Nuclear guide

Small Modular Reactors (SMRs) Explained: Promise and Reality

What SMRs are, which designs are real (BWRX-300, NuScale, Linglong One, floating reactors), what they may cost, and why Southeast Asia is interested.

Updated 6 min read 1,318 words

Small modular reactors get more headlines than almost any other energy technology, and more than 80 designs are under development worldwide according to the IAEA. The idea is attractive: instead of a one-off 1,600 MW megaproject that takes 10–15 years, build smaller reactors in a factory, ship them to site, and add units as demand grows. The reality in 2026 is more mixed. A handful of small reactors run, several Western projects have firm construction dates, and costs remain the big unknown. This guide explains what an SMR is, which projects are real, and what to make of the claims.

What makes a reactor “small” and “modular”

The IAEA calls reactors up to about 300 MWe small, and up to about 700 MWe medium. For comparison, an EPR delivers about 1,650 MWe. “Modular” refers to construction: large components or whole reactor modules built in a factory, then trucked, railed or shipped to the site. Below roughly 20 MWe, designs are usually called microreactors. See SMR in the glossary.

Small reactors are not new. Submarines and icebreakers have used compact reactors since the 1950s, and early civil plants were often under 300 MW. What is new is the pitch: standardised designs, passive safety and serial production to bring costs down, much as solar panels and batteries got cheaper through mass manufacturing.

The main SMR families

Family Coolant Example designs Size (MWe) Typical fuel
Light-water SMRs Water GE-Hitachi BWRX-300, NuScale VOYGR, Rolls-Royce SMR (~470), Holtec SMR-300, CNNC Linglong One 50–470 Standard 3–5% LEU
High-temperature gas Helium China HTR-PM, X-energy Xe-100 80–210 TRISO, HALEU for Xe-100
Sodium-cooled fast Liquid sodium TerraPower Natrium (345, with salt storage) 300–500 HALEU
Molten salt Fluoride or chloride salts Kairos Hermes (test), Terrestrial IMSR 15–200 Various
Microreactors Heat pipes, gas Westinghouse eVinci, Oklo Aurora, BWXT BANR 1–20 HALEU / TRISO

Light-water designs are the nearest to market because they reuse proven fuel, materials and regulatory experience from the existing fleet; see how a nuclear reactor works. The more exotic designs promise higher temperatures for industrial heat and better fuel use, but they need HALEU fuel that is in short supply (see the fuel cycle guide).

What is actually running or being built

Russia: Akademik Lomonosov. A barge carrying two 35 MWe KLT-40S reactors, derived from icebreaker designs. It began commercial operation in May 2020 at Pevek in Chukotka, supplying power and heat to an isolated Arctic grid. Russia also runs RITM-200 reactors on its newest icebreakers.

China: HTR-PM and Linglong One. The HTR-PM at Shidaowan, two helium-cooled reactor modules driving one 210 MWe turbine, entered commercial operation in December 2023. It runs at low output compared with its size but showed that the pebble-bed concept works at grid scale. The Linglong One (ACP100), a 125 MWe pressurised water SMR at Changjiang on Hainan Island, started construction in July 2021 and was approaching start-up in 2026, which would make it the first land-based commercial light-water SMR.

Canada: BWRX-300 at Darlington. Ontario Power Generation received a licence to construct the first BWRX-300 from the Canadian Nuclear Safety Commission in April 2025, and the Ontario government approved a four-unit project in May 2025 with a total budget of about C$20.9 billion. The first unit, around C$6.1 billion plus about C$1.6 billion of shared infrastructure, is targeted for about 2030. The Tennessee Valley Authority plans the same design at Clinch River, and Poland’s OSGE has selected it.

United States: NuScale and others. NuScale’s 50 MWe module was the first SMR to receive US design certification (2023), and an uprated 77 MWe version received NRC standard design approval in May 2025. Its flagship six-module project for the Utah Associated Municipal Power Systems was cancelled in November 2023 when the estimated cost rose to about $9.3 billion for 462 MWe. TerraPower began non-nuclear construction of Natrium in Wyoming in 2024, and Kairos Power is building its Hermes test reactor in Tennessee.

United Kingdom and Europe. Great British Nuclear selected Rolls-Royce SMR in 2025 for the first UK fleet, and several Central European utilities have signed agreements with GE-Hitachi, NuScale or Rolls-Royce. None has started nuclear construction.

The cost question

The core economic problem is simple. Large reactors exist because many costs scale less than linearly with size: a reactor four times bigger does not need four times the concrete, staff or licensing effort. Small reactors give up that advantage and hope to win it back through factory production, shorter construction times and lower financing costs.

Overnight cost per kW = project cost ÷ net capacity

BWRX-300 at Darlington, unit 1 (incl. share of site works): ≈ C$7.7 billion ÷ 300,000 kW ≈ C$25,700/kW (first-of-a-kind) Four-unit average: C$20.9 billion ÷ 1,200,000 kW ≈ C$17,400/kW

NuScale UAMPS (2023 estimate): $9.3 billion ÷ 462,000 kW ≈ $20,100/kW Vogtle 3–4 (AP1000, final): > $30 billion ÷ 2,234,000 kW ≈ $13,500/kW or more

These are early, first-of-a-kind figures with different scopes (some include financing, grid works and owner’s costs), so treat comparisons with care. The direction is clear, though: early SMRs cost at least as much per kW as large reactors. The learning effect only appears if dozens of identical units get built. That is how capital-heavy plants like these behave in a levelised cost model, where financing and construction time dominate. Try different overnight costs and discount rates in the LCOE calculator.

Where SMRs could fit

  • Replacing coal units. A 300 MW reactor can reuse the grid connection, cooling water and site of a retired coal unit of similar size. US and Polish studies have looked at this closely.
  • Remote and island grids where diesel costs $0.25–0.50/kWh, such as mines, Arctic towns or island systems.
  • Industrial heat and hydrogen. High-temperature designs can supply 500–750 °C steam to chemical plants.
  • Data centres. Google, Amazon and Microsoft signed agreements in 2024–2025 with Kairos, X-energy and others, and with existing large plants, for future nuclear supply.
  • Smaller grids where a 1,100 MW unit would be too large a share of the system for safe operation (a rule of thumb keeps the largest unit below roughly 10% of peak demand).

Southeast Asia’s interest

No Southeast Asian country runs a nuclear power plant today, but several are studying SMRs as a way to add firm low-carbon power without large single units.

  • Philippines. Executive Order 164 (2022) added nuclear to the energy mix, and the US–Philippines civil nuclear agreement (the “123 agreement”) entered into force in 2024. Meralco and other utilities have signed studies with NuScale and others.
  • Indonesia includes nuclear in its long-term plans, with a first plant pencilled in for the early 2030s, and has studied designs from ThorCon and NuScale.
  • Vietnam decided in late 2024 to revive its Ninh Thuan nuclear project, cancelled in 2016.
  • Malaysia dissolved its nuclear power agency in 2018 but has since said nuclear, including SMRs, is being studied as an option after 2035. Public acceptance, regulation and siting are open questions. See the Malaysia profile, Malaysia’s energy transition and energy in Southeast Asia.

A balanced verdict

  • SMRs are real hardware now, but only Russia and China have them running, and the first Western units are due around 2029–2032.
  • Early costs per kW are at least as high as large reactors; savings depend on serial orders that do not exist yet.
  • Their strongest cases are coal-site replacement, remote grids, industrial heat and small power systems.
  • Planning on SMRs to deliver large volumes before the mid-2030s is a bet, not a forecast.

For reactor data by country, see the nuclear energy hub, the nuclear technology page and Nuclear Power: Pros and Cons.

Calculators for this topic

Frequently asked questions

What counts as a small modular reactor?
The IAEA defines small reactors as those up to about 300 MWe, against 1,000–1,700 MWe for large modern units. Modular means major components are built in factories and shipped to site. Reactors below roughly 20 MWe are usually called microreactors.
Are any SMRs operating today?
A few. Russia's floating plant Akademik Lomonosov (two 35 MWe reactors) has supplied the Arctic town of Pevek since 2020, and China's HTR-PM high-temperature gas-cooled demonstration (about 210 MWe) entered commercial operation in December 2023. China's Linglong One (125 MWe) was close to start-up in 2026. No Western-designed SMR was generating power by late 2026.
Are SMRs cheaper than large reactors?
Not yet, and maybe not per kW. Small units lose economies of scale, and first-of-a-kind projects have come in expensive: NuScale's Utah project was cancelled in 2023 after its target power price rose to $89/MWh, and Ontario's first BWRX-300 is budgeted at about C$7.7 billion including shared site costs. Savings depend on building many identical units.
Why was the NuScale Utah project cancelled?
The Carbon Free Power Project for the Utah Associated Municipal Power Systems (UAMPS) was cancelled in November 2023 because not enough utilities signed up after the estimated cost rose to about $9.3 billion for 462 MWe and the target price jumped from $58 to $89/MWh. NuScale's design itself remains licensed by the US NRC.
Is Malaysia planning to build an SMR?
Malaysia has no SMR project. The government has said it is studying nuclear power, including small reactors, as a possible option after 2035, and the Philippines and Indonesia are further along with policy work and vendor agreements. Any plant would need a regulator, a legal framework and public acceptance first.
Do SMRs produce less nuclear waste?
Not necessarily. Smaller cores leak more neutrons, and a 2022 Stanford-led study in PNAS estimated some SMR designs could produce more waste per unit of energy than large reactors. Results depend heavily on the design and fuel.

Sources & references

  1. IAEA — Advances in Small Modular Reactor Technology Developments (2024 edition)
  2. Ontario Power Generation — Darlington New Nuclear Project
  3. U.S. NRC — NuScale US600 and US460 design certification and approval
  4. World Nuclear Association — Small Nuclear Power Reactors
  5. Krall, L., Macfarlane, A., Ewing, R. (2022). Nuclear waste from small modular reactors. PNAS 119(23)

Figures in this guide are typical ranges drawn from the sources above; your own numbers depend on location, prices and equipment. Use the linked calculators for site-specific estimates.