Nuclear Sites Reborn as AI Empires

Washington is quietly turning Cold War nuclear weapons sites into massive AI hubs, raising fresh questions about who really benefits when federal land and power are handed to Big Tech.

Story Snapshot

The Trump administration is opening old nuclear-related federal sites for huge AI data centers and power plants.
The Department of Energy has named 16 weapons, nuclear, and lab sites as candidates for fast-track AI infrastructure.
Projects in Kentucky and Ohio would turn uranium enrichment plants into multi-gigawatt, privately run AI campuses.
Supporters promise jobs and lower bills, while critics see another elite-driven deal on contaminated, symbolic land.

Trump’s plan: AI data centers on nuclear-era federal land

President Trump’s administration is pushing a plan to use federal land, including former nuclear weapons and uranium sites, for giant artificial intelligence data centers and their power plants. The White House has backed orders that speed permits on brownfield and Superfund land and tell agencies to allow data centers on appropriate federal sites. This is sold as a way to move fast, sidestep local fights over land use, and use land that Washington already controls for “prosperity and security.”

The Department of Energy has gone beyond talk and built a list of 16 federally owned sites that could host these projects. The list includes places tied directly to America’s nuclear history, such as the Paducah and Portsmouth gaseous diffusion plants, Los Alamos National Laboratory, and the Pantex Plant that assembles and disassembles nuclear weapons. An internal memo and public request for information ask developers to propose AI campuses with on-site power, including nuclear reactors, with the aim of opening by 2027.

Paducah and Portsmouth: from uranium for bombs to power for AI

In Paducah, Kentucky, the Energy Department has identified a 3,500-acre uranium enrichment site, shut down in 2013, as a candidate for a major AI campus. This Cold War-era facility once produced low-enriched uranium for reactors and weapons and is now under environmental cleanup. The site has heavy infrastructure already in place, including roads, rail, and access to a river, and is designed for up to 3 gigawatts of power and tens of millions of gallons of water per day.

Reporting describes a proposed $100 billion project by energy and investment firms to turn Paducah into a combined data center and power hub. Supporters say this would reuse an existing industrial footprint instead of building on untouched land, and promise construction jobs and long-term operations in a region that has lost manufacturing work. But the cleanup is ongoing, and the record so far is about proposals and land offers, not finished buildings or confirmed job numbers.

Piketon, labs, and weapons plants: scaling up AI and energy

In Piketon, Ohio, the former Portsmouth Gaseous Diffusion Plant is set to become one of the largest AI and power campuses in the world. This site enriched uranium for nuclear weapons and power plants during the Cold War and sits on thousands of acres of federal land. The Department of Energy and the Department of Commerce announced a partnership with SoftBank’s energy arm and a utility to build a 10-gigawatt AI data center there, powered mainly by new natural gas plants on site.

The project’s backers frame it as a model for using idle federal weapons land to meet soaring AI power demand and revive local economies. Pike County leaders, facing job losses and low tax bases, have openly welcomed the campus as a path to growth. Yet even here, where local resistance is lower than in some suburbs, people elsewhere question whether making private tech campuses on nuclear-era public land mainly serves citizens or deep-pocketed companies.

Why this hits shared nerves about power, elites, and risk

Many Americans see this federal land push through a broader lens of frustration with Washington and large corporations. Conservatives who distrust “woke” policies and global institutions also worry that elites are cutting special deals for tech giants on land tied to national security. Liberals who focus on inequality and environmental justice fear communities near old nuclear sites are being asked to accept new risks so wealthy firms can secure cheap land and power.

The Department of Energy itself highlights that these sites can support both data centers and new power plants, including nuclear, and that federal control can speed permits. That may help avoid some “not in my backyard” battles that have stalled projects on private land. At the same time, using weapons plants and contaminated sites creates powerful symbolism: land once used for bombs and secret labs now leased to corporations to train advanced AI systems, with long-term energy and water demands that local residents must live with.

Promises, unknowns, and what comes next

Supporters of Trump’s strategy promise several gains: faster AI deployment, more reliable power grids, high-paying jobs, and lower property taxes in struggling counties. The administration has also moved to label AI data centers at Department of Energy sites, and the reactors that power them, as defense-critical infrastructure, tying them directly to national security. For those worried about China or other rivals in AI, this framing makes the projects feel not just commercial but strategic.

Trump Administration Converts Old Nuclear Sites into AI Data Centers https://t.co/2C9v0v9AxG

— Policy Wire (@policy_wire) July 30, 2026

But key questions remain. Many plans still depend on future power plants, long grid connections, and ongoing nuclear cleanup, all of which can take years and face delays. The evidence so far focuses on land selection, partnership announcements, and early permitting, not on completed campuses or proven local benefits. As work moves ahead at Paducah, Portsmouth, and other sites, Americans across the political spectrum will be watching to see whether this bold reuse of nuclear-era land delivers for ordinary people—or mostly for the same elites they already mistrust.