The Department of Energy has announced a partnership to construct an AI data center combined with energy infrastructure at the former Paducah Gaseous Diffusion Plant in Kentucky, marking a significant shift in how the U.S. is approaching the massive power demands of artificial intelligence. This project represents a novel strategy: repurposing decades-old federal nuclear facilities to host the computing clusters that train and run modern AI systems.

What Is the Paducah Data Center Partnership?

The DOE announced a partnership with Big Rivers Electric Power Corporation, Brookfield Corporation, Jackson Purchase Energy Cooperative, NextEra Energy, and Paducah Power System to redevelop parts of the former Paducah Gaseous Diffusion Plant into a data center campus with integrated energy infrastructure. The Paducah site, which operated as a uranium enrichment facility during the Cold War, sits on land with existing electrical infrastructure and grid connections that make it attractive for power-hungry computing operations. This is the first concrete deployment of an AI data center at a federal nuclear site, signaling that policymakers view AI infrastructure as critical national infrastructure worthy of federal investment and coordination.

Why Are Tech Companies and the DOE Targeting Nuclear Sites for AI?

These partnerships address a fundamental challenge facing the AI industry: data centers require enormous amounts of reliable, continuous electrical power. Training large language models and running inference at scale can consume as much electricity as a small city. Nuclear sites offer several distinct advantages that make them attractive for this purpose. First, these locations already have substantial electrical infrastructure built over decades of operation. Second, they sit on federal land with fewer zoning restrictions than private sites. Third, the DOE's involvement signals potential access to advanced nuclear power generation, which can provide carbon-free baseload electricity without the intermittency challenges of renewables.

The timing reflects growing recognition that AI's power demands cannot be met by traditional grid expansion alone. As AI companies race to build larger models and deploy them at scale, electricity availability and cost have become competitive advantages. Companies that can secure reliable, affordable power gain a substantial edge in the race to develop and deploy advanced AI systems.

How Are These Projects Structured?

• Public-Private Partnerships: The DOE is not building or operating these data centers directly. Instead, it is partnering with established energy companies and technology firms to share infrastructure, risk, and operational expertise.
• Energy Integration: Unlike traditional data centers that purchase power from the grid, these projects combine computing facilities with on-site or nearby power generation, reducing transmission losses and improving reliability.
• Site Remediation and Economic Benefits: Repurposing former nuclear weapons production sites creates economic activity and employment at locations that have faced decades of cleanup and decommissioning costs.

The Paducah project involves multiple utilities and energy cooperatives, suggesting a regional approach to power distribution and grid management. This structure allows the DOE to leverage private sector expertise in data center operations while maintaining federal oversight of nuclear facilities and environmental compliance.

What Other Nuclear Projects Are Moving Forward?

Beyond the Paducah data center, the nuclear sector is experiencing broader momentum. NextEra Energy is pursuing the restart of the Duane Arnold nuclear power plant in Palo, Iowa, as early as 2029, with the Nuclear Regulatory Commission posting a draft environmental assessment and finding of no significant impact on August 4, 2026. The Department of Energy's Office of Energy Dominance Financing is a cooperating agency on this restart project, considering providing financial assistance. Additionally, ARC Clean Technology has entered into a Strategic Partnership Project agreement with Battelle Energy Alliance, the management and operating contractor of Idaho National Laboratory, to collaborate on the design, demonstration, and first deployment of the company's ARC-100, a 100-megawatt sodium-cooled fast reactor. This partnership focuses on advanced reactor technology development rather than data center construction, but reflects the broader momentum in nuclear deployment.

The convergence of these projects suggests that policymakers and private companies view nuclear power as essential infrastructure for supporting AI development. The strategy signals a shift from the traditional model where nuclear plants primarily served utility companies and the grid. If these partnerships succeed, they could accelerate similar projects at other DOE sites and encourage private investment in advanced reactor technologies designed specifically for data center applications.