Small Modular Reactor Statistics 2026
By Axis Intelligence Research
Co-authors: Aidan Jad | Last updated: August 30, 2026 | License: CC BY 4.0
The first small modular reactor budget ever approved by a G7 government prices four units at CAD 20.9 billion for 1,200 MW. According to Axis Intelligence Research, that works out to USD 12,653 per kilowatt across the fleet, but USD 18,646 per kilowatt for the first unit alone. The first SMR in the West is more capital-intensive per kilowatt than the large reactors it was supposed to undercut.
Quick Answer
A small modular reactor is a nuclear reactor rated up to 300 MWe per module. As of August 2026, the only Western SMR with a government-approved capital budget is Ontario Power Generation’s Darlington New Nuclear Project: CAD 20.9 billion for four BWRX-300 units. According to Axis Intelligence Research, the completed four-unit fleet prices at USD 12,653 per kilowatt, 13.0 percent below the Western large-reactor benchmark, while the first unit alone prices 28.2 percent above it.
Key Findings
- According to Axis Intelligence Research, Ontario Power Generation’s four-unit Darlington SMR fleet carries a capital intensity of USD 12,653 per kilowatt, against a Western large-reactor benchmark of USD 14,543 per kilowatt.
- According to Axis Intelligence Research, the first Darlington unit plus shared site infrastructure prices at USD 18,646 per kilowatt, which is 28.2 percent more capital-intensive per kilowatt than the average of Vogtle 3 and 4, Flamanville 3, Hinkley Point C and Sizewell C.
- Ontario Power Generation’s own budget implies a 27.9 percent per-unit cost reduction between the first BWRX-300 and the average of units two through four, according to Axis Intelligence Research calculations from the published CAD 6.1 billion, CAD 1.6 billion and CAD 20.9 billion figures.
- According to Axis Intelligence Research, the Darlington fleet price is 2.81 times the USD 4,500 per kilowatt that the International Energy Agency identifies as the 2040 cost level at which SMRs would take off in the United States and Europe.
- Western large reactors completed or committed since 2009 have run a median 1.94 times their announced budget and a median seven years past their announced commissioning date, according to Axis Intelligence Research analysis of sponsor and audit-office filings.
How Much Does a Small Modular Reactor Cost per Kilowatt in 2026?
There is exactly one honest answer available in 2026, and it comes from Canada.
Most SMR cost figures in circulation are vendor targets. A target is a marketing artifact. The Darlington New Nuclear Project is different: Ontario Power Generation publishes a budget that a provincial shareholder approved, that a federal regulator is building against, and that private and Indigenous investors have put equity into. It is the first SMR number in the Western world with real money behind it.
The published structure is three-part. The first BWRX-300 unit costs CAD 6.1 billion. Site systems and services shared by all four units cost CAD 1.6 billion. The total for four units is CAD 20.9 billion, delivering 1,200 MW.
That three-part disclosure is unusually useful, because it lets anyone separate first-of-a-kind cost from fleet cost. Most projects never publish the split.
SMR and large-reactor capital intensity, USD per kilowatt
| Project | Country | Capacity (MW) | Published cost | USD per kW | Source |
|---|---|---|---|---|---|
| Darlington, units 2-4 average | Canada | 300 each | CAD 4.4bn per unit (Axis-derived) | 10,655 | Axis Intelligence Research from OPG |
| Vogtle 3 and 4 | United States | 2,228 | USD 23.6bn (Axis-derived) | 10,607 | Axis Intelligence Research from Southern Company 10-Q |
| Darlington, four-unit fleet | Canada | 1,200 | CAD 20.9bn | 12,653 | Axis Intelligence Research from OPG |
| Hinkley Point C | United Kingdom | 3,260 | GBP 35bn (2015 sterling) | 14,648 | Axis Intelligence Research from EDF |
| Sizewell C | United Kingdom | 3,260 | GBP 38bn (2024 prices) | 15,904 | Axis Intelligence Research from Sizewell C |
| Flamanville 3 | France | 1,630 | EUR 23.7bn (2023 euros) | 16,988 | Axis Intelligence Research from Cour des comptes |
| Darlington, first unit plus common | Canada | 300 | CAD 7.7bn | 18,646 | Axis Intelligence Research from OPG |
Currency conversion uses the Federal Reserve H.10 rates for August 21, 2026: 1.3765 CAD per USD, 1.1684 USD per EUR, 1.3644 USD per GBP.
Read the table from the bottom up and the SMR promise inverts. The single most capital-intensive line in this comparison is not a French EPR that took seventeen years. It is the first small modular reactor in the G7.
Aidan Jad’s take: The pitch for SMRs was never really “small.” It was “repeatable.” A 300 MW unit that gets built once is just an expensive reactor with a small nameplate, because the engineering, the licensing basis, the first-of-a-kind supply chain and the site works all get amortised over one-quarter of the megawatts. The economics only arrive on the second, third and fourth pour. Darlington’s budget says exactly this out loud, and it is the most useful thing any SMR sponsor has published.
Source: Axis Intelligence Research — CC BY 4.0
What Is the SMR Cost Parity Index (SCPI)?
Comparing a Canadian SMR to a French EPR requires a fixed reference point. Axis Intelligence Research built one.
SCPI, the SMR Cost Parity Index, is the ratio of a nuclear project’s capital cost per kilowatt to the WLCB, the Western Large-Reactor Cost Benchmark. An SCPI of 1.00 means the project costs the same per kilowatt as a typical recent Western large reactor. Below 1.00 means cheaper per kilowatt. Above 1.00 means more expensive.
Formula
WLCB = (sum of project capital costs in USD) / (sum of net capacity in kW)
SCPI = (project capital cost per kW in USD) / WLCB
WLCB inputs, as of August 21, 2026
| Component | Capacity (MW) | Cost as published | Cost in USD bn |
|---|---|---|---|
| Vogtle 3 and 4 | 2,228 | Georgia Power 45.7 percent share of USD 10.8bn, grossed up | 23.632 |
| Flamanville 3 | 1,630 | EUR 23.7bn, 2023 euros | 27.691 |
| Hinkley Point C | 3,260 | GBP 35bn, 2015 sterling | 47.754 |
| Sizewell C | 3,260 | GBP 38bn, 2024 prices | 51.847 |
| Total | 10,378 | 150.925 |
WLCB = USD 150.925bn / 10,378,000 kW = USD 14,543 per kW.
The components are capacity-weighted rather than averaged, so a 3,260 MW two-unit station counts for more than a single 1,630 MW unit. Each project is carried at the price basis its own sponsor or auditor publishes; no common deflator is applied across 2015 sterling, 2023 euros, 2024 sterling and nominal dollars, because inventing a cross-currency deflator would add a layer of Axis assumption on top of four already-differing accounting conventions. The readings below therefore compare published budgets as published, which is how a lender or a journalist encounters them.
SCPI readings, August 2026
| Project or target | USD per kW | SCPI | Reading |
|---|---|---|---|
| IEA 2040 cost target, United States and Europe | 4,500 | 0.309 | The level at which the IEA expects SMR uptake to accelerate |
| Darlington units 2-4 average | 10,655 | 0.733 | Fleet economics, once first-of-a-kind is behind it |
| Darlington four-unit fleet | 12,653 | 0.870 | The whole programme, as budgeted |
| Western large-reactor benchmark | 14,543 | 1.000 | Baseline |
| Darlington first unit plus common infrastructure | 18,646 | 1.282 | First-of-a-kind penalty, made explicit |
Source for all readings: Axis Intelligence Research, from Ontario Power Generation, Southern Company, EDF, Sizewell C, the Cour des comptes and the International Energy Agency.
These are baseline readings. SCPI enters its first publication cycle here, and future readings will be dated and added rather than backfilled.
What the index is measuring is capital intensity at the point of commitment, not levelised cost of electricity. It does not capture operating cost, capacity factor, fuel, decommissioning provisions, or financing structure, all of which move the eventual price of a megawatt-hour. A project can post a strong SCPI and still deliver expensive power if it runs at a poor capacity factor or carries a high cost of capital, and a RAB-financed project shifts construction risk to consumers in a way that per-kilowatt capital cost cannot show.
How Do SMR Costs Compare to Vogtle, Flamanville, Hinkley Point C and Sizewell C?
The four benchmark projects are the entire recent Western large-reactor record, and each one arrived at its number differently.
Vogtle 3 and 4 are the only new US reactors in a generation. The US Energy Information Administration records the original expectation as USD 14 billion with commercial operation in 2016 and 2017; construction began in 2009. Unit 3 entered service on July 31, 2023 and Unit 4 on April 29, 2024, per Southern Company’s Form 10-Q, which also puts Georgia Power’s share of the total project capital cost forecast at USD 10.8 billion against a 45.7 percent ownership interest.
Axis Intelligence Research grosses that share up to USD 23.632 billion for the two units. This is a conservative floor: owner-level disclosures are stated net of Toshiba settlement proceeds and inter-owner cost-sharing payments, so gross project spend was higher. The gross-up is shown rather than a larger round number precisely because the arithmetic behind it is checkable.
Flamanville 3 is the most expensive line in the set and the best documented. The Cour des comptes put total cost at completion at EUR 20.4 billion in constant 2015 euros, or EUR 23.7 billion in 2023 euros, against an original EUR 3.3 billion estimate. The French audit office raised its own 2020 figure of EUR 19.1 billion by EUR 1.3 billion after adjusting EDF’s provisions and financing-cost treatment. The reactor connected to the grid on December 21, 2024.
Hinkley Point C moved again this year. In EDF’s 2025 annual results, the completion cost is stated at GBP 35 billion in 2015 sterling and Unit 1 production is adjusted to 2030, with a EUR 2.5 billion impairment booked. That is above the GBP 31 billion to GBP 34 billion range EDF published in its January 2024 Hinkley Point C update, and close to double the GBP 18 billion at final investment decision.
Sizewell C is the interesting one, because it is a deliberate attempt to buy down first-of-a-kind cost by copying a plant. Sizewell C reached final investment decision on July 22, 2025 at a capital cost of around GBP 38 billion in 2024 prices for a design that replicates Hinkley Point C.
And here is the finding that ought to travel further than it has. Sizewell C, the replica, prices at USD 15,904 per kilowatt. Hinkley Point C, the original, prices at USD 14,648 per kilowatt. On published figures at the stated price bases, the second-of-a-kind large reactor is not cheaper per kilowatt than the first. Some of that is price-basis drift between 2015 and 2024 sterling and Axis Intelligence Research does not claim the gap survives a full deflation. But the direction is the opposite of the one the fleet-learning argument predicts, and every SMR business case in Europe rests on that same argument.
Aidan Jad’s take: Everyone in this sector says “series effect” as though it were a law of physics. It is a hypothesis, and the only place it has been tested at scale in the West recently is between Hinkley Point C and Sizewell C. Darlington’s budget assumes 27.9 percent per-unit savings from unit one to the later units. That number is a plan, not a result. The first honest test arrives when OPG publishes an actual cost for unit two.
One project deliberately left out of the benchmark
Olkiluoto 3 in Finland is the textbook nuclear overrun, and it is not in the WLCB. Construction started on August 12, 2005 against a 2009 completion target; the 1,600 MW unit entered commercial operation on April 16, 2023, per TVO reporting.
The cost is a different matter. No single sponsor or audit-office figure covers both TVO’s investment and the losses absorbed by the Areva-Siemens consortium under the fixed-price turnkey contract. Widely circulated totals combine a utility’s balance sheet with a vendor’s write-downs, which are not the same kind of number. Axis Intelligence Research uses Olkiluoto 3 in the schedule analysis, where the dates are unambiguous, and excludes it from the cost benchmark, where they are not. The row appears in the dataset marked NA with that reason attached.
How Long Does It Take to Build a Nuclear Reactor, and Are SMRs Faster?
Capital intensity is only half the problem. The other half is that money sits in the ground for a decade and a half before a single kilowatt-hour is sold.
Construction duration, first concrete to grid or commercial operation
| Project | Type | Start | Grid or commercial | Years |
|---|---|---|---|---|
| Linglong One (ACP100) | SMR, 125 MW | July 13, 2021 | 58-month plan | 4.8 (planned) |
| Vogtle 3 | Large, 1,114 MW | 2009 | July 31, 2023 | 14.6 |
| Flamanville 3 | Large, 1,630 MW | December 2007 | December 21, 2024 | 17.0 |
| Olkiluoto 3 | Large, 1,600 MW | August 12, 2005 | April 16, 2023 | 17.7 |
Sources: EIA, Southern Company, Cour des comptes, CNNC, TVO via World Nuclear News.
According to Axis Intelligence Research, the three completed Western large reactors averaged 16.4 years from first concrete to grid, against the 58-month construction period CNNC planned for the ACP100 at Changjiang. That is a ratio of 3.3 to 1.
Two honest caveats belong with that ratio. The Chinese figure is a plan, not an outcome, and 125 MW is a smaller machine than a 1,630 MW EPR. But the schedule gap is large enough that neither point closes it, and schedule is what determines whether a reactor arrives in time to serve a load that exists.
Schedule slip against announced commissioning dates
| Project | Announced target | Actual or current | Slip (years) |
|---|---|---|---|
| Hinkley Point C, Unit 1 | 2025 | 2030 | 5 |
| Vogtle 3 | 2016 | 2023 | 7 |
| Vogtle 4 | 2017 | 2024 | 7 |
| Flamanville 3 | 2012 | 2024 | 12 |
| Olkiluoto 3 | 2009 | 2023 | 14 |
Source: Axis Intelligence Research, from EIA, Southern Company, EDF and Cour des comptes filings.
Median slip is seven years; the mean is 9.0. Axis Intelligence Research reports the median as the headline because Olkiluoto 3 and Flamanville 3 pull the mean upward, and a distribution with two extreme members is better described by its middle than its average. Hinkley Point C is not finished, so its five years is a floor.
How Much Do Nuclear Projects Overrun Their Budgets?
| Project | Announced budget | Latest or final | Multiple | Price basis |
|---|---|---|---|---|
| Vogtle 3 and 4 | USD 14bn | USD 23.6bn | 1.69x | Nominal USD both ends |
| Hinkley Point C | GBP 18bn | GBP 35bn | 1.94x | 2015 sterling both ends |
| Flamanville 3 | EUR 3.3bn | EUR 20.4bn | 6.18x | 2015 euros both ends |
Source: Axis Intelligence Research, from EIA, EDF and Cour des comptes.
According to Axis Intelligence Research, the median escalation multiple across these three is 1.94, and the mean is 3.27. Only comparisons where both the original and the final figure are stated on the same price basis are included, which is why Sizewell C sits outside the calculation: its GBP 20 billion 2020 planning-stage estimate and its GBP 38 billion 2024-price capital cost are not the same kind of number, and dividing one by the other would produce a ratio with no defensible meaning.
That restraint matters for what comes next. A 1.94x median on large reactors is the escalation risk that SMR programmes are being underwritten to avoid. Whether factory fabrication actually removes it is unproven, because no Western SMR has finished.
Which SMR Projects Are Actually Under Construction in 2026?
Pipeline announcements outnumber pours by a wide margin. The IAEA counts more than 80 SMR designs and concepts globally, and four SMRs in advanced stages of construction, in Argentina, China and Russia.
Canada. Ontario Power Generation holds a licence to construct issued by the Canadian Nuclear Safety Commission on April 4, 2025. Regulatory Hold Point 1, covering the reactor building foundation, was cleared on March 30, 2026, and OPG applied for a 20-year licence to operate on March 25, 2026. Grid connection is targeted for the end of 2030. In June 2026 the Williams Treaties First Nations announced a CAD 700 million investment in the project.
United States. No commercial SMR is under nuclear construction. The Tennessee Valley Authority’s BWRX-300 at Clinch River is the furthest along: the US Department of Energy recorded the NRC docketing the first US utility construction permit application for an SMR in July 2025, and the Nuclear Regulatory Commission completed the final supplemental environmental impact statement on April 6, 2026.
United Kingdom. Great British Energy – Nuclear signed its contract with Rolls-Royce SMR on April 13, 2026 for three units at Wylfa delivering at least 1.4 GWe, with GBP 2.6 billion allocated in the 2025 Spending Review and up to GBP 599 million committed by the National Wealth Fund. By June 2026, GBE-N had awarded nearly GBP 900 million in contracts since 2023, more than 70 percent to UK-registered companies.
No capital cost has been published for Wylfa and no final investment decision has been taken, so Axis Intelligence Research computes no cost per kilowatt and no SCPI reading for it. The dataset carries that row as NA with the reason stated. A programme allocation is not a project budget, and treating one as the other is how bad benchmarks get built.
China. Construction of the 125 MW ACP100, known as Linglong One, began at Changjiang on July 13, 2021 on a 58-month plan.
How Much Is the US Military Spending on Nuclear Microreactors?
On August 26, 2026, the US Army announced that its Janus Program had selected five vendors and five installations, awarding up to a combined USD 2.2 billion in partnership with the Defense Innovation Unit.
| Vendor | Installation | State |
|---|---|---|
| Antares Nuclear | Fort Bragg | North Carolina |
| BWXT Advanced Technologies | Fort Campbell | Kentucky |
| General Atomics Electromagnetic Systems | Fort Hood | Texas |
| Radiant Industries | Fort Benning | Georgia |
| Westinghouse Government Services | Fort Drum | New York |
Source: US Army, August 26, 2026.
The funding runs across FY2027 to FY2031 on a milestone-based model, and the Army expects more than 20 microreactors across Department of War installations once private capital is counted alongside the government contribution. The reactors will be contractor-owned and contractor-operated.
According to Axis Intelligence Research, USD 2.2 billion spread across 20 reactors sets an upper bound of USD 110 million in public funding per microreactor, and the true figure is lower because the Army expects more than 20 units. That is a different order of magnitude from grid-scale SMR economics, and it should be read as a bounded subsidy for a first industrial base rather than as a reactor price.
Aidan Jad’s take: The Janus structure is the most interesting procurement design in the sector, and not because of the dollar figure. Milestone-based payments on contractor-owned assets means the Army pays for demonstrated capability instead of hours worked, which is the exact opposite of how Vogtle and Flamanville were contracted. Whether that transfers to a 300 MW grid-connected unit with a utility rate base behind it is the open question. An installation buying resilience against a grid outage has a very different tolerance for cost per kilowatt than a ratepayer does.
How Big Will the SMR Market Get?
The International Energy Agency identifies plans of varying maturity for up to 25 GW of SMR capacity, and expects the first commercial SMR projects to start operation around 2030.
| Scenario | SMR capacity by 2050 | Notes |
|---|---|---|
| Stated Policies | 40 GW | Today’s policy settings |
| Announced Pledges | 120 GW | More than 1,000 reactors |
| Cost parity case | 190 GW | Requires costs to reach large-reactor parity |
Source: International Energy Agency, The Path to a New Era for Nuclear Energy.
On investment, the IEA sees SMR spending rising from under USD 5 billion today to USD 25 billion by 2030, with USD 670 billion cumulative to 2050 in the Announced Pledges Scenario and USD 900 billion in the cost parity case.
The cost parity case is the one that carries the whole argument, and it has a number attached: USD 2,500 per kilowatt in China and USD 4,500 per kilowatt in the United States and Europe by 2040.
According to Axis Intelligence Research, the Darlington fleet price of USD 12,653 per kilowatt is 2.81 times that USD 4,500 target. On SCPI, the IEA’s 2040 target reads 0.309 against today’s Western large-reactor benchmark, meaning SMRs need to reach roughly 31 percent of current large-reactor capital intensity to unlock the 190 GW outcome. Darlington, the best-documented Western SMR in existence, is at 0.870.
That is the gap, stated in one number. The IEA also notes that SMRs can reach cash flow break-even up to ten years earlier than large reactors, which is a real financing advantage that per-kilowatt capital cost does not capture, and it is why lenders are interested despite the arithmetic above.
Aidan Jad’s take: The 120 GW scenario implies more than a thousand reactors averaging under 120 MW each. Set aside cost for a second and look at that as a manufacturing statement. It requires factory throughput, a licensed serial design, and a regulator willing to approve the hundredth unit without re-litigating the first. The IEA is explicit that of the 52 reactors that started construction worldwide since 2017, 25 were Chinese designs and 23 Russian. The West is not currently short of SMR designs. It is short of a production line, and no amount of announced pipeline substitutes for one.
Methodology
Collection. Every figure in this report was retrieved from a primary document during the production session on August 30, 2026. Project costs come from the entity that owns the budget or the audit body that reviewed it: Ontario Power Generation for Darlington, Southern Company’s SEC filings for Vogtle, EDF for Hinkley Point C, Sizewell C for Sizewell C, and the Cour des comptes for Flamanville 3. Regulatory milestones come from the CNSC and the NRC. Defence figures come from the US Army. Scenario data comes from the IEA. Design counts come from the IAEA.
Currency. All conversions use Federal Reserve H.10 rates for August 21, 2026, published in the release dated August 24, 2026. A single FX snapshot is used throughout so that every reading in this report is internally consistent and independently reproducible from one dated source.
Price bases. Published costs are carried at the price basis their source states, which differs across projects: nominal USD for Vogtle, 2023 euros for Flamanville 3, 2015 sterling for Hinkley Point C, 2024 sterling for Sizewell C, 2024 Canadian dollars for Darlington. No cross-currency deflator is applied. Comparisons of escalation multiples use only pairs stated on the same basis at both ends.
WLCB formula. Capacity-weighted mean of four Western large-reactor projects: total capital cost in USD divided by total net capacity in kW. Inputs, capacities and conversions are shown in full in the section above and as rows in the dataset.
SCPI formula. Project capital cost per kW in USD, divided by WLCB. Readings are dated. Methodology changes will be disclosed and re-baselined rather than applied silently to past readings.
What is excluded and why. Olkiluoto 3 is excluded from the cost benchmark because no single sponsor or auditor figure covers both TVO’s investment and the Areva-Siemens consortium’s losses. Wylfa is excluded from cost comparisons because no project capital cost has been published and no final investment decision has been taken. Sizewell C is excluded from the escalation calculation because its 2020 and 2025 figures are stated on different bases. Akkuyu, Rooppur, Paks II and Barakah are absent because Axis Intelligence Research could not obtain sponsor-published or audit-published total project costs for them during this session, and estimated figures do not enter this dataset.
What these readings do not capture. SCPI measures capital intensity at commitment. It does not measure levelised cost of electricity, and it is silent on capacity factor, fuel cost, operating cost, decommissioning provisions, financing structure and cost of capital. Two projects with identical SCPI readings can deliver electricity at very different prices. Projects still under construction carry cost estimates that can move, and every large reactor in this dataset moved.
About This Dataset
Contents. 136 rows covering SMR and large-reactor capital costs, capacity, construction and commissioning dates, regulatory milestones, defence microreactor awards, IEA deployment scenarios, IAEA design counts, and the Axis-calculated WLCB and SCPI readings. Every row carries source organisation, source document, source URL, retrieval date, primary-source flag, Axis-calculated flag and, where applicable, a method note.
Coverage. Canada, United States, United Kingdom, France, Finland, China and global aggregates. Temporal span 2005 to 2050 scenario horizons, with cost and status readings as of August 2026.
License. CC BY 4.0.
Cite as: Axis Intelligence Research, “Small Modular Reactor Statistics 2026,” axis-intelligence.com/small-modular-reactor-statistics/
Citation Block
APA Axis Intelligence Research. (2026). Small modular reactor statistics 2026: Cost per kilowatt, build times, and how SMRs compare to large reactors. https://axis-intelligence.com/small-modular-reactor-statistics/
MLA Axis Intelligence Research. “Small Modular Reactor Statistics 2026: Cost per Kilowatt, Build Times, and How SMRs Compare to Large Reactors.” Axis Intelligence, 30 Aug. 2026, axis-intelligence.com/small-modular-reactor-statistics/.
Chicago Axis Intelligence Research. “Small Modular Reactor Statistics 2026: Cost per Kilowatt, Build Times, and How SMRs Compare to Large Reactors.” Axis Intelligence. August 30, 2026. https://axis-intelligence.com/small-modular-reactor-statistics/.
Frequently Asked Questions
Is the first Darlington BWRX-300 more expensive per megawatt than a large reactor?
Yes. According to Axis Intelligence Research, the first unit plus shared site infrastructure prices at USD 18,646 per kilowatt, against a Western large-reactor benchmark of USD 14,543 per kilowatt, an SCPI reading of 1.282. The four-unit fleet reads 0.870. Whether an SMR beats a large reactor per kilowatt depends entirely on whether you price the first unit or the programme.
What does the CAD 20.9 billion Darlington figure actually include?
Ontario Power Generation states CAD 6.1 billion for the first BWRX-300, CAD 1.6 billion for systems and services common to all four units such as roads, cooling water tunnels and ancillary buildings, and CAD 20.9 billion in total for four units delivering 1,200 MW. The figure is expressed in 2024 dollars and includes interest and contingencies.
How much cheaper does OPG expect units two, three and four to be?
Axis Intelligence Research derives CAD 13.2 billion for units two through four from OPG’s published figures, an average of CAD 4.4 billion per unit against CAD 6.1 billion for the first. That is a 27.9 percent per-unit reduction. It is a budget assumption, not an outcome, and the first evidence either way will be OPG’s actual cost for unit two.
Do modular construction and factory fabrication actually cut build times?
No Western SMR has finished, so the honest answer is that it is untested outside China and Russia. CNNC planned a 58-month build for the 125 MW ACP100 from first concrete. According to Axis Intelligence Research, the three completed Western large reactors in this dataset averaged 16.4 years, a ratio of 3.3 to 1, though a 125 MW unit and a 1,630 MW EPR are not the same construction problem.
Why can no one produce a reliable total cost for Olkiluoto 3?
Because the plant was built under a fixed-price turnkey contract, the overrun landed partly on TVO’s balance sheet and partly on the Areva-Siemens consortium as losses. Widely quoted totals add a utility’s investment to a vendor’s write-downs, which are different accounting objects. Axis Intelligence Research uses Olkiluoto 3 for schedule analysis and excludes it from cost comparison.
Did Sizewell C get cheaper per kilowatt by copying Hinkley Point C?
Not on published figures. Hinkley Point C prices at USD 14,648 per kilowatt on EDF’s GBP 35 billion in 2015 sterling; Sizewell C prices at USD 15,904 per kilowatt on GBP 38 billion in 2024 prices. Part of that gap is price-basis drift between 2015 and 2024 sterling, and Axis Intelligence Research does not claim it survives full deflation. But the replica is not visibly cheaper per kilowatt, which is worth weighing against fleet-learning assumptions.
What cost do SMRs have to reach to be deployed at scale?
The IEA’s cost parity case puts it at USD 4,500 per kilowatt in the United States and Europe and USD 2,500 per kilowatt in China by 2040, which would take global SMR deployment to 190 GW by 2050. According to Axis Intelligence Research, that USD 4,500 target reads 0.309 on SCPI, and Darlington’s fleet price is 2.81 times it.
Are Army microreactors licensed the same way as commercial SMRs?
No. The Janus Program reactors are prototypes regulated by the Army rather than through the NRC’s commercial licensing route, and they are contractor-owned and contractor-operated. The Army is targeting September 2028 for a first operating reactor on a military installation, under the target set in Executive Order 14299.
Which companies have SMR units actually committed, not just announced?
As of August 2026: GE Vernova Hitachi has four BWRX-300 units budgeted at Darlington and one under NRC construction permit review at TVA’s Clinch River site; Rolls-Royce SMR has a signed contract for three units at Wylfa ahead of a final investment decision; CNNC has the ACP100 in construction at Changjiang; and Antares Nuclear, BWXT, General Atomics, Radiant and Westinghouse hold US Army Janus awards for microreactors. Design counts run far higher: the IAEA lists more than 80 designs and concepts.
Does a strong SCPI reading mean cheap electricity?
No. SCPI measures capital cost per kilowatt at the point of commitment. Delivered electricity cost also depends on capacity factor, fuel, operating cost, financing structure and decommissioning provisions. A regulated asset base model, as used at Sizewell C, moves construction risk onto consumers in a way no capital-intensity figure can show.