The AI revolution is giving fracking a renaissance, a surprising turn of events for an industry that, even during its boom in the early 2010s, was criticized by environmentalists for polluted water sources, induced seismic activity, and the continued reliance on fossil fuels.
AI firms are establishing vast data centers near major natural gas production areas, frequently generating their own electricity by directly utilizing fossil fuels. This trend has been overshadowed by news regarding the convergence of AI and healthcare (and addressing climate change), but it has the potential to reshape — and pose complex questions for — the communities that house these facilities.
Consider the most recent illustration. This week, the Wall Street Journal reported that AI coding assistant startup Poolside is building a data center complex on over 500 acres in West Texas — approximately 300 miles west of Dallas — an area equivalent to two-thirds the size of Central Park. The facility will produce its own energy by utilizing natural gas from the Permian Basin, the nation’s most productive oil and gas region, where hydraulic fracturing is not only prevalent but essentially the only option.
The project, known as Horizon, will generate two gigawatts of computing power. This is equivalent to the total electrical capacity of the Hoover Dam, but instead of utilizing the Colorado River, it is burning fracked gas. Poolside is developing the facility with CoreWeave, a cloud computing company that provides access to Nvidia AI chips and is supplying access to over 40,000 of them. The Journal describes it as an “energy Wild West,” which seems fitting.
However, Poolside is not alone. Almost all the major AI players are adopting similar strategies. Last month, OpenAI CEO Sam Altman visited his company’s main Stargate data center in Abilene, Texas — about 200 miles from the Permian Basin — where he openly stated, “We’re burning gas to run this data center.”
According to the Associated Press, the complex needs around 900 megawatts of electricity across eight buildings and includes a new gas-fired power plant with turbines similar to those used to power warships. The companies state that the plant only provides backup power, with the majority of electricity sourced from the local grid. For the record, that grid draws from a combination of natural gas and the vast wind and solar farms in West Texas.
However, the people residing near these projects are not exactly reassured. Arlene Mendler resides across the street from Stargate. She informed the AP that she wishes she had been asked for her opinion before bulldozers removed a large area of mesquite shrubland to make way for what is being built on top of it.
Techcrunch event
San Francisco
|
October 27-29, 2025
“It has completely altered our way of life,” Mendler told the AP. She relocated to the area 33 years ago in search of “peace, quiet, tranquility.” Now, construction serves as the background music, and bright lights on the scene have ruined her nighttime views.
Then there’s the water. In drought-prone West Texas, locals are particularly concerned about the potential impact of new data centers on the water supply. During Altman’s visit, the city’s reservoirs were at approximately half capacity, with residents on a twice-weekly outdoor watering schedule. Oracle asserts that each of the eight buildings will only require 12,000 gallons per year after an initial million-gallon fill for closed-loop cooling systems. However, Shaolei Ren, a University of California, Riverside professor who studies AI’s environmental impact, told the AP that this is misleading. These systems demand more electricity, which translates to greater indirect water consumption at the power plants generating that electricity.
Meta is pursuing a similar approach. In Richland Parish, the poorest region of Louisiana, the company intends to construct a $10 billion data center the size of 1,700 football fields, which will require two gigawatts of power solely for computation. Utility company Entergy will invest $3.2 billion to construct three large natural-gas power plants with 2.3 gigawatts of capacity to supply the facility by burning gas extracted via fracking in the nearby Haynesville Shale. Residents of Louisiana, like those in Abilene, are not excited about being surrounded by bulldozers around the clock.
(Meta is also building in Texas, albeit elsewhere in the state. This week, the company announced a $1.5 billion data center in El Paso, near the New Mexico border, with one gigawatt of capacity expected to be operational in 2028. El Paso is not near 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 Memphis facility has faced considerable controversy this year, has ties to fracking. Memphis Light, Gas and Water — which currently provides power to xAI but will eventually own the substations xAI is constructing — purchases natural gas on the spot market and transports it to Memphis via two companies: Texas Gas Transmission Corp. and Trunkline Gas Company.
Texas Gas Transmission is a bidirectional pipeline transporting natural gas from Gulf Coast supply areas and numerous major hydraulically fractured shale formations through Arkansas, Mississippi, Kentucky, and Tennessee. Trunkline Gas Company, the other Memphis supplier, also transports natural gas from fracked sources.
If you’re wondering why AI firms are taking this approach, they will tell you it’s not just about electricity; it’s also about outcompeting China.
That was the argument made by Chris Lehane last week. Lehane, a seasoned political strategist who joined OpenAI as vice president of global affairs in 2024, outlined the case during an interview on stage with TechCrunch.
“We anticipate that in the near future, at least in the U.S., and indeed globally, we will need to generate approximately a gigawatt of energy each week,” Lehane stated. He highlighted China’s massive energy expansion: 450 gigawatts and 33 nuclear facilities built in the past year alone.
When TechCrunch inquired about Stargate’s decision to build in economically disadvantaged areas such as Abilene, or Lordstown, Ohio, where more gas-powered plants are planned, Lehane reverted to geopolitics. “If we [as a country] execute this effectively, we have the opportunity to re-industrialize nations, bring manufacturing back, and also transform our energy systems to achieve the necessary modernization.”
The Trump administration certainly agrees. The July 2025 executive order expedites gas-powered AI data centers by simplifying environmental permits, providing financial incentives, and opening federal lands for projects utilizing natural gas, coal, or nuclear power — while explicitly excluding renewables from support.
Currently, most AI users remain largely unaware of the carbon footprint behind their impressive new tools for leisure and work. They are more concerned with capabilities like Sora 2 — OpenAI’s hyperrealistic video-generation product, which requires significantly more energy than a basic chatbot — than with the origin of the electricity.
The companies are relying on this. They have presented natural gas as the practical, unavoidable solution to AI’s escalating power demands. However, the speed and scale of this fossil fuel development warrant greater attention than they are receiving.
If this is a bubble, the consequences will be severe. The AI sector has become a self-destructive system of dependencies: OpenAI requires Microsoft, which requires Nvidia, which requires Broadcom, which requires Oracle, which requires data center operators, who in turn require OpenAI. They are all engaging in transactions with one another in a self-perpetuating cycle. The Financial Times noted this week that if the foundation collapses, there will be a substantial amount of expensive infrastructure left unused, both in digital form and in the form of gas-burning facilities.
The outlet stated that OpenAI’s ability to fulfill its obligations alone is “increasingly a concern for the broader economy.”
A crucial question that has been largely overlooked in the discussion is whether all this new capacity is even necessary. A Duke University study discovered that utilities typically only utilize 53% of their available capacity throughout the year. According to a report earlier this year by MIT Technology Review, this suggests substantial potential to accommodate new demand without constructing new power plants.
The Duke researchers estimate that if data centers reduced electricity consumption by approximately half for a few hours during annual peak demand periods, utilities could manage an additional 76 gigawatts of new load. This would effectively absorb the 65 gigawatts that data centers are projected to require by 2029.
Such flexibility would enable companies to launch AI data centers more rapidly. More importantly, it could provide a respite from the urgency to construct natural gas infrastructure, allowing utilities time to develop cleaner alternatives.
However, according to Lehane and numerous others in the industry, this would mean ceding ground to an autocratic regime, so instead, the natural gas construction spree appears likely to burden regions with more fossil-fuel plants and leave residents with escalating electricity bills to finance current investments, even long after the tech companies’ contracts expire.
For example, Meta has guaranteed that it will cover Entergy’s costs for the new generation infrastructure in Louisiana for 15 years. Poolside’s lease with CoreWeave is for 15 years. What happens to customers when those contracts expire remains an unresolved issue.
Things may eventually evolve. Significant private capital is being invested in small modular reactors and solar installations with the expectation that these cleaner energy alternatives will become more prevalent energy sources for these data centers. Fusion startups such as Helion and Commonwealth Fusion Systems have also secured substantial funding from those at the forefront of AI, including Nvidia and Altman.
This optimism extends beyond private investment circles. The enthusiasm has spread to public markets, where several “non-revenue-generating” energy companies that have successfully gone public possess truly anticipatory market capitalizations, based on the expectation that they will eventually power these data centers.
In the meantime — which could still span decades — the most pressing concern is that the people who will be left to bear the consequences, both financially and environmentally, never requested any of this in the first place.
