What are small modular reactors? A beginner’s guide to SMRs

A small modular reactor (SMR) is a nuclear power plant designed to be built in a factory and assembled on site, rather than constructed piece by piece as a bespoke project. The reactors are smaller than conventional ones, and the intention is that building many identical units drives costs down in a way that one-off megaprojects never have. 

The idea is not new. What has changed is that several designs have moved from paper to licensing and construction, while demand for firm, low-carbon capacity has risen sharply. Whether the economics work at scale remains unproven. 

How SMRs differ from traditional nuclear reactors 

Three differences matter. 

1. Output per unit is the most obvious. A conventional reactor typically generates between 1,000 and 1,600 MW. An SMR is defined as producing up to around 300 MW, though several commercially prominent designs sit somewhat above that threshold while retaining the modular approach. 

2. Fabrication is the more consequential difference. Conventional nuclear is built on-site, with components manufactured individually and assembled in place – an approach that has led to repeated cost overruns and schedule slippage across recent European projects. SMR designs move most of that work into factory conditions, delivering modules to site for assembly. The argument is that factory production allows learning across units, whereas each bespoke build effectively starts again. 

3. Passive safety is the third. Many designs rely on natural circulation and gravity rather than powered systems to remove heat after shutdown, reducing the equipment required to maintain safety and potentially shrinking emergency planning zones, thereby widening the range of viable sites. 

The trade-off is that a smaller reactor produces less output for a similar level of regulatory work, licensing effort and site infrastructure. SMR economics depend entirely on whether series production compensates. On the first unit, it does not. 

Published cost estimates reflect that uncertainty. Levelised cost benchmarks for SMR designs span a wide range, broadly from $60 to $120 per MWh, depending on design, financing assumptions, and whether the unit is a first-of-a-kind or a later one in a series. The width of that range is itself the story: at the bottom end, SMRs are competitive with most firm alternatives, and at the top end they are not. 

Why interest in SMRs is accelerating in 2026 

Four forces have converged. 

Firm capacity has become scarcer and more valuable. As thermal plants retire and renewable share rises, systems need generation that is available regardless of weather. Capacity markets have made that value explicit, as covered in What are capacity markets? How countries pay to keep the lights on 

Data centre demand has changed the buyer set. Hyperscalers seeking round-the-clock carbon-free power have collectively committed to more than 10 GW of new nuclear capacity globally, which introduces creditworthy offtakers willing to sign long-term contracts - historically the missing ingredient in nuclear financing. That demand growth is examined in What is driving the surge in AI data centre power demand? 

Energy security has moved firmly up the agenda. Domestic generation by firms looks considerably more attractive to European governments than it did before the gas crisis, and the policy environment has shifted accordingly. 

Government funding has followed. Several European states have committed direct capital and, more importantly, established financing structures that share construction risk - the single largest barrier to nuclear investment. 

Who is building them, and where 

Deployment is further advanced than the technology’s reputation suggests. 

In Canada, GE Hitachi’s BWRX-300 at Darlington in Ontario became the first grid-scale SMR to enter construction in a G7 country, following the granting of a licence to construct in 2025. It is the closest the sector has to a live demonstration of the delivery model. 

In the UK, Rolls-Royce SMR was selected as preferred technology partner by Great British Energy - Nuclear, with a contract signed to deliver the country’s first three units at Wylfa on Anglesey. The design has been progressing through the Generic Design Assessment process, and the company has also secured commitments in Czechia, where it is working with ČEZ towards deployment at Temelín, and in Sweden. 

In the United States, TerraPower’s Natrium project in Wyoming has secured a construction permit, and several designs are progressing through a regulatory framework that has been revised to accommodate advanced reactors. 

The Nordics are following a distinct path, with Sweden reforming its financing model to support new nuclear and Finland pursuing smaller reactors for district heating as well as electricity. 

China is furthest ahead in absolute terms, with the Linglong One expected to become the world’s first land-based commercial SMR in operation. 

What SMRs could mean for grid reliability and baseload power 

The system value of a firm low-carbon capacity is easy to assert but harder to quantify. Three effects are worth distinguishing. 

The first is adequacy. Firm capacity available during still, cold periods is precisely what high-renewable systems struggle to provide, and it is what capacity mechanisms are procuring at an increasing cost. 

The second is the forward curve. Adding significant inflexible low-carbon output alters the shape of the merit order, displacing gas from more hours and reducing the frequency with which gas sets the price. This compresses the average price level while doing relatively little to reduce the peaks that occur when the system is genuinely tight. 

The third aspect is balancing. Nuclear power has traditionally operated steadily, but in a system with high renewable energy, this becomes more problematic. An overly windy system combined with inflexible nuclear plants can lead to oversupply, which in turn causes negative power prices, as explained in What are negative power prices? Why they happen and who benefits. 

Several SMR designs claim meaningful load-following capability, and some pair generation with thermal storage so that output to the grid can vary while the reactor runs steadily. If that works as advertised, it would address the main system-integration objection to nuclear in renewable-heavy markets. It has not yet been demonstrated commercially at scale. 

Siting is a further consideration that gets less attention than it deserves. Smaller units with reduced emergency planning requirements can in principle be built closer to demand, including on retired coal and industrial sites where grid connections, cooling water and transmission infrastructure already exist. Reusing an existing connection point is a substantial advantage in markets where new connections are the binding constraint. 

Not all of the output need be electricity. Several designs target industrial process heat or district heating, applications that are difficult to decarbonise by other means and that value steady output rather than flexibility. Finland’s interest in smaller reactors for heating illustrates the point: the value case does not depend on the power market at all. 

The challenges still standing in the way 

The obstacles are substantial and mostly not technical: 

• Licensing timelines, since regulators assess each design thoroughly and processes designed for large bespoke reactors do not straightforwardly accommodate a fleet approach 

• First-of-a-kind cost, where the first units carry the full weight of design, licensing and factory establishment, and current cost estimates span a wide range 

• Supply chain, which contracted severely during the decades when little nuclear was built, particularly for large forgings and specialist components 

• Fuel, with several advanced designs requiring high-assay low-enriched uranium whose Western production capacity remains well short of projected demand 

• Public consent, which varies enormously by country and can determine whether a project proceeds regardless of its economics. 

Fuel deserves particular attention because it is the constraint least amenable to a quick fix. Enrichment capacity takes years to build, and reactors requiring higher enrichment cannot operate without it. Several otherwise well-advanced projects are exposed to a supply chain that does not yet exist at the required scale. 

The deeper commercial question is whether the modular premise holds. The cost case depends on building many identical units and improving with each one. That requires a committed order book large enough to justify factory investment before the first unit has demonstrated anything – which is exactly the chicken-and-egg problem the sector has been trying to solve for fifteen years. 

Conclusion 

SMRs are no longer a concept. Units are under construction, designs are progressing through the licensing process, and governments and technology companies have committed real capital. 

What remains unproven is the proposition on which the whole model rests: that factory production and series learning can deliver nuclear at a cost conventional construction never achieved. The first units will not demonstrate this, because first units never do. The evidence will come from whether the second and third are meaningfully cheaper and faster. 

Three markers are worth tracking. Final investment decisions indicate genuine commitment rather than optionality. Regulatory milestones determine whether licensing can move at the pace deployment schedules assume. And first power dates, compared against the dates originally announced, will show whether this generation of nuclear has broken the industry’s long record of optimistic scheduling. 

Until then, the honest position is that SMRs are a credible answer to a real problem in a system that badly needs firm low-carbon capacity, priced on a set of assumptions that nobody has yet tested.