Nuclear Renaissance Accelerates as AI Power Demands Resurrect Dead Projects

According to Utility Dive, major nuclear projects once considered permanently abandoned are being revived to meet surging power demands from artificial intelligence and data centers. In South Carolina, Brookfield Asset Management has signed a letter of intent with Santee Cooper to potentially restart construction on the V.C. Summer nuclear expansion, which was originally proposed in 2008 for under $12 billion but abandoned in 2017 when costs threatened to reach $25 billion. The partnership establishes a six-week feasibility period to evaluate resuming construction of two nuclear units using Westinghouse AP1000 technology. Meanwhile, Google and NextEra announced they expect to bring Iowa’s Duane Arnold nuclear plant back online by early 2029, with NextEra having received Federal Energy Regulatory Commission approval in August to restart the facility it shuttered in 2020. These developments come as the U.S. Department of Energy pushes for accelerated grid interconnections, creating a landscape where resurrecting existing nuclear infrastructure may become the fastest path to meeting AI’s massive energy requirements.

The Unprecedented Scale of AI’s Energy Appetite

What makes these nuclear restarts particularly significant is the sheer magnitude of power required by next-generation AI systems. While data centers have always been energy-intensive, modern AI training and inference workloads operate on an entirely different scale. A single large language model training run can consume more electricity than 100 U.S. homes use in an entire year, and companies like Google are running thousands of these operations simultaneously. The artificial intelligence revolution isn’t just creating incremental power demand—it’s creating step-function changes that traditional energy planning cycles simply can’t accommodate. Nuclear power, with its high capacity factors and 24/7 reliability, offers the only currently scalable solution that matches AI’s constant, massive energy requirements without corresponding carbon emissions.

The Regulatory Landscape Transformation

The Department of Energy’s recent directive to FERC represents a fundamental shift in energy policy thinking. For decades, the focus has been on managing declining or stable electricity demand through efficiency improvements and renewable integration. Now, regulators are scrambling to accommodate what could be the largest surge in electricity demand since the advent of air conditioning. The Federal Energy Regulatory Commission faces the challenge of balancing rapid interconnection needs with grid reliability concerns. This creates a unique opportunity for nuclear assets—both existing and partially constructed—because they can deliver massive baseload power without the intermittency issues that plague renewable sources when trying to serve constant, high-density AI workloads.

The Changed Economic Calculus for Nuclear

What’s fundamentally different today versus when these projects were originally abandoned is the economics of power purchase agreements. AI companies aren’t just looking for cheap power—they’re looking for reliable, carbon-free power at almost any price. Google’s partnership with NextEra likely involves long-term contracts at premium rates that make nuclear economics work in ways they didn’t just five years ago. According to Google’s announcement, restarting operational plants represents “the fastest path to unlock large-scale nuclear power to meet AI growth in the near-term.” This reflects a strategic calculation that the opportunity cost of delayed AI development outweighs the premium for guaranteed nuclear power. For Santee Cooper and Brookfield, maintaining the V.C. Summer site equipment for eight years now appears prescient rather than wasteful, as the preserved infrastructure provides a multi-year head start on new construction.

The Execution Challenges Ahead

While the announcements are promising, the history of nuclear construction in the U.S. suggests significant execution risks remain. The V.C. Summer project’s original failure wasn’t due to technology problems but construction management and cost control issues. The same electric generator technology that makes these assets attractive—Westinghouse’s AP1000—was involved in both the Vogtle expansion that faced massive cost overruns and the original V.C. Summer cancellation. Brookfield’s six-week feasibility assessment, as detailed in Santee Cooper’s announcement, will need to thoroughly evaluate whether supply chain, labor, and regulatory conditions have materially improved enough to avoid repeating past mistakes. Similarly, NextEra’s planned restart of Duane Arnold faces challenges around recommissioning equipment that has been offline for years and securing the specialized nuclear workforce needed for safe operations.

Broader Market and Technology Implications

These nuclear restarts could catalyze a broader re-evaluation of energy infrastructure across the country. If successful, they demonstrate that existing nuclear sites—even those considered permanently closed—represent valuable strategic assets in the AI era. This could lead to similar efforts at other shuttered nuclear facilities and potentially change the calculus for nuclear plant retirements scheduled in the coming years. The deals also create a new template for technology-energy partnerships, where large tech companies essentially underwrite the revival of capital-intensive energy projects in exchange for guaranteed, clean power. As AI continues to drive unprecedented electricity demand growth, we’re likely to see more creative partnerships between technology giants and energy providers, fundamentally reshaping how large-scale power projects get financed and developed in the United States.