How AI is giving nuclear power a second chance

MPower is a small modular nuclear reactor project that was once considered a promising source of affordable nuclear energy, The Washington Post reports. 

About $400 million was invested in the project before BWXT ended development in 2017, as cheap natural gas made nuclear power less competitive and the Fukushima disaster reduced interest in new nuclear plants. 

The 120-foot-tall test facility outside of Lynchburg, Virginia, then sat unused for nearly a decade. Now, the rapid growth of artificial intelligence and data centers has created greater demand for reliable, low-emissions electricity, giving the technology a new opportunity. 

BWXT is partnering with Applied Atomics, a startup founded by former SpaceX engineers, to pursue the project. Unlike some companies developing more experimental reactor designs, mPower uses familiar light-water reactor technology similar to that used in existing nuclear plants. 

Applied Atomics hopes its first 195-megawatt reactor could eventually provide electricity to data centers and other industrial facilities, while the companies are also exploring floating nuclear plants.

However, significant challenges remain, including construction costs, regulatory approval, radioactive waste storage and safety concerns. Critics question whether small reactors can overcome the nuclear industry’s history of delays and cost overruns. The developers believe the main challenge is not the basic nuclear technology but building reactors efficiently and economically. 

The Washington Post has the full story. A subscription may be required.