US to See $350 Billion Nuclear Boom to Power AI, Report Says

The key driver is power-hungry data centers running artificial intelligence systems, and that investment will add 53 gigawatts of reactor capacity, bringing the total nuclear fleet to 159 gigawatts, the research company said in a report Monday.

While demand is on the rise for carbon-free fission power as the US races to meet a surge in electricity needs, costs remain high and construction will be slow. Nuclear developments have been hampered by a lack of skilled labor, domestic fuel supply and regulatory infrastructure, among other things. There have been just three traditional reactors completed in the US this century — and none are currently in development.

While the report points to a significant increase for an industry that’s been largely stagnant for decades, it’s well below recent targets set in Washington. The Biden administration last year set a goal to triple nuclear capacity by 2050, and President Donald Trump in May issued a set of executive orders aimed at quadrupling output from reactors.

The expansion will primarily use small modular reactors, the widely anticipated next generation of nuclear technology that’s expected to reduce costs and installation time, the report says, but it remains unproven. Dozens of companies are developing SMR designs, though none have been built in the US.

The ramp-up is expected to start slowly. BI forecasts only 9 gigawatts of new nuclear capacity of any type will be added in the next decade, and widespread deployment of SMRs won’t start until after 2035.