The Unseen Energy Beast: How AI is Breathing New Life into Fracking
The Artificial Intelligence (AI) revolution is transforming our world at an astonishing pace. From revolutionizing healthcare to automating complex tasks, AI promises a brighter, more efficient future. But behind the dazzling headlines about sentient chatbots and groundbreaking medical discoveries lies a less-discussed, yet incredibly potent, energy hunger. And in a surprising and controversial twist, this hunger is leading the AI industry to forge an unlikely alliance with an industry many thought was on its last legs: fracking.
For years, hydraulic fracturing, or fracking, has been a lightning rod for environmental activism. Critics have pointed to its role in water contamination, the triggering of seismic activity, and its perceived stubbornness in perpetuating our reliance on fossil fuels. Yet, as AI companies scramble to meet the insatiable demand for computing power, they are increasingly turning to the very energy source that has been so vilest critiqued.
Data Centers: The New Oil Rigs?
Across the United States, particularly in regions rich with natural gas reserves, a new kind of infrastructure is quietly taking root: colossal data centers. These aren’t your average server rooms; they are vast complexes, often sprawling across hundreds of acres, designed to house the thousands of specialized AI chips that power our digital age. And many of these AI giants are opting to generate their own power, often by tapping directly into the abundant fossil fuels beneath their feet – fuels primarily extracted through fracking.
This trend, overshadowed by the more glamorous aspects of AI development, is poised to reshape the landscapes and economies of the communities that host these behemoths. It’s a development that raises critical questions about sustainability, environmental responsibility, and the true cost of our technological progress.
The "Energy Wild West" of West Texas
One of the most striking examples of this phenomenon is unfolding in West Texas. AI coding assistant startup Poolside is constructing a massive data center complex on over 500 acres, a footprint two-thirds the size of New York’s Central Park. This ambitious project, codenamed Horizon, won’t be drawing power from distant renewable farms or the traditional grid. Instead, it will generate its own electricity by tapping directly into the Permian Basin, the nation’s most prolific oil and gas field, where fracking is not just common, but the primary method of extraction.
Horizon is designed to produce a staggering two gigawatts of computing power. To put that into perspective, it’s equivalent to the entire electrical output of the Hoover Dam. But instead of harnessing the mighty Colorado River, it will be fueled by burning fracked natural gas. Poolside is partnering with CoreWeave, a cloud computing company that rents out access to Nvidia’s cutting-edge AI chips, to equip the facility with over 40,000 of these powerful processors.
The Wall Street Journal aptly describes this emerging landscape as an "energy Wild West," and the Poolside project is far from an isolated incident.
OpenAI’s "Burning Gas" Revelation
Many of the biggest names in AI are pursuing similar strategies. Last month, Sam Altman, the CEO of OpenAI, toured his company’s flagship "Stargate" data center in Abilene, Texas. Located a couple of hundred miles from the Permian Basin, Altman was remarkably candid about the facility’s energy source: "We’re burning gas to run this data center." The Stargate complex, a sprawling network of eight buildings, demands approximately 900 megawatts of electricity. While the company claims the on-site gas-fired power plant, utilizing turbines similar to those on warships, serves only as a backup, the primary electricity source is still the local grid.
This grid, it’s important to note, is a mixed bag, drawing power from both natural gas and the extensive wind and solar farms scattered across West Texas. However, for the residents living in the shadow of these colossal AI hubs, the presence of new fossil fuel infrastructure is hardly a comforting thought.
The Human Cost of AI’s Energy Appetite
Arlene Mendler, who lives directly across the street from the Stargate data center, shared her dismay with the Associated Press. She expressed a wish that her opinion had been sought before a vast expanse of mesquite shrubland was bulldozed to make way for the facility. Mendler moved to the area 33 years ago seeking "peace, quiet, tranquility." Now, her serene existence is punctuated by the constant hum of construction and the glare of bright lights that have disrupted her nighttime views.
Water Woes in a Dry Landscape
Beyond the noise and light pollution, a more pressing concern for residents in drought-prone West Texas is the impact these data centers will have on their already strained water supply. During Altman’s visit, the city’s reservoirs were only at about half capacity, leading to residents being placed on a strict twice-weekly outdoor watering schedule.
Oracle, a company involved in data center cooling technology, claims that each of the eight buildings at Stargate will require a modest 12,000 gallons per year, after an initial million-gallon fill for their closed-loop cooling systems. However, Shaolei Ren, a professor at the University of California, Riverside, who specializes in AI’s environmental footprint, cautions that this figure might be misleading. These advanced cooling systems are more electricity-intensive, which indirectly translates to increased water consumption at the power plants generating that electricity.
Meta’s Louisiana Gambit: A Ten-Billion-Dollar Bet on Gas
Meta, the parent company of Facebook and Instagram, is also embracing a similar energy strategy. In Richland Parish, one of the poorest regions in Louisiana, Meta is planning a colossal $10 billion data center complex. This facility, the size of 1,700 football fields, will require a staggering two gigawatts of power just for computation.
To meet this immense demand, the local utility company, Entergy, is investing a substantial $3.2 billion to construct three large natural-gas power plants. These plants, with a combined capacity of 2.3 gigawatts, will burn natural gas extracted through fracking in the nearby Haynesville Shale formation.
Similar to the residents in Abilene, the communities in Louisiana are not exactly thrilled about the prospect of being surrounded by round-the-clock construction and the ongoing operation of new fossil fuel power plants.
(Meta is also expanding its data center footprint in Texas, albeit in El Paso near the New Mexico border. This facility, expected to be operational in 2028, will have a one-gigawatt capacity and Meta claims it will be powered by 100% clean and renewable energy. This is a notable point in Meta’s favor.)
Elon Musk’s xAI and the Pipeline Problem
Even Elon Musk’s AI venture, xAI, has found itself entangled in the fracking energy debate. Its facility in Memphis, Tennessee, has been the subject of significant controversy. While Memphis Light, Gas and Water (MLGW) currently supplies power to xAI, and will eventually own the substations xAI is constructing, it purchases natural gas on the spot market. This gas is piped to Memphis via two companies: Texas Gas Transmission Corp. and Trunkline Gas Company.
Both of these pipeline companies transport natural gas sourced from major hydraulically fractured shale formations, highlighting a pervasive connection between AI energy needs and the fracking industry across different regions.
The Geopolitical Justification: Beating China?
When questioned about the aggressive pursuit of natural gas-powered data centers, AI companies often point to a perceived geopolitical imperative: the need to compete with China.
Chris Lehane, a seasoned political operative who joined OpenAI as vice president of global affairs in 2024, articulated this argument forcefully. He stated that in the near future, the U.S. will require "in the neighborhood of a gigawatt of energy a week" to keep pace with global AI development. He cited China’s massive energy infrastructure buildout, which includes 450 gigawatts and 33 nuclear facilities constructed in the last year alone, as a stark warning.
Lehane argued that by embracing natural gas power for AI, the U.S. has an opportunity to "re-industrialize countries, bring manufacturing back and also transition our energy systems so that we do the modernization that needs to take place."
Government Support for the Fossil Fuel-AI Nexus
This geopolitical argument appears to resonate with some levels of government. An executive order signed in July 2025 aims to fast-track the development of gas-powered AI data centers. It streamlines environmental permits, offers financial incentives, and opens federal lands for projects utilizing natural gas, coal, or nuclear power, while explicitly excluding renewables from such support.
The Unsuspecting Consumer: Ignorance is Not Bliss
For the vast majority of AI users, the energy consumption and its origins remain largely invisible. The focus is on the dazzling capabilities of AI, such as OpenAI’s Sora 2, a hyperrealistic video generation product that demands exponentially more energy than a simple chatbot. The source of the electricity powering these innovations is a secondary concern, if it’s a concern at all.
The AI companies are capitalizing on this disconnect. They have strategically positioned natural gas as the pragmatic, almost inevitable, solution to the escalating power demands of AI. However, the sheer speed and scale of this fossil fuel infrastructure buildout warrants far greater public attention.
A House of Cards: The Interconnected AI Economy
The AI sector has become a complex web of mutual dependencies. OpenAI relies on Microsoft, which relies on Nvidia, which relies on Broadcom, which relies on Oracle, which relies on data center operators, who in turn need AI companies like OpenAI. This self-reinforcing loop creates a potentially fragile ecosystem.
The Financial Times has pointed out that if this foundation falters, it could leave behind a significant amount of expensive, stranded infrastructure – both digital and the fossil-fuel-burning kind. The report highlighted concerns that OpenAI’s ability to meet its growing obligations could be "increasingly a concern for the wider economy."
Is All This New Capacity Truly Necessary?
A crucial question that has been largely absent from the public discourse is whether all this new energy-generating capacity is even essential. Research from Duke University suggests that utilities typically operate at only 53% of their available capacity throughout the year. This indicates a substantial buffer that could accommodate new demand without the immediate need to construct entirely new power plants.
According to MIT Technology Review, Duke researchers estimate that if data centers could reduce their electricity consumption by approximately half during just a few hours of annual peak demand periods, utilities could accommodate an additional 76 gigawatts of new load. This would effectively absorb the projected 65 gigawatts that data centers are expected to require by 2029.
Such flexibility would not only allow AI companies to launch their data centers more rapidly but, more importantly, could provide a much-needed reprieve from the rush to build new natural gas infrastructure. It would grant utilities the time to develop cleaner energy alternatives. Yet, the prevailing sentiment within the industry, driven by the fear of losing ground to autocratic regimes, seems to favor the accelerated construction of natural gas facilities.
The Long-Term Burden on Communities
The consequence of this approach is likely to be regions saddled with more fossil-fuel power plants and residents facing escalating electricity bills. These bills will, in part, be used to finance the investments made today, long after the contracts between tech companies and utilities expire. For instance, Meta has guaranteed it will cover Entergy’s costs for the new Louisiana power generation for 15 years, and Poolside’s lease with CoreWeave is also for 15 years. The long-term financial and environmental burden on these communities remains an open and concerning question.
A Glimmer of Hope: The Quest for Cleaner Energy
Despite the current reliance on fossil fuels, there are signs of a shift towards cleaner energy solutions. Significant private investment is flowing into small modular reactors (SMRs) and advanced solar installations, with the expectation that these technologies will become central to powering data centers in the future. Fusion startups like Helion and Commonwealth Fusion Systems have also attracted substantial funding from key players at the forefront of AI, including Nvidia and Sam Altman himself.
This optimism is not confined to private investment circles. The excitement has permeated public markets, with several nascent energy companies, some still not generating revenue, experiencing significant market capitalization based on the projected future demand from AI data centers.
The Unasked Question: Who Pays the Price?
In the meantime, and this period could still span decades, the most pressing concern is that the very communities being impacted – financially and environmentally – by this AI-driven energy buildout never had a say in the matter. They are the ones left to bear the consequences of a technological revolution whose energy appetite is, for now, inextricably linked to the controversial practice of fracking.
This is a complex issue with no easy answers, but one thing is clear: the hidden energy costs of AI are no longer a secret to be ignored. The future of our planet and the well-being of our communities depend on us confronting these challenges head-on.
Leave a Reply