Skip to content
The surging power demand of the Artificial Intelligence boom has driven leading tech companies, including Meta, Amazon, Microsoft, and Google, toward nuclear energy, with each announcing major initiatives over the past year. Their motivations are clear: they need reliable, low-emission energy sources to sustain thousands of new and expanding data centers while meeting ambitious climate goals. These efforts often materialize as purchase agreements with existing plants and investments in next-generation nuclear technologies promising safer, more efficient reactors.
Nuclear power is attractive to these companies because it delivers constant, carbon-free electricity, a quality critical for data centers that require uninterrupted 24/7 operation. Yet, a significant mismatch exists between the swift energy needs of Artificial Intelligence infrastructure and the slow pace of nuclear plant development. Constructing new reactors, particularly those based on innovative designs such as small modular technology, can take up to a decade. While data center power consumption in the US could quadruple to 400 terawatt-hours by 2030, existing nuclear facilities have provided a steady, but limited, 800 terawatt-hours annually for two decades with little room for rapid expansion.
Tech firms have launched deals with nuclear startups and established utilities to address these gaps. Google partnered with Kairos Power to purchase up to 500 megawatts by 2035, while Amazon is supporting X-energy´s modular reactor demonstration in Washington State. However, these are pilot efforts unlikely to meet the vast upcoming demand on their own. Some companies, such as Microsoft, are also buying power from older plants being relicensed or reopened, though those too are in finite supply. Experts agree that to substantially influence the Artificial Intelligence sector´s energy use, dozens of new reactors will be required, but these won’t be viable at scale until the 2030s. In the interim, tech companies may resort to fossil fuels to fill the shortfall, potentially compromising their climate targets. The outcome of today’s decisions—whether to back nuclear, renewables, or fossil fuels—will shape the power grid for decades, as no single technology is poised to meet the colossal, rapidly growing energy needs of the Artificial Intelligence revolution alone.
