The 10 kinds of nuclear reactors
How each one works, what it does well, where it falls short, and where it is running or being built today, by a nuclear engineer who has worked on both PWRs and BWRs.
Watch: the narrated video (6 minutes)
Plain-English explainer by Joe Miller, a nuclear engineer. Not investment advice. Figures are the latest public numbers we could find and change fast; sources are at the bottom of each page.
First, the basics
1. Pressurized water reactor (PWR)
How it works: Water under high pressure (about 2,250 psi) cools the core and stays liquid at about 600°F. It heats a second loop of water in a steam generator, and that steam spins the turbine.
Strengths: The most common reactor on Earth: about 7 in 10 of the world's roughly 440 power reactors, and about two-thirds of the 94 in the U.S. Decades of experience; the radioactive water stays in the first loop.
Weak spots: Big, expensive and slow to build. Vogtle 3 and 4 in Georgia (AP1000s, finished 2023 and 2024) ran about seven years late and billions over budget.
Where you'll find it: Vogtle 3 and 4 (Georgia); Palisades (Michigan) restarting; the Westinghouse AP1000 is being offered worldwide.
2. Boiling water reactor (BWR)
How it works: Water boils right in the core, and the steam goes straight to the turbine. One loop instead of two.
Strengths: Simpler, with fewer big components and a lower operating pressure (about 1,000 psi). About a third of U.S. reactors are BWRs.
Weak spots: The turbine sees slightly radioactive steam, so it needs shielding during maintenance. Fukushima Daiichi's units were early BWRs.
Where you'll find it: River Bend (Louisiana, where I directed nuclear engineering), Grand Gulf, Clinton; the BWRX-300 small reactor is being built at Darlington, Ontario.
3. Heavy water reactor (CANDU)
How it works: Uses heavy water, water made with a heavier form of hydrogen, to slow neutrons. That lets it run on natural uranium with no enrichment.
Strengths: No enrichment plant needed; it can be refueled while running; long track record in Canada and India.
Weak spots: Heavy water is expensive, and the design makes some tritium that has to be managed.
Where you'll find it: Canada (Bruce, Darlington, Pickering), India, Argentina, Romania, South Korea.
4. High-temperature gas reactor (HTGR)
How it works: Helium gas cools the core, and the fuel is TRISO: poppy-seed-size uranium kernels wrapped in ceramic layers, packed into pebbles or blocks.
Strengths: Runs hot (about 750°C and up), good for industrial heat as well as electricity. TRISO fuel holds its radioactivity even at very high temperatures, so it is very hard to melt down.
Weak spots: Large core for its power, special fuel supply, and helium leaks. The U.S. tried this before (Fort St. Vrain, Colorado, 1979 to 1989) with mixed results.
Where you'll find it: China's HTR-PM has run commercially since 2023; X-energy's Xe-100 is planned for Dow in Texas and for Amazon in Washington State. I wrote one of my first papers on HTGR fuel management in 1975.
5. Molten salt reactor
How it works: Either the fuel is dissolved in hot liquid salt, or the salt is just the coolant for solid fuel. The salt runs at near-normal pressure.
Strengths: Low pressure, high temperature, and the salt does not boil away. Some designs can use thorium, which is more plentiful than uranium.
Weak spots: Hot salt corrodes metal, and regulators have little experience with it. The U.S. ran one test reactor at Oak Ridge from 1965 to 1969.
Where you'll find it: China's 2-megawatt thorium test reactor at Wuwei converted thorium into uranium fuel in 2025; Kairos Power (salt-cooled, for Google) is building its Hermes test reactor in Oak Ridge, Tennessee.
6. Sodium-cooled fast reactor
How it works: Liquid sodium metal cools the core, and there is no moderator, so the neutrons stay fast. Fast neutrons can burn long-lived waste and even make more fuel than they use.
Strengths: Very low pressure, excellent heat transfer, and much better use of uranium.
Weak spots: Sodium burns in air and reacts with water, so leaks are serious. Needs higher-enriched fuel (HALEU), which is in short supply.
Where you'll find it: TerraPower's Natrium plant in Kemmerer, Wyoming (345 MW, boosted to 500 MW with molten-salt heat storage) got its NRC construction permit in March 2026, aiming for about 2030; Meta has signed for up to eight units. Russia's BN-800 is running now.
7. Lead-cooled fast reactor
How it works: Melted lead cools a fast-neutron core.
Strengths: Lead does not burn or react with water, boils at about 1,750°C and shields radiation itself.
Weak spots: Lead is very heavy and corrosive, and it freezes at about 327°C, so it must be kept hot.
Where you'll find it: Russia's BREST-OD-300 is under construction in Seversk; Soviet Alfa-class submarines used lead-bismuth reactors; Westinghouse is developing a lead-cooled design.
8. Small modular reactors (SMRs)
How it works: Not a coolant, a size: reactors under about 300 megawatts, built mostly in a factory and shipped to the site. Most use proven light-water designs.
Strengths: Smaller up-front cost, faster to build if the factory model works, and fits where a giant plant won't.
Weak spots: First units cost more per megawatt than big plants; savings only come after many are built.
Where you'll find it: GE Vernova Hitachi BWRX-300 at Darlington, Ontario (target 2029-30) and planned at TVA's Clinch River; NuScale (NRC approval of its 77-MW module in 2025); Holtec SMR-300 at Palisades (early 2030s).
9. Microreactors
How it works: Tiny reactors of about 1 to 20 megawatts that fit on a truck, for military bases, mines, remote towns and data centers.
Strengths: Can run for years without refueling, and can be moved.
Weak spots: Expensive per megawatt, and new to regulators.
Where you'll find it: In 2026, four test microreactors reached criticality under the Energy Department's pilot program, including Antares (sodium heat-pipe) and Valar Atomics (helium gas). Oklo's sodium-cooled Aurora is planned for Meta.
10. Fusion
How it works: Joins light atoms, forms of hydrogen, at over 100 million degrees, the way the sun does, instead of splitting heavy ones.
Strengths: No chain reaction to run away, no meltdown, and no long-lived high-level waste; fuel from seawater and lithium.
Weak spots: No machine has yet made more electricity than it uses. Materials, tritium supply and cost are unsolved.
Where you'll find it: Lawrence Livermore's NIF reached ignition in 2022 (8.6 MJ record in 2025); Commonwealth Fusion's SPARC is about 80% built, aiming at first plasma in 2027; Helion is building a plant for Microsoft with a 2028 target; ITER in France plans research operations around 2034.
Side by side
| # | Type | Coolant | Neutrons | Status |
|---|---|---|---|---|
| 1 | Pressurized water (PWR) | Water, high pressure | Slow | Most of the world's reactors |
| 2 | Boiling water (BWR) | Boiling water | Slow | About a third of U.S. reactors |
| 3 | Heavy water (CANDU) | Heavy water | Slow | Canada, India and others |
| 4 | High-temperature gas | Helium | Slow | Running in China; U.S. units planned |
| 5 | Molten salt | Liquid salt | Slow (most) | Test reactors |
| 6 | Sodium-cooled fast | Liquid sodium | Fast | Running in Russia; Wyoming plant permitted |
| 7 | Lead-cooled fast | Liquid lead | Fast | Under construction in Russia |
| 8 | Small modular (SMR) | Usually water | Usually slow | First units under construction |
| 9 | Microreactors | Gas, sodium or heat pipes | Varies | Test units reached criticality in 2026 |
| 10 | Fusion | Not fission | Fusion | Experiments; no net electricity yet |
My take as a nuclear engineer
The workhorses aren't going anywhere. PWRs and BWRs make almost all nuclear power today and will for at least another decade. Keeping them running, and restarting the few that can be restarted, is the fastest and cheapest clean power we have.
SMRs are the next real step, because most use the same water technology regulators and operators already know. The BWRX-300 at Darlington is the one to watch.
The advanced designs are real physics with real engineering left to do. Sodium, gas and salt reactors have all run before; the hard parts now are fuel supply (especially HALEU), cost and licensing. Expect first commercial units around 2030 and real numbers in the 2030s.
Fusion is the long game. The science has crossed real milestones, but no one has delivered net electricity. I'd plan on the 2040s for fusion at scale.
Sources
Checked November 15, 2026.
- American Nuclear Society: NRC approves TerraPower construction permit (Mar. 2026)
- The Register: TerraPower gets permission to build sodium-cooled reactor (Mar. 5, 2026)
- American Nuclear Society: The deadline arrives, checking in on the Reactor Pilot Program (July 2026)
- U.S. DOE: Department of Energy celebrates fourth criticality ahead of July 4th goal
- SMR Intel: State of Small Modular Reactors 2026
- POWER: China's molten salt reactor reaches thorium-uranium conversion milestone
- NucNet: Russia's BREST-OD-300 fast reactor scheduled for operation
- Discovery Alert: Fusion's commercial gates, what is proven and what is not (Oct. 20, 2026)
- U.S. Energy Information Administration: Nuclear explained
- World Nuclear Association: Nuclear power reactors