In Kemmerer, Wyoming, a town of roughly 2,400 people best known for a shuttering coal plant, TerraPower has begun the long, regulated work of building the first utility-scale advanced reactor in the United States. The Nuclear Regulatory Commission issued the construction permit for the Natrium demonstration plant in 2025 [Department of Energy, 2025], and the Wyoming Industrial Siting Council followed with approval covering the portions of the project outside NRC jurisdiction [ANS Nuclear Newswire, 2026]. Operations are targeted for 2030 [TerraPower website, October 2025]. After nearly two decades of design work, the company is now pouring concrete on a real site, with a real workforce, against a real deadline.
The bet
TerraPower, founded in 2006 by Bill Gates, Nathan Myhrvold, and John Gilleland, sells one core product: the Natrium reactor, a 345-megawatt sodium-cooled fast reactor paired with a molten-salt energy storage system that can briefly boost output to 500 megawatts. The pitch to utilities is dispatchable carbon-free power that follows the grid's load curve rather than running flat. The Kemmerer Power Station Unit 1, developed in partnership with PacifiCorp, is the proof point. Beyond the demonstration plant, TerraPower has a second business line in medical isotopes, including actinium-225 for cancer treatment [TerraPower website].
The commercial wedge widened considerably this year. TerraPower entered an agreement with Meta covering eight Natrium plants, including early development for two units with rights for up to six additional [TerraPower website]. The company also signed a 10-year offtake agreement for high-assay low-enriched uranium (HALEU) reported at $3.5 billion [LinkedIn, Nick Lawson], addressing what has been the single biggest supply-chain question hanging over the advanced reactor sector.
Why it could be big
The timing is unusually favorable for nuclear hardware. Hyperscaler power demand is rising faster than utilities can build generation, and data center operators have started signing directly with reactor developers rather than waiting for grid capacity. TerraPower's 2024 raise of roughly $650 million pulled in NVentures (Nvidia's venture arm), HD Hyundai, and SK Group [Crunchbase, 2024], a cap table that reads as a bet on the AI-energy nexus rather than a traditional cleantech thesis. Gates remains board chair and has been the company's most visible long-term capital source.
| Metric | Value |
|---|---|
| Total disclosed funding | 650 $M |
| Meta offtake (8 plants, announced) | 8 reactors |
| HALEU offtake agreement | 3500 $M |
| Kemmerer permanent jobs | 250 jobs |
If Kemmerer comes online near schedule and the Meta units follow, TerraPower would be one of very few companies in the world with a deployed, permitted, fueled advanced reactor design and a backlog of identified sites. The regulatory moat alone, given how long NRC review took, is meaningful. The Natrium design has also entered the UK's Generic Design Assessment process [World Nuclear News], which would open a second jurisdiction.
The team and traction
Chris Levesque is President and CEO [TerraPower website]. John Gilleland, one of the original co-founders, serves as Chief Technical Officer and a board member [Craft.co]. Nathan Myhrvold is Vice Chairman of the Board. The Kemmerer project is projected to bring 250 permanent jobs to the region [Business Wyoming], on top of a peak construction workforce in the high hundreds. Active hiring on the company's job board includes a Principal Fuel Fabrication Engineer, a Nuclear Security Systems Resident Engineer, and a Quality Assurance Manager for the U.S. operations [Greenhouse / TerraPower], roles consistent with a company moving from design into licensed construction.
Utility partnerships extend beyond PacifiCorp. TerraPower and Evergy signed an agreement with the State of Kansas to explore Natrium deployment [TerraPower website], and a separate agreement with Southern Company covers a fast-spectrum molten salt research reactor at Idaho National Laboratory [TerraPower website]. None of those add up to firm orders the way the Meta agreement does, but they form a credible pipeline.
The honest counterfactual
The most cited concern is pace. Critics quoted in E&E News argue TerraPower's path to a commercial reactor by the early 2030s is aggressive given how few sodium-cooled fast reactors have operated commercially anywhere [E&E News by POLITICO], and a more recent piece raised the long-running question of sodium's reactivity with water and air in the event of a coolant breach [The Boston Times, 2026]. Sodium fires are a real failure mode, and the operating history of fast reactors globally is mixed. The bull answer is specific: the Natrium design separates the nuclear island from the power and storage island, runs at near-atmospheric pressure rather than the high pressures of light-water reactors, and has now passed an NRC final safety evaluation followed by an issued construction permit [ANS Nuclear Newswire, 2025; Department of Energy, 2025]. That is the most rigorous safety review any U.S. advanced reactor design has cleared. The schedule risk is real; the design-basis risk has been litigated in front of the regulator and the regulator said yes.
What to watch
The next twelve months are about execution on three fronts: continued construction progress at Kemmerer toward the 2030 operations target, conversion of the Meta agreement's two early-development units into a firm site selection and permit filing, and the first deliveries against the HALEU offtake. A follow-on funding round is plausible given the capital intensity of building eight more reactors, and any movement on the UK GDA process would broaden the addressable market. The Kemmerer pour is the milestone that matters most. If concrete keeps moving and the fuel arrives on time, TerraPower stops being a design company and starts being an operator.
Technical breakdown
Natrium is a pool-type sodium-cooled fast reactor rated at 345 MWe baseload, coupled to a molten salt thermal storage loop that can deliver 500 MWe for several hours. Sodium operates as coolant at near-atmospheric pressure, which simplifies the containment design relative to pressurized water reactors but introduces chemical reactivity risks that drive the intermediate sodium loop and the inert-gas cover requirements. Fuel is HALEU at enrichments up to 19.75 percent U-235, which is why the $3.5B offtake matters: domestic HALEU supply is the binding constraint for nearly every advanced reactor in the U.S. pipeline. The storage coupling is the commercial innovation more than the reactor physics, because it lets a single unit bid into both baseload and peaking markets.
What could go wrong at scale
The sober assessment: a single demonstration plant on schedule does not prove an eight-unit fleet on schedule. Fast reactor programs in France, Japan, and Russia have all hit multi-year delays driven by sodium handling incidents, fuel qualification, and supply-chain bottlenecks for specialized components like intermediate heat exchangers and electromagnetic pumps. If Kemmerer slips two or three years, the Meta units slip with it, and the HALEU offtake becomes a working capital problem rather than a strategic asset. The NRC has approved one design at one site; it has not approved a manufacturing line. Scaling from one to nine is where most reactor programs in history have stalled, and TerraPower will be doing it with a workforce, a fuel supply chain, and a regulatory framework that have never been exercised at that cadence before.