Small Modular Reactors (2026): Why SMRs Have Not Delivered

Small modular reactors (SMRs) are nuclear reactors of up to about 300MW each, built largely from factory-made modules, and as of September 2026 they have not yet delivered cheaper nuclear power, or any commercial power in the United States. The leading US project at the time, NuScale's 462MW plant in Idaho, was cancelled in 2023 after its estimated cost rose from $5.3 billion to about $9.3 billion; China's 125MW Linglong One missed its target of starting commercial operation in the first half of 2026; and NRC construction permits have so far gone to Kairos Power's Hermes reactors (2023 and 2024), TerraPower's Natrium plant in Wyoming (March 2026) and TVA's BWRX-300 in Tennessee (September 29, 2026). Their realistic role in the US grid is a 2030s story.

What makes a reactor "small" and "modular"?

Conventional nuclear power plants are large. A typical pressurized water reactor produces 1,000–1,600 MW of electricity and takes 10–20 years and $10–$30 billion to build. Small modular reactors, by definition, produce no more than 300 MW of electrical output, and most designs target 50–175 MW. The "modular" part refers to factory fabrication: instead of building a unique, custom reactor on-site, SMR components are manufactured in a factory to standardized specifications and shipped to the site for assembly.

The premise is compelling. Factory manufacturing should improve quality control, reduce construction time, and allow cost reductions through repetition, the same logic that made commercial aviation economical over time. Smaller size should reduce financial risk, since a developer commits a few billion dollars rather than $30 billion. And SMRs' smaller footprint should make them deployable on sites unsuitable for conventional nuclear plants.

Whether that premise holds in practice is still unproven, and the first commercial units are only now arriving.

Is there a working commercial SMR?

China's ACP100 reactor, branded "Linglong One," is a 125 MWe pressurized water reactor under construction at Changjiang, Hainan Province. It finished its non-nuclear commissioning tests in December 2025 with commercial operation targeted for the first half of 2026 (World Nuclear News); that date passed, and in September 2026 World Nuclear News reported it was still undergoing pre-commissioning tests (WNN). Once running, it would be one of the first commercial land-based small modular reactors in the world. (Russia's floating Akademik Lomonosov plant, which uses two small reactors, has supplied power since 2020.)

The ACP100 was designed by the China National Nuclear Corporation specifically as an SMR: integral reactor design (steam generators inside the reactor vessel rather than external), passive safety systems, and compact size. Construction began in 2021.

The ACP100's performance data: its actual construction time, final cost, and operational reliability once it comes online, will be closely watched by the global nuclear industry. China has the regulatory framework, industrial supply chain, and financing conditions to execute nuclear projects efficiently. Whether the ACP100's economics prove out SMR theory in a favorable environment will say a great deal about what is achievable elsewhere.

What is the Rolls-Royce SMR?

Rolls-Royce SMR is developing a 470 MWe reactor design (larger than most SMR definitions, sometimes called an "advanced modular reactor") for deployment in the United Kingdom. The state-owned Great British Energy - Nuclear picked it as preferred bidder in June 2025, and in November 2025 the government named Wylfa, on the island of Anglesey in North Wales, as the first site, for three units (World Nuclear News).

The two signed a contract for the Wylfa units in April 2026, when the design was in the third and final step of the UK regulators' Generic Design Assessment (WNN). The government aims for Wylfa's reactors to supply power from the mid-2030s, with a final investment decision expected in 2029.

Why did NuScale cancel its SMR project?

The US SMR landscape suffered a significant setback in November 2023 when NuScale Power cancelled its VOYGR project, a 6-module, 462 MWe plant planned for the Idaho National Laboratory site, citing cost overruns. Project costs had escalated from initial estimates of $5.3 billion to revised estimates approaching $9.3 billion, and the Utah Associated Municipal Power Systems utility consortium that had agreed to purchase the power withdrew when costs made the project uncompetitive against alternatives.

The NuScale cancellation was a genuine blow to US SMR credibility. NuScale had spent over a decade developing its design, received the first SMR design approval from the US Nuclear Regulatory Commission in 2020, and was widely considered the most advanced US SMR project. The cost escalation reflected real challenges: NuScale's factory-fabricated modules were larger than easy transportation permitted, requiring field assembly that limited the efficiency gains of factory manufacturing.

NuScale has kept going. In May 2025 the NRC approved an uprated design with 77MW modules (Federal Register), and in September 2025 TVA and NuScale's partner ENTRA1 Energy announced a framework to deploy up to 6GW of NuScale plants in TVA's region, with ENTRA1 financing and owning the plants and selling the power to TVA (NuScale).

Which US companies are still developing SMRs?

TerraPower: Founded with backing from Bill Gates, TerraPower is developing the Natrium reactor, a 345 MWe sodium-cooled fast reactor with a molten salt thermal energy storage system that allows it to vary output to complement wind and solar. In March 2026 the NRC issued the construction permit for its Kemmerer, Wyoming plant, which DOE called the first ever issued for a commercial non-light-water power reactor, and DOE lists 2030 as the target for operation (DOE). Nuclear construction started in April 2026 (ANS). In September 2026 TerraPower and its construction contractor, Bechtel, parted ways; TerraPower said it would rebid the contract and did not expect the schedule to change (World Nuclear News).

X-Energy: X-Energy is developing the Xe-100, an 80 MWe pebble bed high-temperature gas reactor. Its first customer is Dow, which applied in March 2025 for an NRC construction permit for four Xe-100s to supply power and steam at its Seadrift, Texas site, a potentially important use case since industrial heat is hard to decarbonize. X-energy's mid-2026 SEC filing expects that permit in the first quarter of 2027 and its first commercial reactor delivery in the early 2030s; DOE's Advanced Reactor Demonstration Program shares half of up to $2.4 billion in eligible costs. Amazon is backing a separate four-reactor, 320MW project with Energy Northwest in Washington State (X-energy 10-Q).

Kairos Power: Developing a fluoride salt-cooled high-temperature reactor. Has received NRC approval to construct a test reactor called Hermes at Oak Ridge, Tennessee, the first non-water-cooled reactor approved for US construction in decades. Hermes will not generate commercial electricity but will validate the technology before commercial deployment. Its follow-on, Hermes 2, received a construction permit in November 2024 as the first power-producing Gen IV reactor permitted by the NRC, has broken ground, and is designed to supply up to 50MW to TVA's grid for Google data centers (Kairos Power).

TVA: On September 29, 2026 the NRC issued TVA a construction permit for a 300 MWe GE Vernova Hitachi BWRX-300 at its Clinch River site in Oak Ridge, Tennessee, the first US permit for that design; TVA still needs a separate operating license before it can load fuel (GE Vernova). In December 2025 DOE selected TVA and Holtec (for two SMR-300 units at Palisades in Michigan) for up to $400 million each in cost-shared funding (World Nuclear News). The first BWRX-300 is being built at Ontario Power Generation's Darlington site, licensed for construction in April 2025, with commercial operation targeted by the end of 2030 (OPG).

Test reactors: A May 2025 executive order set DOE a goal of getting at least three advanced test reactors to criticality by July 4, 2026. Four small units, from Antares, Valar Atomics, Deployable Energy and Aalo Atomics, got there by the deadline (World Nuclear News), but these are test reactors, not plants selling power.

How much do small modular reactors cost?

Nuclear economics operate on a fundamental tension: nuclear power plants have high upfront capital costs but low fuel and operating costs, giving them a levelized cost that depends heavily on discount rates and construction risk. The history of Western nuclear construction, particularly the Vogtle Units 3 and 4 in Georgia, which came online in 2023 and 2024 at a total cost near $35 billion by the Associated Press's count, versus an original $14 billion estimate (EIA), demonstrates how badly construction cost overruns can destroy the economics of nuclear projects.

SMRs theoretically solve this by making nuclear projects smaller and faster to build. But the cost advantage has not yet materialized in practice:

• Small reactors lose the economies of scale of larger plants, you need more reactors, each with their own control room, safety systems, and regulatory compliance overhead, to produce equivalent output
• Factory manufacturing savings depend on high production volumes, which require a large market, which does not yet exist
• Regulatory costs are substantial regardless of reactor size
• Construction labor costs in Western countries are high and difficult to reduce through modularization

First-of-a-kind reactors are expected to cost considerably more per kilowatt than later units. DOE's 2024 advanced nuclear Liftoff report modeled early units at overnight capital costs of $7,000 to $20,000 per kW and expects repeat builds of one design to drive costs down substantially. That is several times the roughly $1,700 per kW that recent US onshore wind projects have cost (Berkeley Lab). Nuclear's case therefore rests on factors other than capital cost: 24/7 availability and a small land footprint.

Why does nuclear still matter?

The argument for nuclear is not that it is cheap. It is that the grid needs 24/7 carbon-free power, and there are not many ways to provide it. Hydropower is the main existing source of dispatchable clean electricity, but it is geographically limited and largely already built out. Geothermal baseload is confined to specific geologies. Long-duration storage is improving but not yet scalable enough to firm a winter grid entirely powered by wind and solar.

Nuclear's land footprint is also dramatically smaller than renewable energy. The Energy Department says a typical 1,000 MW US nuclear plant needs a little more than 1 square mile, and cites the Nuclear Energy Institute's estimate that wind farms need 360 times as much land area to produce the same electricity (DOE), though most of the land inside a wind farm stays usable for farming or grazing. In densely populated regions or areas with land constraints, this matters.

SMRs add two capabilities conventional nuclear lacks: the ability to follow load (some designs can ramp output more quickly than large plants) and deployment in smaller increments. A utility can add 100 MW of SMR capacity, then add another 100 MW a few years later as demand grows, rather than committing to a 1,600 MW project all at once.

When will SMRs matter for the US grid?

For SMRs to be commercially significant in US electricity generation, the realistic timeline is the mid-to-late 2030s, assuming regulatory approvals proceed, Linglong One and other first movers demonstrate favorable economics, and construction supply chains develop. This is not a technology that will materially affect electricity supply in the next 5 years. For the 2030 grid, wind, solar, and storage are the dominant story. For the 2040 grid, SMRs may play a meaningful role in providing firm clean power to complement variable renewables.

Related reading: Tidal and wave energy · Onshore vs offshore wind energy · Geothermal energy explained · The future of home energy