The rise of generative AI, powering everything from chatbots like ChatGPT to complex machine learning models, has sent electricity consumption skyrocketing. Data centers, the backbone of AI operations, are projected to consume over 1,000 terawatt-hours globally by 2026, doubling from 460 terawatt-hours in 2022. A single advanced AI chip, like Nvidia’s Blackwell, can draw up to two kilowatts—enough to power a typical household. With tech giants like Microsoft, Google, and Amazon racing to scale AI capabilities, the strain on the electrical grid is unprecedented. Nuclear power, with its high energy density and near-zero carbon emissions, is emerging as a leading solution to meet this demand while aligning with corporate net-zero goals.
Why Nuclear and Why Now?
Nuclear energy offers unmatched reliability compared to intermittent renewables like solar or wind. A single uranium fuel pellet, the size of a pencil eraser, generates as much energy as one ton of coal, making it ideal for the constant, high-capacity needs of data centers. The Paducah facility aims to supply HALEU for small modular reactors (SMRs), which are smaller, factory-built reactors designed for faster deployment and lower costs. Companies like Kairos Power, backed by Google, are developing SMRs to deliver 500 megawatts of carbon-free power by 2030. Microsoft’s deal to restart the Three Mile Island Unit 1 reactor by 2028 further signals the industry’s pivot to nuclear as a stable, low-carbon energy source.
Challenges in the Nuclear Revival
Reviving nuclear power isn’t without hurdles. The Paducah plant requires significant investment to restore its gas centrifuge technology, with initial production expected to take years. The U.S. faces a shortage of skilled nuclear workers, a legacy of plant closures over decades. Uranium mining and enrichment also raise environmental concerns, from groundwater contamination risks in regions like South Texas to the unresolved challenge of storing radioactive waste. No permanent repository exists in the U.S. for long-lived nuclear waste, and critics warn that scaling nuclear capacity could exacerbate these issues. Additionally, producing HALEU increases the risk of nuclear proliferation, as enriched uranium could potentially be diverted for weapons if not tightly regulated.
Big Tech’s Nuclear Push
Tech giants are doubling down on nuclear investments. Google’s agreement with Kairos Power aims to bring SMRs online by 2030, with more by 2035. Meta is seeking bids for nuclear plants to power its data centers, while Amazon and Microsoft have signed deals for nuclear capacity to support their AI ambitions. These moves align with broader policy shifts, including executive orders from the Trump administration to quadruple U.S. nuclear output by 2050 and streamline reactor approvals. The DOE’s recent selection of four federal sites for nuclear-powered AI data centers underscores the public-private momentum. However, critics argue that the high capital costs—$6,417 to $12,681 per kilowatt for nuclear versus $1,290 for natural gas—could slow progress.
Balancing Innovation and Risk
The nuclear-AI synergy offers a path to sustainable energy for tech’s future, but it demands careful navigation. Proponents highlight innovations like in-situ leach mining, which minimizes environmental impact, and SMRs’ passive safety systems that reduce meltdown risks. Yet, skeptics point to historical accidents like Three Mile Island, where a 1979 partial meltdown exposed safety flaws and cost $973 million to clean up. As AI continues to reshape industries, the push for nuclear power must address workforce shortages, waste management, and proliferation risks to deliver on its promise of clean, reliable energy.
