TerraPower LLC

Overview

TerraPower is a private nuclear energy company founded in 2008 by Bill Gates and Nathan Myhrvold, the former Microsoft Chief Technology Officer, with the goal of developing advanced reactor technology that could operate on spent nuclear fuel and run for decades without refueling. The company was built around the traveling wave reactor (TWR), a sodium-cooled fast reactor concept in which a self-sustaining fission wave propagates slowly through a core loaded with depleted uranium, breeding and burning fuel in place. The TWR never reached construction. A 2015 partnership with China National Nuclear Corporation to prototype the design collapsed in 2019 when the Trump administration restricted nuclear technology transfers to China, forcing TerraPower to pivot its entire development program. The company redesigned around Natrium, a simpler sodium fast reactor paired with a molten salt thermal energy storage system, and won a $2 billion U.S. Department of Energy cost-share award in 2020. President and CEO Chris Levesque has led the company since 2017.

The Natrium design combines a 345-megawatt sodium-cooled fast reactor with what TerraPower describes as gigawatt-scale molten salt thermal energy storage, a combination that gives the plant a capability no conventional nuclear reactor offers: dispatchability. The reactor runs at full thermal output continuously, charging the molten salt storage when grid demand is low. When demand spikes, whether during summer afternoons, winter evenings, or any period when wind and solar generation drops, the stored heat supplements the reactor output. PacifiCorp's October 2025 regulatory filing describes the resulting plant as capable of flexible operation from 100 to 500 megawatts for up to five and a half hours. The storage system turns a baseload nuclear plant into a resource that can follow the grid in real time, making it compatible with a high-renewable system in a way that conventional nuclear plants are not.

TerraPower broke ground on non-nuclear site preparation at Kemmerer, Wyoming in June 2024, adjacent to PacifiCorp's retiring Naughton coal plant, and received the first state permit ever issued to a commercial-scale advanced nuclear project from the Wyoming Industrial Siting Council in January 2025. The NRC issued its final environmental impact statement in October 2025 and completed its final safety evaluation on December 1, 2025. The Commission authorized the construction permit on March 4, 2026, the first the agency had granted any commercial reactor in nearly a decade and the first ever for a commercial non-light-water design. Official nuclear construction began April 23, 2026, with Bechtel as the engineering, procurement, and construction partner. The project targets commercial operation in 2030 to 2031. At peak, approximately 1,600 construction workers will be on site; the completed plant will employ about 250 people permanently, replacing much of the workforce from the coal plant it is transitioning.

Technology & projects

From traveling wave to Natrium

TerraPower was originally built around one of the most intellectually ambitious reactor concepts in modern nuclear history: the traveling wave reactor. The TWR's core insight was that a fast neutron chain reaction could be designed to breed fissile plutonium from fertile depleted uranium at the front of the reaction zone while burning previously bred fuel at the rear, with the reaction wave propagating slowly through the core over decades. A fully developed TWR would run on depleted uranium, of which the United States has approximately 700,000 metric tons stockpiled as a waste product of uranium enrichment, and produce power for 40 to 60 years without refueling or removing fuel for reprocessing. The spent fuel produced would contain significantly less long-lived transuranics than conventional LWR waste. Gates and Myhrvold believed the TWR was the reactor concept that could genuinely address nuclear power's cost, proliferation, and waste concerns simultaneously.

The TWR required proof-of-concept at scale, and building a prototype in the United States faced the same slow regulatory pathway that had made new nuclear construction economically unattractive for decades. In 2015, TerraPower signed an agreement with China National Nuclear Corporation (CNNC) to build a 600-megawatt TWR prototype at Xiapu in Fujian province, with construction planned for 2018 to 2025. The arrangement was practical: China was building nuclear capacity aggressively, had the construction infrastructure and regulatory willingness to move quickly, and was willing to fund a substantial share of the demonstration cost. The deal would give TerraPower the operational data needed to refine the TWR design for commercial deployment.

In 2018 and 2019, the Trump administration tightened restrictions on nuclear technology transfers to China amid broader trade tensions and nuclear nonproliferation concerns. TerraPower was required to terminate the CNNC partnership. The company lost four years of development time and the capital the Chinese partnership would have provided. Bill Gates was publicly critical of the decision, arguing that it prevented a valuable nuclear demonstration without meaningfully addressing proliferation risk. Regardless of the policy merits, the effect on TerraPower was concrete: the company had to find a new path to demonstrating its technology without a Chinese partner, in a U.S. nuclear regulatory environment that had not licensed a commercial non-light-water reactor since the 1970s.

The pivot to Natrium was a pragmatic response. The TWR's core physics, while theoretically elegant, required achieving very high fuel burnup levels and demonstrating novel fuel metallurgy that would take years to qualify. A sodium fast reactor without the traveling wave feature, operating more like established designs such as Russia's BN-800 or Idaho National Laboratory's EBR-II, which ran successfully for 30 years from 1964 to 1994, could be built and licensed on the ARDP timeline. The molten salt thermal energy storage system was Natrium's genuine innovation: it reused salt storage technology proven in concentrating solar plants and applied it to nuclear to produce a dispatchable zero-carbon plant capable of competing with gas peakers for grid services. The DOE ARDP selection in October 2020 validated the design as fundable and buildable in the United States.

TerraPower has stated that the TWR concept is not abandoned. If Natrium demonstrates that a commercial sodium fast reactor can be built and operated in the United States, it establishes the platform, meaning the regulatory precedent, the supply chain, the workforce, and the fuel cycle infrastructure, on which a TWR could be built as a second-generation design. Natrium is, in this framing, the proof-of-concept that the TWR needed China to provide but can now be supplied domestically.

Funding and capital structure

TerraPower has never published a cumulative fundraising total, and the figures in circulation disagree: trade press has cited more than $2.2 billion of private capital raised between 2022 and 2025, while private-market databases put the total closer to $1.6 billion. The disclosed rounds are firmer ground. TerraPower announced $750 million in August 2022, a round co-led by Bill Gates and SK Group, whose SK Inc and SK Innovation together committed about $250 million; a further $80 million followed in November 2022; and $650 million was announced on June 18, 2025, with NVentures, the venture capital arm of NVIDIA, joining alongside returning investors Gates and HD Hyundai. Earlier backers include Khosla Ventures and Charles River Ventures. No valuation was disclosed for the 2025 round; the last publicly reported figure is $3.8 billion from 2022. TerraPower has no IPO filing or announced timetable, though its shares change hands on secondary platforms among accredited investors.

One transaction is easily misread as a fundraise and is not one. On January 20, 2026, Korea Hydro & Nuclear Power joined the shareholder base by purchasing part of SK Innovation's existing stake, with CFIUS clearance the month before. It was the first direct investment by a Korean state energy enterprise in a leading advanced reactor developer, but it was a secondary transfer between shareholders rather than new primary capital into the company. The Korean relationship deepened commercially on May 20, 2026, when TerraPower named HD Hyundai Heavy Industries a preferred manufacturer for Natrium reactor enclosure system components and signed a broader collaboration with HD Hyundai and Hyundai Engineering & Construction covering design, manufacturing, supply chain, and delivery of multiple units. TerraPower has also begun exporting the licensing case: Natrium entered the United Kingdom's Generic Design Assessment process in June 2026, having been accepted into the queue in February, with the Office for Nuclear Regulation, the Environment Agency, and Natural Resources Wales conducting the review.

The capital structure of the Kemmerer project matters for the replication case. At roughly $4 billion for a 345-megawatt plant, the overnight capital cost is approximately $11,600 per kilowatt, several times the cost of a combined-cycle gas plant but in line with recent conventional nuclear construction in the United States and lower than the final cost of Vogtle Units 3 and 4. That $4 billion figure dates to the 2021 project announcement and has never been publicly re-baselined, which is worth stating plainly: no cost overrun has been reported, but neither has the estimate been reaffirmed against five years of design maturation, inflation, and a nuclear construction market that has tightened considerably. TerraPower argues that the first plant is a demonstration project that will not reflect the economics of serial production, where factory-built modular components, a trained workforce, and an established supply chain would reduce costs substantially. Whether that argument holds depends on whether the first plant actually comes in at or near its estimate, which is far from guaranteed given the track record of nuclear first-of-a-kind construction in the United States.

Strategy & outlook

TerraPower's medium-term strategy is to complete Kemmerer Unit 1 on schedule and budget, establish the Natrium design as a licensed and demonstrated technology, and develop a pipeline of subsequent units. The company has said it expects to site additional Natrium plants adjacent to retiring coal facilities across the western United States, a deployment model that takes advantage of existing transmission infrastructure, experienced industrial workforces, and communities that need economic replacement for coal employment. Rocky Mountain Power, now contracted for Kemmerer Unit 1's output under a 40-year agreement, remains the most natural utility customer; broader deployment would require licensed designs and supply chains that only exist once Kemmerer operates.

Data center power demand has materially changed the commercial case for new nuclear in the period since TerraPower's ARDP award, and the Meta agreement of January 2026 converted that from a thesis into a signed arrangement. A 345-megawatt sodium fast reactor with the ability to flex to 500 megawatts on demand is a close match for a large data center campus that needs firm, carbon-free, around-the-clock power, and NVIDIA's June 2025 investment through NVentures pointed the same direction. The significance for TerraPower is the funding path: private customer capital supporting early development of units two and three is what a follow-on fleet needs if it is to be built without ARDP-style federal co-investment. The caution is that the Meta arrangement funds development of two units and options six more, no site has been chosen, and the first delivery is targeted for 2032 at the earliest. It is a commitment to develop, not revenue.

Key considerations

Nuclear first-of-a-kind construction in the United States has a poor cost and schedule track record. Vogtle Units 3 and 4, the most recent large nuclear project, took 15 years from license application to commercial operation, came in at roughly twice the projected cost, and required a federal loan guarantee and Georgia ratepayer support to survive. TerraPower's project has the advantage of Bechtel as a contractor with nuclear construction experience, a DOE cost-share that removes half the financial risk, and a simpler reactor design than Westinghouse's AP1000. But the Kemmerer project is still a first-of-a-kind sodium fast reactor built to a new NRC license basis in a country that has not built a commercial non-LWR in more than 50 years. The supply chain for sodium-compatible components, meaning specialized pumps, heat exchangers, and instrumentation rated for liquid sodium service, is limited. The fuel qualification process for Natrium's sodium-compatible fuel is ongoing. Cost and schedule overruns are not a failure scenario; they are the historical base case for advanced nuclear construction.

Regulatory and policy continuity is a second consideration. The DOE ARDP commitment of up to $2 billion is a multi-year cost-share that spans administrations and remains subject to annual appropriations; DOE has not published how much of it has actually been obligated or disbursed. Political support for advanced nuclear has been bipartisan across the Biden and Trump administrations: the Nuclear Energy Innovation and Modernization Act of 2018 directed the NRC to develop a licensing pathway for non-light-water reactors, the ADVANCE Act followed in 2024, and Executive Order 14300 imposed licensing deadlines in 2025. The construction permit authorized in March 2026 is a durable regulatory action that cannot easily be revoked, and TerraPower reached it through the conventional NRC Part 50 route rather than through the DOE Reactor Pilot Program, in which it does not participate. But future policy changes affecting DOE disbursements, uranium supply chains, or offtake agreements with regulated utilities would affect the project's economics.

The replication case is the most important long-run question. A single Natrium demonstration at $4 billion, even if it succeeds perfectly, does not by itself transform the economics of advanced nuclear. The path from one demonstration plant to 10 or 50 deployed units requires a supply chain, a trained workforce, standardized permitting, and commercial power purchase agreements at prices that reflect the cost of subsequent units rather than the cost of a demonstration. TerraPower is building all of those prerequisites simultaneously with the Kemmerer project. If Kemmerer completes on schedule and at or near cost, TerraPower will have accomplished something no advanced nuclear developer has done in the United States in half a century. That outcome is not certain, but the regulatory, financial, and engineering groundwork being laid between 2024 and 2026 is more credible than any advanced reactor project in recent U.S. history.

Sources