The thesis
For two decades, advanced nuclear in the United States was a story of designs on paper. TerraPower is the company that turned one of those designs into poured concrete. On March 4, 2026 the Nuclear Regulatory Commission issued Kemmerer Unit 1 the first construction permit for a commercial-scale advanced reactor in the country — and the first for a non-light-water power reactor in more than four decades — and on April 23, 2026 TerraPower and contractor Bechtel broke ground on the nuclear island in rural Wyoming. Whatever happens next, that milestone is real in a way the sector's usual renderings and roadmaps are not: TerraPower is now building, not just proposing.
The bet underneath the steel is that the grid needs a nuclear plant that behaves less like a nuclear plant. TerraPower's Natrium design pairs a sodium-cooled fast reactor with a molten-salt heat store, so the reactor can run flat-out at 345 megawatts while the plant dispatches anywhere up to 500 megawatts on demand — a machine engineered to ramp alongside wind and solar rather than fight them. In an era when data centers and electrification are driving the sharpest surge in firm-power demand in a generation, that flexibility is the pitch. The company is founded by Bill Gates, backed by the U.S. Department of Energy, and, as of 2026, further along toward commercial operation than almost any advanced-reactor rival on Earth.
From Bill Gates's invention lab to a sodium reactor
TerraPower grew out of Intellectual Ventures, the Bellevue, Washington invention firm co-founded by former Microsoft CTO Nathan Myhrvold, and was established as a standalone company in 2008 with Bill Gates as chairman and principal backer. Its first idea was audacious: the traveling wave reactor, a design that would burn depleted uranium — nuclear waste, essentially — in a slow-moving wave of fission, promising decades of operation on fuel nobody else wanted. The concept drew serious money and serious engineering talent, and even a 2015 development agreement with China National Nuclear Corporation, but it never reached a shovel.
The pivot that mattered came in 2020, when TerraPower and GE Hitachi Nuclear Energy unveiled Natrium, a far more pragmatic machine: a 345-megawatt sodium-cooled fast reactor mated to a molten-salt storage system for over five hours of dispatchable output. Where the traveling wave reactor was a moonshot, Natrium was buildable — sodium fast reactors have real operating history, and the storage tank borrowed proven technology from concentrated solar plants. Later that year the Department of Energy chose Natrium as one of two winners of its Advanced Reactor Demonstration Program, committing to a roughly 50/50 cost share, and in 2022 TerraPower selected the retiring Naughton coal plant near Kemmerer, Wyoming as the site — a deliberate choice to put a clean-energy future on the ground of a fading fossil-fuel town.
The technology: a fast reactor wired to a battery
Natrium's defining trick is the decoupling of heat generation from power delivery. The reactor core produces a steady 345 megawatts of electrical output, but instead of sending that heat straight to a turbine, the plant can route it into a bank of molten nitrate salt. When demand and prices spike, the stored heat is drawn down to drive the turbine harder, boosting output to 500 megawatts for several hours. That lets a single reactor act as both baseload generator and peaking plant — a profile conventional light-water reactors, which run best at a constant clip, simply cannot match. Because the coolant is liquid sodium at near-atmospheric pressure rather than high-pressure water, the design also sheds much of the heavy containment and safety plumbing that makes traditional plants so expensive to build.
The catch is the fuel. Natrium runs on high-assay low-enriched uranium, or HALEU, enriched to just under 20 percent — far richer than the roughly 5 percent fuel today's reactor fleet uses, and a grade almost no one in the West produces at commercial scale. That single dependency has shaped TerraPower's trajectory more than any engineering choice, and it turned a fuel-supply question into the project's central strategic risk.
The HALEU problem that cost two years
When TerraPower laid out its original schedule, it assumed it could buy HALEU from the world's only commercial supplier: Russia. Then Russia invaded Ukraine in February 2022, and that assumption collapsed. By late 2022 the company was forced to announce that its demonstration plant would slip at least two years — from a 2028 target toward 2030 — not because of any technical failure, but because the fuel to run it no longer had a viable source. It was a stark illustration of how a single supply-chain node can hold an entire clean-energy project hostage.
TerraPower's response was to help build a domestic HALEU industry more or less from scratch, spreading its bets across multiple suppliers — Centrus Energy, Framatome, GNF-A, and ASP Isotopes among them — and leaning on the Department of Energy's HALEU Availability Program, which named the company among its first allocation recipients. The partnership with Framatome reached a milestone when a metallization line at Framatome's Richland, Washington facility produced the metallic uranium 'pucks' that Natrium's fuel requires, proving out a step in the domestic supply chain. Securing enough qualified HALEU on schedule remains one of the harder problems standing between the Kemmerer plant and first criticality.
From one demonstration to a commercial fleet
Through 2025, TerraPower's public story was essentially a single plant in Wyoming. In 2026 it became a fleet strategy. In January, the company signed a landmark agreement with Meta to develop up to eight Natrium plants in the United States — as much as 2.8 gigawatts of baseload power, boostable toward 4 gigawatts with storage — with Meta funding early development of two units, holding rights to six more, and first units targeted as early as 2032. It is one of the clearest signals yet that hyperscale technology companies, not traditional utilities, may be the customers who finally pull advanced nuclear into the market.
To industrialize that fleet, TerraPower turned to Korea. In May 2026 it signed commercialization agreements with HD Hyundai and Hyundai Engineering & Construction: HD Hyundai Heavy Industries — a shipbuilding giant that had already invested $30 million in TerraPower back in 2022 — was named a preferred manufacturer for components of the Natrium Reactor Enclosure System, while a broader tripartite arrangement covers design, supply chain, construction, and delivery of a standardized fleet. A month later, in June 2026, Natrium formally entered the United Kingdom's Generic Design Assessment and TerraPower launched a British subsidiary, opening a second national market. The moat TerraPower is building is less any single patent than this accumulating stack of regulatory approvals, sites, fuel contracts, and manufacturing partners that a newly funded rival cannot buy off the shelf.
Financials: a $4 billion demonstration against zero revenue
TerraPower is still a private, pre-revenue company, and the Kemmerer demonstration is an expensive proposition — publicly estimated at roughly $4 billion, roughly half of it covered by the Department of Energy's Advanced Reactor Demonstration Program cost share of up to $2 billion, with TerraPower and its partners responsible for the rest. That structure is the entire reason the plant is being built now rather than a decade from now: without the federal match, the first-of-a-kind risk would likely have been uninvestable.
On the private side, TerraPower has raised more than a billion dollars from Bill Gates and early venture backers over the years, and in 2022 closed a $750 million round led by SK Group and Gates that the company explicitly tied to both its reactor work and a growing medical-isotopes business. That isotopes arm — producing materials such as actinium-225 for targeted cancer therapies — is a reminder that TerraPower is not a pure-play reactor company, though power generation remains the core of the thesis. For now the balance sheet is a developer's: heavy investment, government support, and no commercial electricity sales until the early 2030s.
What to watch
The near-term story is execution at Kemmerer. Project owner US SFR Owner LLC has told the NRC it expects to complete construction by February 2031, before entering the operating-license phase and reaching commercial operation in the early 2030s. Because Natrium is being built under a two-step Part 50 process — construction permit first, operating license later — the company can pour concrete now while the operating-license review proceeds in parallel, but any slip in the build, the workforce ramp toward roughly 1,600, or the HALEU deliveries would ripple straight through the schedule. First sodium criticality, when it comes, would be the first new sodium-cooled fast reactor to operate in the U.S. in decades.
The longer arc is whether the fleet materializes. Watch whether the Meta agreement hardens from framework into financed, under-construction units on its 2032 timeline; whether the HD Hyundai manufacturing and EPC partnerships actually compress the cost and schedule of plants two through eight; and whether the UK Generic Design Assessment opens a credible second market. TerraPower has done the hardest thing in advanced nuclear — start building — but the value case now depends on turning one demonstration into a repeatable product, on time and on budget, in an industry with a long history of doing neither.
Sources
- TerraPower — construction begins on America's first utility-scale advanced nuclear plant
- U.S. Department of Energy — NRC issues construction permit for Natrium
- World Nuclear News — TerraPower and GE Hitachi introduce Natrium
- GeekWire — TerraPower warns of two-year delay over HALEU supply
- TerraPower — commercialization agreements with HD Hyundai and HDEC
- CNBC — TerraPower raises $750M for nuclear and medical isotopes