The Trump administration announced Tuesday a $17.5 billion conditional loan program to accelerate orders for 10 Westinghouse AP1000 nuclear reactors, aiming to meet surging U.S. electricity demand driven by artificial intelligence and data centers. The Energy Department said the funding would support up to five projects, each deploying two of the 1.1-gigawatt reactors, with the goal of cutting deployment timelines by up to three years.
Energy Secretary Chris Wright told reporters that the loans have drawn strong interest from data-center "hyperscalers," utilities, and energy firms. The program targets long-lead equipment such as reactor vessels and steam generators, which can take years to build and ship, allowing construction to begin sooner. The department said this approach could shave up to three years off the typical timeline for nuclear projects.
U.S. power demand is rising after a long period of flat growth, largely due to the rapid expansion of data centers powering cloud and AI services. The Electric Power Research Institute estimates that U.S. data centers could consume up to 9% of total electricity by 2030, more than double their estimated 4% share in 2023. A separate analysis backed by Lawrence Berkeley National Laboratory puts the 2030 range between 9.5% and 15%.
The loans are not full construction financing but are intended to cover early-stage costs. Each project requires Westinghouse and its partner to commit $500 million in equity—$1 billion per project—before accessing federal funds. Westinghouse has signed letters of intent with seven potential partners, all of whom have selected sites, though the department did not name them.
The program deepens the federal government's ties to Westinghouse, which is owned by Brookfield Asset Management and Cameco. The loan plan follows an $80 billion deal in which the government helps secure financing and permits for Westinghouse reactors in exchange for a share of future profits. Brookfield CEO Connor Teskey called the loans a "catalyst for nuclear."
The move comes after costly delays at Georgia Power's Plant Vogtle, where new units finished years behind schedule and billions over budget. Wright said batch building at multiple sites could help establish a supply chain and construction expertise that Vogtle lacked. Only two large reactors have been built from scratch in the U.S. in recent decades before the current push.
Westinghouse is not alone in chasing an AI-driven nuclear surge. TerraPower, Kairos Power, and X-energy are advancing small modular reactor projects with federal cost-sharing, and have signed supply or offtake deals with Google, Meta, the Tennessee Valley Authority, and Dow Chemical. The new DOE loan program, however, only covers large conventional reactors, not small modular ones.
Risks remain significant. Nuclear projects require heavy upfront spending, slow permits, and complex supply chains. Delays could hit utilities, ratepayers, or taxpayers hard if demand changes or costs escalate. Travis Fisher of the Cato Institute criticized the government for picking winners, arguing that firms should build plants that "pass the market test."
Policy hurdles persist. The DOE said its conditional commitment signals intent to lend, but projects must still meet technical, legal, environmental, and financial conditions before funds are released. The department did not provide a timeline for selecting the five projects. The Trump administration aims to start 10 large reactors by 2030 and reach 400 gigawatts of nuclear capacity by 2050, but the pressure is on utilities and data-center firms to order, approve, and complete these reactors before the next AI-driven leap in power demand.
