TerraPower Natrium Reactor Explained: How a Molten Salt Battery Became America’s First Advanced Nuclear Plant

Key takeaway: The TerraPower Natrium reactor is a 345 MW sodium-cooled fast reactor paired with a molten salt thermal storage system that lets the plant surge to 500 MW for more than five hours. It is under construction at Kemmerer Power Station Unit 1 in Wyoming, beside a retiring coal-fired plant, backed by up to $2 billion from the Department of Energy’s Advanced Reactor Demonstration Program on a 50/50 match, and targeted for operation in 2030.

The TerraPower Natrium reactor is a 345-megawatt sodium-cooled fast reactor with an integrated molten salt thermal storage system, a combination that allows the plant to lift electrical output from 345 MW of baseload to roughly 500 MW of peak generation for more than five hours on demand. It is being built as Kemmerer Power Station Unit 1 in Kemmerer, Wyoming, on the site of a retiring coal-fired plant, with commercial operation targeted for 2030. The salt battery exists because a fast reactor is most economic at constant output while grid demand is volatile, therefore the heat is stored and the electricity is time-shifted into the hours that pay most.

About this analysis: Written by the billgates.pk editorial team, which tracks Bill Gates’ climate technology and global health investments with a focus on funding structures, named recipients and verifiable timelines. All figures are current as of 6 November 2026.

Transparency note: TerraPower is a privately held company founded by Bill Gates in 2008. It is not a Gates Foundation grantee, and Gates Foundation funds are not deployed in it. Output, workforce and schedule figures below are operator disclosures from TerraPower unless attributed to the Nuclear Regulatory Commission, the Department of Energy or independent trade press. Nothing here is investment advice.

Natrium is not a molten salt reactor — and the distinction matters

Natrium is not a molten salt reactor. The salt never touches the fuel. Natrium is a sodium-cooled fast reactor whose heat is transferred out of the core and into a separate molten salt storage tank, which sits on the non-nuclear side of the plant and functions purely as a thermal battery.

That distinction carries a regulatory consequence. In a true molten salt reactor the fissile material is dissolved into the salt itself, which means the salt loop is a nuclear system and every tank, pump and valve falls inside the reactor safety boundary. In the Natrium design the salt is a heat-transfer and storage medium only, therefore the storage tanks sit outside that boundary and the licensing case covers a conventional industrial fluid rather than a fuel-bearing one. The effect is a materially simpler safety argument for the largest components on site.

The architecture reflects this split. The nuclear island houses the sodium-cooled core producing 345 MW of baseload electrical output. The energy island houses the salt tanks, the steam generator and the turbine. Because the two are decoupled, the turbine is sized well above the reactor’s steady output and can run harder than the core for limited periods — the physical basis of the surge to 500 MW.

The design is not a blank sheet. Sodium fast reactor operating and test experience in the United States gave the Nuclear Regulatory Commission a technical foundation to review against, which is one reason the agency was able to complete its safety review of the Kemmerer project ahead of schedule in December 2025.

Dimension	True molten salt reactor	Natrium (TerraPower)
Role of the salt	Fuel carrier and coolant	Heat storage medium only
Coolant in the core	Molten salt	Liquid sodium
Fuel form	Fuel dissolved in the salt	Solid fuel, separate from the salt
Where the salt sits	Inside the nuclear island	Energy island, outside the reactor safety boundary
Reason for the salt	Core heat transport and fuel chemistry	Storing heat to time-shift electricity output

What molten salt energy storage actually is

Molten salt energy storage works in four movements: heat passes from the liquid sodium loop into molten nitrate salt, the hot salt is held in an insulated tank, the salt is drawn off to generate steam when demand or prices peak, and the cooled salt returns to a cold tank to be reheated. No electricity is stored at any point — only heat.

The causal logic is straightforward. A fast reactor is most economic running flat out continuously, therefore surplus heat during low-demand hours has to go somewhere other than a condenser. Consequently the hot tank absorbs it, and the turbine releases it during the evening ramp when solar output collapses and prices spike.

1. The sodium-cooled core produces heat continuously at 345 MW-equivalent electrical output.
2. Heat transfers through an intermediate loop from liquid sodium into molten nitrate salt, with no contact between salt and fuel.
3. Hot salt is held in an insulated storage tank through low-demand, low-price hours.
4. On demand, hot salt is pumped to the steam generator, driving the turbine to roughly 500 MW for more than five hours before returning to the cold tank.

This is a different asset class from a lithium-ion battery, and the comparison is frequently muddled. Lithium-ion stores electrons; molten salt stores heat. The storage medium is cheap industrial nitrate salt already used at commercial scale in concentrated solar power plants, which means the materials risk is lower than for a novel chemistry. And the duration is measured in hours of 500 MW output rather than the minutes-to-low-hours range in which most grid batteries deliver frequency support and arbitrage.

Energy systems researchers, including groups at the University of Washington working on decarbonised grid modelling, have consistently identified multi-hour firm capacity as the hardest gap to fill in a renewables-heavy system. Dispatchability is therefore the commercial point of the design: it converts a flat-output asset into one that sells into peak pricing, and that revenue profile is the argument TerraPower is making to utilities and hyperscalers alike.

How a sodium-cooled fast reactor works

A sodium-cooled fast reactor uses liquid sodium metal as coolant, allowing operation near atmospheric pressure at roughly twice the outlet temperature of a light-water reactor, which raises thermal efficiency and removes the need for a high-pressure containment design.

“Fast” refers to the neutron spectrum. There is no moderator slowing neutrons down, so fission is sustained by fast neutrons — a spectrum that uses fuel more efficiently than a moderated core. “Sodium-cooled” refers to the coolant: a liquid metal with a high boiling point that stays liquid without being pressurised. Because the coolant does not need pressurisation to remain in the liquid phase, the plant avoids the pressure-driven failure modes that dominate light-water safety analysis, and the containment structure is designed against a different and narrower set of scenarios.

The temperature advantage then feeds the economics. A higher outlet temperature means more mechanical work extracted per unit of heat, which is why the same 345 MW reactor supports a turbine sized for 500 MW and why the stored salt heat converts to saleable electricity at a competitive efficiency.

Sodium carries a real liability that should not be glossed over: it reacts vigorously with water and air. The design response is to keep sodium in sealed loops with an intermediate circuit between the core and the water-steam system. TerraPower also built a dedicated Sodium Test & Fill Facility at Kemmerer — a non-nuclear support building designed to hold 400,000 gallons of liquid sodium — specifically to validate handling, pumping and purification procedures before nuclear operation. That facility is evidence of mitigation and of regulatory seriousness; it is not evidence that the chemical hazard has been removed.

Inside Kemmerer Unit 1, the Wyoming construction site

Kemmerer Power Station Unit 1 sits in Kemmerer, Wyoming, beside a retiring coal-fired plant in PacifiCorp’s service territory. Full construction began on 23 April 2026, following the Nuclear Regulatory Commission’s construction permit the previous month, and the project targets commercial operation in 2030. The siting logic is economic: transmission capacity, water rights, permitting precedent and a skilled industrial workforce already exist at a retiring coal site, therefore the TerraPower Natrium reactor inherits infrastructure rather than building it from scratch. Replacing retiring coal generation also carries the largest single emissions benefit available in the power sector, a pattern visible in EPA power sector greenhouse gas data.

On 15 October 2026, founder Bill Gates and chief executive Chris Levesque hosted a construction update at the site with Wyoming Governor Mark Gordon, U.S. Senator John Barrasso and the governors of Utah and Idaho. The delegation toured the nuclear island excavation and the near-complete Test & Fill facility.

“This is the moment our industry has been working toward for a generation. We’re not just breaking new ground on a first-of-a-kind nuclear plant in Wyoming; we’re building the next generation of America’s energy infrastructure.”
— Chris Levesque, President and CEO, TerraPower

The local economic case is the other half of the political coalition behind the project. TerraPower reports roughly 1,600 workers at peak construction falling to 250 permanent operating staff, and those are the figures Wyoming officials cite when defending a coal-to-nuclear transition in a county whose tax base was built on extraction.

Specification	Figure
Baseload output	345 MW
Peak output and duration	~500 MW for more than five hours
Homes powered at peak	Roughly 400,000
Reactor type	Sodium-cooled fast reactor with molten salt storage
Location and host site	Kemmerer, Wyoming; retiring coal-fired plant site
Construction start	23 April 2026
Peak construction workforce	~1,600 workers
Permanent jobs	250
Liquid sodium at Test & Fill facility	400,000 gallons
Target operation date	2030

The 4 March 2026 NRC permit and why 40 years mattered

On 4 March 2026 the U.S. Nuclear Regulatory Commission issued a construction permit for the Natrium plant at Kemmerer — the first construction permit the agency has issued for a commercial non-light-water power reactor in more than 40 years, a milestone independently reported by World Nuclear News.

A construction permit is not an operating licence, and the difference governs the schedule. TerraPower is authorised to build; it must return to the NRC for separate authorisation to load fuel and operate. That is precisely why 2030 is a target rather than a commitment.

The permit’s commercial effect was immediate. Regulatory uncertainty had kept component suppliers from committing tooling and capital to a design that might never be built, so once the permit landed, that constraint lifted — and in August 2026 TerraPower announced 12 major supplier contracts to build out the domestic supply chain.

“Securing supplier contracts is fundamental to establishing the resilient and sustainable supply chain needed to build advanced nuclear plants.”
— Pat Young, Senior Vice President and Natrium Project Director, August 2026

1. December 2025 — NRC completes its safety review of the Kemmerer project ahead of schedule.
2. 4 March 2026 — NRC issues the construction permit, the first for a commercial non-light-water reactor in over 40 years.
3. 23 April 2026 — Full construction begins on Kemmerer Unit 1.
4. August 2026 — TerraPower announces 12 supplier contracts and a framework agreement with Hyundai Engineering & Construction and SK Innovation.
5. September 2026 — Bechtel steps away from the next construction phase; the EPC contract goes back to bid.
6. 15 October 2026 — Site update with Bill Gates, Chris Levesque, Governor Mark Gordon and Senator John Barrasso.

Who is paying — the DOE cost-share and the Bechtel exit

The Department of Energy’s Advanced Reactor Demonstration Program is investing up to $2 billion in the project, matched dollar-for-dollar by TerraPower and private investors under a 50/50 public-private structure. The matching requirement is the operative detail: federal money arrives only as private capital arrives beside it, which means the ARDP award functions as leverage rather than as a grant. TerraPower has not published a single all-in capital cost for the TerraPower Natrium reactor, therefore the cost-per-kilowatt figures circulating in commentary should be treated as third-party estimates rather than disclosures.

The schedule risk is concrete and worth stating without softening. Bechtel served as the original engineering, procurement and construction partner, helped secure the NRC permit and built the Sodium Test & Fill Facility — then announced in September 2026 that it was stepping away from the next phase. Consequently the remaining Kemmerer build is being rebid. A first-of-a-kind nuclear project changing principal contractor mid-build carries genuine schedule exposure, and the 2030 date depends on how quickly that contract closes and how much re-mobilisation the new contractor requires.

There is a counterweight. In August 2026 TerraPower signed a framework agreement with Hyundai Engineering & Construction and SK Innovation, selecting HDEC as EPC contractor for up to eight future Natrium plants globally. Contractor appetite for the design clearly exists at scale, which reduces the likelihood of a prolonged bidding gap at Kemmerer.

Element	Party	Amount or status
Federal cost-share	DOE Advanced Reactor Demonstration Program	Up to $2 billion
Private match	TerraPower and private investors	Dollar-for-dollar, 50/50 structure
Host utility	PacifiCorp (Kemmerer site)	Host site and service territory
Original EPC contractor	Bechtel	Stepped away September 2026
Current EPC status	Not awarded	Out to rebid as of November 2026
International EPC framework	Hyundai E&C with SK Innovation	Up to eight future plants
Published all-in capital cost	TerraPower	Not stated

The AI load problem the salt battery solves

In January 2026 Meta signed an agreement with TerraPower for up to eight Natrium plants by 2035, targeting up to 4 GW of peak power for the company’s AI infrastructure. That deal is the clearest commercial signal the design has received, and what Meta is buying is dispatchability rather than megawatt-hours.

The reason sits in the load shape. AI training runs and inference traffic swing sharply hour to hour, so a flat 345 MW contract is worth less to a hyperscaler than a 500 MW surge capability it can call on. Nuclear output that follows a data centre’s demand curve removes the need to overbuild firm capacity for the handful of hours that define the peak.

The grid-side effect compounds it. Interconnection studies and demand charges are driven by the single highest hour, which means a generator able to lift output precisely during that hour captures value out of all proportion to the energy it delivers. That asymmetry is the economic engine behind the entire salt-battery concept, and it explains why the Natrium design attracted hyperscaler interest that conventional baseload nuclear has struggled to win.

One honest caveat belongs here: the Meta agreement is a framework for plants that do not yet exist beyond Kemmerer. It validates demand convincingly. It does not confirm delivery.

Natrium vs small modular reactors vs light-water plants

Natrium occupies a middle position. At 345 MW baseload it is larger than most designs marketed as small modular reactors, smaller than the gigawatt-scale light-water fleet, and — among leading US designs — distinctive for shipping with integrated thermal storage rather than treating storage as a separate grid asset.

Dimension	Natrium	Typical small modular reactor	Light-water reactor
Electrical output	345 MW baseload, ~500 MW peak	Generally under 300 MWe by industry definition	Roughly 1,000 MWe and above
Coolant and operating pressure	Liquid sodium, near atmospheric	Most leading designs water-cooled and pressurised	Pressurised or boiling water, high pressure
Integrated energy storage	Molten salt thermal storage	Not part of typical designs	None
Dispatch flexibility	Surge to ~500 MW for 5+ hours	Limited; design dependent	Designed for steady baseload
US deployment status, November 2026	Under construction, Kemmerer Unit 1	No utility-scale unit in operation	Established operating fleet

The consequence is that these designs are not straight substitutes. SMRs compete on factory fabrication, modular transport and siting flexibility for constrained or remote locations. Natrium competes on dispatch, which means a buyer optimising for peak-hour firmness and a buyer optimising for small-footprint deployment are evaluating different products rather than negotiating the same purchase.

How Natrium fits the Breakthrough Energy thesis

TerraPower is Bill Gates’ privately held nuclear company, founded in 2008 and funded by Gates personally alongside investors including Breakthrough Energy Ventures. It is not a Gates Foundation grantee, and foundation funds are not deployed in it — a separation worth restating because conflating the two is the most common factual error in coverage of this project.

The investment thesis follows from that structure. Breakthrough Energy backs technologies capable of removing emissions at gigatonne scale with a credible path to cost parity, which is why a capital-intensive build with a decade-long horizon fits a portfolio that conventional venture funds would reject on duration alone. Public research funding typically precedes private deployment capital in this sector; agencies including the National Science Foundation have funded the advanced reactor and long-duration storage research base that commercial projects now draw on.

The sequencing at Kemmerer is the pattern applicants should read most closely. Federal cost-share de-risked the first unit, therefore the private capital requirement fell to a level investors would accept. The commercial offtake — Meta, in January 2026 — followed the regulatory milestone rather than preceding it, which is the reverse of the order software founders are trained to pursue. Our climate tech and AI health investing coverage tracks that sequencing across the wider portfolio.

• Emissions-reduction scale matters more than near-term revenue in the Breakthrough Energy screen.
• Pairing private capital with a federal cost-share programme strengthens rather than weakens an application.
• Thermal and long-duration storage is treated as infrastructure, which means longer payback horizons are acceptable.
• A named commercial offtake partner is the strongest validation signal available to a hardware-heavy project.

Timeline to 2030 and what to watch next

TerraPower targets 2030 for commercial operation of Kemmerer Unit 1, conditional on three dependencies that gate the schedule rather than merely influence it. The rebid EPC contract must close before major nuclear island construction scales up, because no contractor means no schedule. High-assay low-enriched uranium (HALEU) must be secured for first core loading, because a fast reactor cannot start on conventional fuel. And the NRC must grant an operating licence, because the March 2026 permit authorises construction only.

1. Award of the rebid EPC contract for the remainder of the Kemmerer build.
2. Confirmed HALEU fuel supply chain for first core loading.
3. Completion of nuclear island civil works and commissioning of the Test & Fill facility.
4. NRC operating licence application and review.
5. 2030 — targeted commercial operation of Kemmerer Unit 1.
6. 2035 — targeted delivery of up to eight further plants, including the Meta framework and Hyundai E&C units.

Each milestone proves something specific. The EPC award demonstrates that contractor confidence survived Bechtel’s exit. Confirmed HALEU supply demonstrates that the domestic fuel cycle has caught up with reactor construction. An operating licence would mark the first for a commercial non-light-water reactor in the modern regulatory era.

Frequently asked questions

What is the TerraPower Natrium reactor?

It is a 345 MW sodium-cooled fast reactor with integrated molten salt thermal storage that lets output surge to roughly 500 MW for more than five hours. It is under construction as Kemmerer Power Station Unit 1 in Kemmerer, Wyoming, developed by TerraPower, the company Bill Gates founded in 2008, with commercial operation targeted for 2030.

Is Natrium a molten salt reactor?

No. Natrium is a sodium-cooled fast reactor, and the molten salt is used only for energy storage outside the nuclear island, where it never contains fuel. The consequence is regulatory: because the salt is non-nuclear, the storage tanks sit outside the reactor safety boundary, which simplifies the licensing case for the plant’s largest components.

How does the molten salt battery actually store nuclear heat?

Heat moves from liquid sodium into molten nitrate salt through an intermediate loop and is held in an insulated hot tank. When demand or prices peak, the hot salt is pumped to the steam generator, driving the turbine to roughly 500 MW for more than five hours before the cooled salt returns to a cold tank to be reheated.

When will America’s first advanced nuclear plant come online?

TerraPower targets 2030 for Kemmerer Unit 1. Full construction began on 23 April 2026, and the schedule depends on closing the rebid EPC contract, securing HALEU fuel supply and obtaining an NRC operating licence, because the March 2026 permit authorises construction only.

Who is paying for the Natrium plant?

The Department of Energy’s Advanced Reactor Demonstration Program is investing up to $2 billion, matched dollar-for-dollar by TerraPower and private investors. The structure is a 50/50 public-private partnership. No single all-in capital cost has been published, therefore circulating project totals should be read as third-party estimates rather than disclosures.

Why did Meta sign a deal with TerraPower?

Meta signed a January 2026 agreement for up to eight Natrium plants by 2035, targeting up to 4 GW of peak power. The reason is load shape: AI training and inference demand swings sharply, therefore dispatchable surge capacity is worth more to a hyperscaler than an equivalent block of flat baseload generation.

Is a sodium-cooled reactor safe?

Sodium cooling removes the high-pressure failure modes central to light-water safety analysis, because the coolant stays liquid near atmospheric pressure — but sodium reacts with water and air, so it is kept in sealed loops with an intermediate circuit. TerraPower built a dedicated Test & Fill facility holding 400,000 gallons of liquid sodium to validate handling. Safety determinations rest with the NRC, not with the developer.

How is Natrium different from a small modular reactor?

Natrium is larger than most SMR designs at 345 MW baseload and is the leading US design shipping with integrated thermal storage. That means it competes on dispatch flexibility and peak-hour value, while SMRs compete on factory fabrication, modular transport and siting flexibility — different products for different buyers.

How much does Bill Gates own of TerraPower?

TerraPower is privately held, therefore ownership stakes fall outside the public reporting requirements administered by the Securities and Exchange Commission and are not disclosed. What is documented: Gates founded the company in 2008, chairs it, and has funded it personally and alongside investors including Breakthrough Energy Ventures. It operates separately from the Gates Foundation.

Can startups get funding from Breakthrough Energy for energy storage?

Yes — long-duration and thermal storage falls within Breakthrough Energy Ventures’ stated focus areas. The screen emphasises gigatonne-scale emissions potential and a credible route to cost parity, and it tolerates longer payback horizons than conventional venture capital, which is why infrastructure-shaped hardware projects receive consideration there.

Limitations, risks and the alternatives

Schedule and cost carry the clearest exposure. Kemmerer Unit 1 is a first-of-a-kind build that changed principal contractor in September 2026, therefore the 2030 target carries genuine slip risk until the rebid closes. First-of-a-kind nuclear projects have historically exceeded initial budgets, and because no all-in capital cost has been published for Natrium, there is no disclosed baseline against which overruns could even be measured.

Fuel is the second dependency. Natrium requires high-assay low-enriched uranium, and the domestic HALEU supply chain is still being built out, which means first core availability rests partly outside TerraPower’s control.

Alternatives compete for the same contracts. Light-water SMRs, grid-scale lithium-ion paired with wind and solar, and gas with carbon capture are all bidding for dispatchable low-carbon supply. Lithium-ion is deployable today at lower capital intensity and faster permitting, but it does not currently match multi-hour firm capacity at this scale and duration — that trade-off between deployability and firmness is what buyers are actually weighing.

Scope: all figures are current as of 6 November 2026 and describe a plant under construction, not an operating one. Nothing here constitutes investment advice or a safety assessment; those judgements belong to qualified financial advisers and to the Nuclear Regulatory Commission respectively.

Conclusion

The TerraPower Natrium reactor matters because it attacks nuclear power’s inflexibility rather than only its cost, and the molten salt battery is the component that makes that possible — storing heat so a constant-output core can sell variable-output electricity. Two variables now decide the outcome: how quickly the rebid EPC contract closes, and whether the NRC grants an operating licence in time to support the 2030 target. For founders and researchers in climate technology, the structure is as instructive as the engineering. Federal cost-share absorbed first-unit risk, private capital matched it, and the commercial offtake followed the regulatory milestone. Read more in our ongoing coverage of Breakthrough Energy and climate tech funding at billgates.pk.

References

• U.S. Department of Energy — Advanced Reactor Demonstration Program — the up to $2 billion award and 50/50 cost-share structure.
• U.S. Nuclear Regulatory Commission — December 2025 safety review completion, the 4 March 2026 construction permit, and the construction permit versus operating licence distinction.
• TerraPower — Natrium technology and Kemmerer project updates — operator-disclosed output, workforce, sodium inventory, supplier contracts, the 15 October 2026 site update, and quotes from Chris Levesque and Pat Young.
• World Nuclear News — independent reporting on the March 2026 permit and the September 2026 EPC transition.
• U.S. Environmental Protection Agency — power sector greenhouse gas emissions data.
• U.S. National Science Foundation — federal research funding context for advanced reactors and long-duration storage.
• U.S. Securities and Exchange Commission — disclosure requirements for public versus privately held companies.
• University of Washington — energy systems research on firm capacity in decarbonised grids.