The global race to power artificial intelligence (AI) has driven major technology companies toward nuclear energy as a potential solution to soaring electricity demands. With data centers and AI chips consuming ever greater amounts of power, firms like Microsoft, Google, and Amazon are searching for reliable, carbon-free sources of electricity. Nuclear energy offers two key advantages—constant, uninterrupted output and zero direct carbon emissions—making it appealing to an industry facing criticism for its growing environmental footprint. Yet despite the excitement, experts caution that nuclear solutions remain far from ready, and public skepticism continues to loom large.
The story of nuclear energy in the United States has long been shaped by the shadow of past accidents. The 1979 partial meltdown at Three Mile Island in Pennsylvania, although far less devastating than Fukushima or Chernobyl, left a deep mark on the national psyche. The timing of the disaster—coinciding with the release of the anti-nuclear thriller The China Syndrome—cemented a negative image of nuclear power for decades. Now, in an ironic twist, Three Mile Island has reemerged in the headlines as Microsoft signs a deal to purchase electricity from its remaining operational reactor. The move signals not just a business decision, but a symbolic shift in how the tech world views nuclear power’s role in a clean energy future.
The enthusiasm extends beyond traditional reactors to newer technologies such as small modular reactors (SMRs). These compact systems, promoted by companies like Google-backed Kairos Power, promise safer operation, faster construction, and potentially lower costs. However, SMRs face substantial hurdles. Most remain at the conceptual or prototype stage, and experts like former U.S. Nuclear Regulatory Commission chair Allison Macfarlane warn that economies of scale still favor large plants. Smaller reactors generate less electricity per unit of fuel, making commercial viability a distant goal. Even if successful, SMRs could take years to contribute meaningfully to the grid, lagging far behind the immediate energy needs of the AI boom.
Meanwhile, public acceptance remains fragile. In North Tonawanda, New York, residents recently voted to ban nuclear development within city limits after a proposal to build a small reactor for cryptocurrency mining triggered backlash. Environmental and safety concerns persist, as do worries about radioactive waste—Stanford researchers note that SMRs may actually produce more waste than conventional reactors due to higher neutron leakage. Still, surveys suggest attitudes may be softening: a Pew Research Center study found that a narrow majority of Americans now favor expanding nuclear energy, reflecting its perceived importance in climate policy and energy independence.
Despite these developments, many scientists argue that innovation within AI itself may be a faster route to sustainability. Researchers like Mosharaf Chowdhury at the University of Michigan are working on reducing the energy footprint of AI models and chips rather than relying solely on new power sources. The International Energy Agency projects that data center electricity use could double within five years, underscoring the urgency of both supply- and demand-side solutions. In the end, even as big tech invests heavily in nuclear energy, industry leaders acknowledge that no single technology—nuclear, renewable, or otherwise—will be enough to meet the colossal energy and decarbonization challenges posed by the AI revolution.
https://www.bbc.com/worklife/article/20251008-why-big-tech-is-going-nuclear