with CoreWeave spans 15 years. The implications for customers once these contracts expire remain uncertain.

Changes may eventually occur. A significant amount of private funding is being directed toward small modular reactors and solar projects, assuming these cleaner energy solutions will become more significant power sources for these data centers. Fusion startups like Helion and Commonwealth Fusion Systems have also secured substantial funding from AI leaders, including Nvidia and Altman.

This positivity isn’t confined to private investment circles. The enthusiasm has extended to public markets, where several “non-revenue-generating” energy firms that have managed to go public boast what appear to be forward-thinking market valuations, based on the belief that they will eventually power these data centers.

In the meantime — which could still span decades — the most pressing concern is that those who will ultimately bear the financial and environmental burdens never requested any of this in the first place.