Major artificial intelligence (AI) companies are increasingly developing large-scale data centers in the United States, primarily in Texas and Louisiana, that rely on natural gas, often sourced through hydraulic fracturing, to meet their escalating power requirements. This trend signifies a substantial infrastructure investment to support AI computing demands.
This week, the Wall Street Journal reported that AI coding assistant startup Poolside is constructing a data center complex spanning over 500 acres in West Texas. The facility, dubbed "Horizon," is projected to produce two gigawatts of computing power by tapping natural gas from the Permian Basin. Poolside is developing this project in conjunction with CoreWeave, a cloud computing firm that provides access to Nvidia AI chips, supplying over 40,000 units for the facility.
Similarly, OpenAI CEO Sam Altman visited the company's flagship "Stargate" data center in Abilene, Texas, which requires approximately 900 megawatts of electricity. Altman stated, "We're burning gas to run this data center." The complex includes a new gas-fired power plant intended for backup power, though local residents, such as Arlene Mendler, have expressed concerns regarding the environmental and lifestyle impacts of the construction. Shaolei Ren, a University of California, Riverside professor, noted that closed-loop cooling systems for such facilities, while appearing to conserve water directly, necessitate more electricity, leading to increased indirect water consumption at power generation sites.
Meta is also pursuing this strategy, with plans for a $10 billion data center in Richland Parish, Louisiana, which will require two gigawatts of power. Utility company Entergy plans to invest $3.2 billion to construct three new natural-gas power plants with 2.3 gigawatts of capacity to supply this facility, drawing gas from the Haynesville Shale. Separately, Meta announced a $1.5 billion data center in El Paso, Texas, expected online in 2028, which the company states will be matched with 100% clean and renewable energy. Even xAI's facility in Memphis has connections to natural gas pipelines carrying gas from hydraulically fractured sources.
Industry leaders, including OpenAI Vice President of Global Affairs Chris Lehane, attribute this reliance on natural gas to the urgent need for massive energy generation, citing geopolitical competition with China's energy infrastructure buildout. A July 2025 executive order from the Trump administration reportedly accelerates federal permitting for gas-powered AI data centers by streamlining environmental regulations and offering financial incentives, while excluding renewables from support.
The rapid expansion raises questions regarding existing energy capacity and economic implications. A Duke University study indicated that utilities typically utilize only 53% of their available capacity annually. Researchers estimate that if data centers reduced electricity consumption by half for a few hours during peak demand, utilities could accommodate an additional 76 gigawatts, potentially absorbing the 65 gigawatts data centers are projected to need by 2029. Analysts at the Financial Times have noted a self-reinforcing dependency loop within the AI sector, raising concerns about potential vulnerabilities in this infrastructure. Furthermore, long-term contracts, such as Meta's 15-year guarantee to cover Entergy's costs, leave open questions about future electricity rates for residents after these agreements conclude.
Looking ahead, significant private investment is directed toward developing small modular reactors, solar installations, and fusion energy startups like Helion and Commonwealth Fusion Systems, signaling an anticipated shift toward cleaner energy alternatives for data centers in the future. Public markets have also seen several non-revenue-generating energy companies achieve substantial market capitalizations based on expectations of future demand from tech companies.
