America’s Nuclear Legacy Is Being Rebuilt for AI

America’s old nuclear weapons sites are quietly being turned into giant AI data center campuses, raising new questions about power, safety, and who really benefits.

Story Snapshot

The Trump administration is opening Cold War-era uranium and weapons sites for massive AI data centers and power plants.
The Department of Energy has named 16 federal sites, many tied to the nuclear weapons program, as candidates for AI infrastructure.
Huge public‑private projects at Paducah, Kentucky, and Portsmouth, Ohio, aim to build some of the world’s largest AI data campuses.
Supporters promise jobs and faster permitting, while critics fear contamination risks, energy costs, and a new wave of elite‑driven mega‑projects.

Trump’s Plan: Turn Nuclear Legacy Sites into AI Hubs

President Trump’s administration is pushing a major plan to reuse old federal nuclear and weapons sites as land for huge artificial intelligence data centers and their power plants. The White House has ordered agencies to speed permits and lean on federally owned land, including brownfield and Superfund sites, to meet soaring demand for computing. These moves fit a broader strategy to treat AI infrastructure as critical for national security and the economy, and to bypass slow local zoning fights.

The Department of Energy has gone beyond general ideas and named 16 specific federal properties as candidates for “AI infrastructure.” Many are tied to the country’s nuclear history, including Los Alamos National Laboratory in New Mexico, the Pantex Plant in Texas, the Kansas City National Security Campus, and the Savannah River Site in South Carolina. An internal memo describes these sites as places where companies could build both data centers and new power plants, including nuclear reactors, on the same ground.

Paducah and Portsmouth: From Uranium for Bombs to Powering AI

In Paducah, Kentucky, the Department of Energy is eyeing parts of a 3,500‑acre former uranium enrichment complex as land for a major AI data center campus. The Paducah Gaseous Diffusion Plant, opened in the 1950s to enrich uranium for nuclear weapons and later reactor fuel, shut down in 2013 and has been in environmental cleanup for years. Federal documents say the site is already designed for up to 3 gigawatts of power and 30 million gallons per day of water, with roads and rail in place to support new industry.

The Department of Energy’s Office of Environmental Management has issued a request for offers from U.S. companies to lease parts of the Paducah site for AI data centers and on‑site energy. Reporting describes plans by major investors like NextEra Energy and Brookfield to turn Paducah into a more than $100 billion AI campus and power complex, blending private capital with federal land. Supporters say this can bring jobs and tax relief to a struggling region, while critics see another massive project where big firms benefit from public assets.

Portsmouth Megaproject: AI, Gas Plants, and Nuclear Symbolism

Southern Ohio is seeing an even larger plan at the Portsmouth Gaseous Diffusion Plant near Piketon, which once enriched uranium for nuclear bombs and reactor fuel during the Cold War. The Department of Energy and the Department of Commerce have announced a partnership with Japan’s SoftBank unit SB Energy and utility AEP Ohio to build what they call the world’s largest AI data center and power campus there. The project targets roughly 10 gigawatts of data center capacity and 9.2 gigawatts of new natural gas generation on federal land.

Local leaders in Pike County have welcomed the project, praising thousands of expected jobs and new tax base after years of economic decline. At the same time, national coverage notes that this development sits on ground that once armed America’s nuclear arsenal, raising questions about contamination, long‑term cleanup, and how safe and fair it is to turn such sites into private technology hubs. Some residents elsewhere fear noise, land use changes, water strain, and a sense that powerful companies are again writing the rules.

Why Federal Nuclear Sites Appeal to AI Developers

Energy and administration officials argue that reusing these federal sites makes sense because they already host heavy industry, grid connections, and large tracts of land away from dense neighborhoods. The Department of Energy says the 16 chosen locations have existing power infrastructure and room for new plants, with federal control allowing faster permits, including for nuclear reactors and gas plants. This approach also tries to address growing public backlash to data centers in suburbs and rural towns that feel overwhelmed by traffic, noise, and land grabs.

What if AI's real bottleneck is angry neighbors?

The overlooked AI infrastructure constraint is not GPUs. It is local permission to consume power.

Gallup reports that 71% of Americans oppose an AI data center near them. Opposition to a local nuclear plant is lower, at 53%. The…

— Nox Vector (@NoxVectorAI) July 29, 2026

Many Americans across the political spectrum worry that this push shows how deeply AI, energy, and federal agencies are now tied to big corporate deals. Conservatives see echoes of past globalist projects and heavy spending. Liberals see more “America First” development that can still widen the gap between elites and everyone else. Both sides share a core concern: decisions about nuclear legacy sites, power, and AI are being made far from the communities that will live with the risks and the bills.

Open Questions: Power, Safety, and Who Wins

Despite the bold announcements, most of these projects are still in early stages, with requests for information, leases, and plans rather than finished data centers and operating power plants. Long wait times for new power plants, grid connections, and transformers could slow the promised AI build‑out and test claims of “fast‑track” approval. Environmental cleanup at sites like Paducah and Portsmouth remains complex; any new contamination finding or delay could fuel public doubt about safety.

For now, the federal government is clearly opening historic nuclear and weapons sites to private AI developers and energy firms, betting that this will keep America ahead in the global AI race. Whether that bet pays off for ordinary citizens—through real jobs, fair power prices, and safe communities—will depend on details that have not yet fully emerged: final contracts, cleanup results, community benefit agreements, and how much say local people get before the bulldozers arrive.