The first test for new Small Modular Reactors (SMR)
Abstract
This article takes data surrounding last week’s announcement of the first SMR farm in Ontario, Canada to assess whether this new approach to nuclear shows any signs of reducing costs below the level of larger reactors and renewable generators. In brief, the conclusion is that it does not.
SMRs
A small number of experimental SMRs have been built or are under construction in Russia, China and Argentina.[1] The history of these reactors has been full of delays and huge cost overruns.
Nevertheless recent months and years have seen an explosion of interest in constructing new reactors, often to serve specific customers with a high and consistent power need such as the operation of data centres. The International Atomic Energy Agency reported that there are 80 different designs and design concepts for new types of SMRs around the world although no formal decisions had then been taken to proceed with any new projects. Many commentators see SMRs as a realistic route to decarbonising electricity production; in its recent report, the UK’s Tony Blair Institute wrote that ‘The new generation of small modular reactors offers hope for the renaissance of nuclear power’.[2]
We saw a big step forward last week with the final governmental approval given for the construction of the first SMR farm at an existing nuclear reactor site in Ontario, Canada. The information released when the announcement was made gives us useful information about the possible cost of SMRs. Despite the upbeat press releases from participants in the project, the numbers provided will not increase optimism about the future of this route to a zero carbon energy system.[3]
The plans
The Darlington project plans to eventually put 4 SMRs, each of 300 MW capacity, onto the site. The Ontario government estimates the project’s total cost at 20.9 billion Canadian dollars (CAN$) and is supporting it financially. Expressed in US dollars, the price is about $15bn at early May 2025 exchange rates for a total of 1.2 gigawatts of electricity.
The technology provider is GE Hitachi using its BWRX 300 boiling water reactor, a simplified version of the company’s full scale nuclear plants. The BWRX 300 has been identified as one of the lowest cost competitors in the race to dominate the SMR industry and the design is one of those currently being evaluated in the UK as a potential recipient of substantial government support. GE Hitachi claims numerous advantages over other SMR technologies, including lower steel and concrete costs, passive cooling and the use of well-understood and widely available nuclear fuel.
Groundwork has been underway for more than 2 years near the existing Darlington nuclear power station in readiness for the formal safety approval finally given in April 2025. Construction is projected to be completed in 2030.[4] In a challenge to the widespread view that SMRs will be simple to construct, one of the challenges facing the first Darlington reactor is creating a tunnel to carry cooling water that will stretch for 3.4 kilometres with a diameter of 6 metres.[5]
The costs of building and operating the SMRs in Ontario.
Capital cost
The first reactor to be built is projected to have a cost of 6.1bn CAN$ (about US$4.4bn). Additional infrastructure will be needed at the site that will be shared between this SMR and the three equivalents that will be eventually built there. The capital cost of these shared facilities is separately budgeted at 1.6bn CAN$ (about US$1.15bn).
The total budget for the entire 4 reactor site is 20.9bn CAN$ (about US$15bn). So the three SMRs to be installed after the first reactor are projected to cost a total of about 13.2bn CAN$, or 4.4bn CAN$ each, a roughly 28% reduction on the ‘first of a kind’ (FOAK).
The cost per kilowatt of capacity for the first reactor (excluding the price of the shared services is about 20,300 CAN$, or US$14,600. Taken together, all four come in at about US$12,900.
Just a year ago, the Institute for Energy Economics and Financial Analysis (IEEFA) published a report on SMRs that showed 2020 forecast for a kilowatt of capacity for a BWRX 300 of just under US$2,900 for an ‘nth of a kind’ reactor.[6][7] This number was expressed in 2023 dollars to account for inflation. The suggested cost of the first Darlington, Ontario reactor is thus about five times the projected NOAK cost in 2020, just five years ago. (However this comparison could be considered unfair to the GE Hitachi reactor because it compares NOAK and FOAK budgets).
The IEEFA report also provides high and low cost estimates from 2023 for the BWRX 300 reactor from external sources. The low figure is just over US$7,400 and the high number is around US$12,350. Thus the Ontario cost for the first of the four reactors is almost 19% above the ‘high’ estimate from just two years ago and almost double the ‘low’ forecast.
As an aside, GE Hitachi in 2020 was estimating its reactor could be constructed in 24-36 months. The building work at the Darlington site commenced in autumn 2022 and is projected to finish in 2030, and therefore is scheduled to take at least 7 years.
Costs compared to renewables
How do the costs per kilowatt of capacity compare to solar and wind equivalents? The Ontario government says that it estimates that the 1.2 GW to be eventually provided in SMR capacity could be replaced by about 8.9 GW of solar and wind power. As far as I can see, it has not split this number between the two renewable sources of electricity.
In the UK, and using capacity factors appropriate to southern England, it would take about 10 GW of solar farms to create an equivalent amount of annual electricity to 1.2 GW of nuclear.[8] Given that Darlington, Ontario sits at a latitude of about 43 degrees north – equivalent approximately to Marseille - compared to London, England at 51 degrees average solar productivity will be probably be higher at the Canadian location.
So let’s assume that 8.9 GW of solar PV in Ontario would provide the same amount of power as the Darlington reactors. This would cost no more than about US$9bn in the UK today, and probably much less by the time the reactors are constructed. In other words, solar alone would be less than half the capital cost of the four SMRs to provide the same amount of electricity. (However the province of Ontario does make the valid point that solar PV would probably require far more new grid capacity. It would also need long term storage). But the conclusion has to be that SMRs, at least as shown by the GE Hitachi prices, are not likely to be cheaper to construct than solar or wind.
Comparison with other nuclear sources
Equally important, are SMRs going to be cheaper than full-scale large nuclear reactors?
The new European Pressurised Reactor (EPR) at Flamanville on the Normandy coast of France was connected to the grid in late 2024. The estimated cost of this reactor is about €19.1 billion, or about US$21.5 million.[9] At maximum capacity it produces just over 1.6 gigawatts of electricity, compared to the 1.2 GW projected for the four SMRs at Darlington, Ontario.
Thus the hugely expensive and long-delayed reactor at Flamanville, the fourth EPR to be completed in the world, was cheaper to construct than the estimates for the first SMR at Darlington and only marginally more expensive than the projected costs for the eventual group of 4 reactors when expressed as capital cost per kilowatt of capacity.
Do SMRs offer savings in operating costs over large nuclear power stations and renewable plants?
Most sources suggest that SMRs will have operating costs at roughly the same level as large nuclear plants per unit of capacity. However we cannot know this with any more certainty than the comparison of capital costs.
One careful piece of academic research put it as follows: ..it is expected that O&M and fuelling costs will be very similar to that of LRs.[10] This conclusion was reached some years ago and estimates may now be out of date. However I have seen no data suggesting that SMRs will be cheaper to operate than larger nuclear power plants.
The operating costs include nuclear fuel, which needs to be regularly but infrequently fed into the reactor. At current uranium prices, fuel is unlikely to be a significant portion of total costs.
More important is the cost of staff and here the limited evidence is that SMRs will require more people than larger reactors. Ontario predicts that 2,500 people will work on the SMRs at the Darlington site when the 4 reactors are complete. By comparison, Sizewell B, an existing nuclear power station in the UK, employs about 900 workers and has approximately the same annual output as the 4 reactor Canadian site will have.[11]
However if we optimistically assume that the SMRs will cost the same to run as larger nuclear sites how will they compare with solar? The US Energy Information Agency writes that the cost of running a nuclear power plant is around $22 per megawatt hour of output, or 2.2 cents per kilowatt hour. Fuel is about 0.61 cents per kilowatt hour with operations at around 0.95 cents and maintenance at 0.64 cents.
Most estimates for the operating costs of full-size solar parks are around US$15 per kilowatt of capacity per year. A solar park in Ontario is likely to produce at least 900 kWh per kilowatt annually, implying a cost of around 1.67 cents per kilowatt hour for total operating costs. The conclusion has to be that SMRs are therefore very unlikely to be cheaper to run than large solar farms in the same area as the new nuclear site.
Summary
The data provided by the press releases announcing the world’s first new generation SMR park give us some information about costs. Of course we cannot know whether these estimates will be correct. But If they are actually achieved, SMRs will be only marginally cheaper than the current generation of large nuclear power stations. Operating costs will be similar to their larger cousins but possibly much more.
Solar and wind parks are likely to be less expensive in both capital and operating terms. Capital costs for solar may be no more than half the figure for the Darlington project and operating costs around three quarters of the level of SMRs.
By itself, this information does not imply that the rush towards SMRs is misplaced. They may offer cheaper grid connections and will be better at providing power directly to 24 hour electricity customers. But the arguments advanced by commentators such as the Tony Blair Institute need to adjusted recognise that there is no evidence today that SMRs will reduce electricity costs compared to continuing rapid investment in wind and solar.
May 11th 2025
[1] https://www.iaea.org/topics/small-modular-reactors
[2] https://institute.global/insights/climate-and-energy/the-climate-paradox-why-we-need-to-reset-action-on-climate-change
[3] https://news.ontario.ca/en/release/1005889/ontario-leads-the-g7-by-building-first-small-modular-reactor
[5] Same as reference 3 above.
[6] https://ieefa.org/sites/default/files/2024-05/SMRs%20Still%20Too%20Expensive%20Too%20Slow%20Too%20Risky_May%202024.pdf
[7] The 2020 figure appears to have been derived from this GE Hitachi web page and adjusted for inflation - https://www.gevernova.com/content/dam/gepower-nuclear/global/en_US/documents/product-fact-sheets/GE%20Hitachi_BWRX-300%20Fact%20Sheet.pdf
[8] Assuming a capacity factor of 90% for the SMR and 11% for UK solar.
[9] https://www.lemonde.fr/les-decodeurs/article/2024/05/09/les-derapages-de-l-epr-de-flamanville-en-graphiques-le-cout-multiplie-par-six-la-duree-du-chantier-par-quatre_5480745_4355771.html
[10] https://www.sciencedirect.com/science/article/abs/pii/S0301421517300538?fr=RR-2&ref=pdf_download&rr=93da6ca1cd7fe1fc
[11]https://www.bbc.com/news/articles/c93qz4dlqlgo#:~:text=Work%20on%20the%20construction%20of,literally%20from%20the%20ground%20up%22.