Meta Platforms, the parent company of Facebook, has entered into an agreement to purchase geothermal power from Sage Geosystems, a move aimed at powering its U.S. data centers as the company scales up its energy-intensive artificial intelligence (AI) infrastructure.
The first phase of the 150-megawatt project is expected to be operational by 2027, expanding the use of geothermal energy in the United States. While the exact location of the project has not yet been determined, it will be situated east of the Rocky Mountains. Financial details of the deal were not disclosed.
Geothermal energy, a renewable source that harnesses the Earth’s internal heat to generate electricity and heat water, is becoming an attractive option for tech giants like Meta, which face growing electricity demands due to the rapid adoption of AI technologies. The 150-megawatt output from this project would be sufficient to power approximately 38,000 homes.
Meta’s announcement was made in conjunction with a U.S. Department of Energy event focused on geothermal development, highlighting the Biden administration’s push for large technology companies to invest in clean energy to meet their escalating electricity needs. The administration’s goal of decarbonizing the power sector by 2035 could be complicated by the soaring energy requirements of AI-driven data centers.
Sage Geosystems, a Houston-based startup backed by oil and gas firms Chesapeake Energy and Nabors Industries, as well as venture capital firms Virya and Helium-3 Ventures, is developing next-generation geothermal technology. Unlike traditional geothermal systems that rely on naturally occurring underground hot water reservoirs, Sage’s technology is designed to be deployed in a wider range of locations. This project represents Sage’s largest endeavor to date, with the company having validated its technology in the field just two years ago.
A Meta spokesperson confirmed that the geothermal power from Sage Geosystems would contribute to the overall power grid, rather than directly supplying any specific data center. This agreement comes as Meta has been rapidly overhauling and expanding its infrastructure to support the AI workloads that require specialized chips, leading to a significant rise in the company’s expenses.
In July, Meta projected its capital expenditures for 2024 to range between $37 billion and $40 billion, attributing much of this increase to the infrastructure costs associated with its AI expansion. The company also cautioned that these costs would continue to be a “significant driver” of expense growth into 2025, Reuters news report said.