Meta goes nuclear, securing 6.6 gigawatts of power to fuel AI superintelligence

Company becomes leading corporate buyer of nuclear energy, striking landmark deals with Vistra, TerraPower, and Oklo.

Meta Platforms, Inc. has executed a series of landmark agreements with nuclear energy companies that will provide up to 6.6 gigawatts (GW) of new and existing carbon-free power by 2035. The move immediately establishes Meta as one of the largest corporate purchasers of nuclear energy in U.S. history and signals the tech industry’s growing reliance on reliable, high-density power sources to meet the insatiable energy demands of generative AI.

The contracts are essential to fueling Meta's ambitious AI strategy, including the Prometheus supercluster - a 1-gigawatt data center being built in New Albany, Ohio, and its longer-term goal of achieving artificial superintelligence.

The trio of nuclear partnerships

To meet the vast and immediate power needs of its rapidly expanding data center footprint, Meta has finalized deals with three distinct nuclear partners:

• Vistra Corp. (Existing Power)

• Meta signed 20-year Power Purchase Agreements (PPAs) to secure over 2.6 GW of energy from Vistra's existing nuclear power plants in Ohio (Davis-Besse and Perry) and Pennsylvania (Beaver Valley).

Crucially, the funding helps support the extension of the operational licenses for these existing plants for another 20 years, guaranteeing immediate, reliable power.

TerraPower (New Reactor Development)

Meta is providing funding to support the development of Bill Gates-backed TerraPower's advanced Natrium reactor units. The agreement covers the development of two initial units, capable of generating up to 690 megawatts (MW) by 2032, and grants Meta rights to power from up to six additional units by 2035.

Oklo (Advanced Nuclear)

Meta has committed to support the development of Sam Altman-backed Oklo’s 1.2 GW advanced nuclear technology campus in Pike County, Ohio, with the first reactors slated to become operational as early as 2030.

The AI power challenge: Why nuclear?

The shift toward nuclear power underscores the energy crisis facing hyperscalers as they race to build AI infrastructure faster than the traditional grid can expand.

• Firm and reliable: Unlike intermittent renewables (solar and wind), nuclear provides "firm" or baseload power 24/7. This reliability is critical for the continuous, non-stop training and inference required by LLMs.

• Decarbonization goal: Meta has committed to matching 100% of the electricity used in its data centers with clean and renewable energy. As AI models surge energy use, nuclear offers a carbon-free solution for the massive amounts of power needed, helping the company bridge the gap to its net-zero goals.

• Scale and density: A single gigawatt of AI capacity requires enough energy to power roughly 750,000 homes. By investing in nuclear, Meta is securing city-sized amounts of power at a single, relatively compact location near its major data center clusters in the Midwest.

The unprecedented scale of these nuclear agreements signals a broader trend where Big Tech is now acting as a primary catalyst and financial backer for the next generation of U.S. power infrastructure.