BasicsEnergyOntario Power Generation is building four GE Vernova Hitachi BWRX-300 units at Darlington for C$20.9 billion, about C$17,400 per kilowatt across 1,200 MW. That figure includes interest, escalation and contingency, so it is not an overnight cost. It is still the clearest price signal the Western SMR sector has produced. Whether it ends up on your bill depends on what units two through four cost, and on who is paying for the financing.
This article tracks that question across the projects that matter before 2035. Each is at a different stage, and the stage determines how much weight its cost claims can bear.
What Darlington has actually priced
The Ontario government approved construction in May 2025. The first unit carries C$6.1 billion, plus C$1.6 billion for systems shared by all four, according to QP Briefing. OPG says the first unit should reach the grid by the end of 2030, with units two to four following in the mid-2030s.
Staged execution is the cost thesis. OPG built its CANDU refurbishment the same way, one unit at a time, and the last of four refurbished units returned to service in March 2026, according to Nuclear News. OPG says it will refine the cost of the remaining three units after learning from the first. That is a plan, not a result.
Compare it with GE Hitachi’s early design target of US$700 million per reactor, which the Globe and Mail reported. The Darlington estimate is several times that. The gap is the first thing a skeptical analyst should price in.
Where the other projects stand
Here the milestones need separating. A permit is not a contract, and a contract is not a financial close.
TVA Clinch River, Tennessee: The NRC issued a construction permit for one BWRX-300 on September 29, 2026, after a 14-month review, per GE Vernova’s release. TVA has announced neither a construction date nor a cost, and it still needs a separate operating license.
Rolls-Royce SMR, Wylfa, Wales: Great British Energy-Nuclear signed a contract in April 2026 for three 470 MWe PWRs, per NucNet reporting. A final investment decision is not expected until 2029.
TVA and ENTRA1, up to 6 GW: ENTRA1 would finance and own plants built from NuScale 77 MWe modules, and TVA would buy the output under power purchase agreements that have not been signed. No price or timeline has been disclosed.
Dow and X-energy, Seadrift, Texas: The companies filed a construction permit application in March 2025 for four 80 MWe Xe-100 units, with review of up to 30 months. This research found no reported permit decision.
Only Darlington is under construction. Clinch River is permitted, Wylfa is contracted for development work, the ENTRA1 deal is a framework, and Seadrift is under review.
The arithmetic behind a lower bill
Capital cost dominates. Bank of America puts the sensitivity at roughly $15 to $20/MWh of LCOE for every $1,000/kW of overnight capital cost. A Fermi Energia analysis of the BWRX-300 estimates an overnight cost near EUR 8,223/kW, which it translates to about EUR 67/MWh, other factors equal. That is the author’s estimate, not an OPG disclosure.
Gas-fired generation is typically cited at $40 to $70/MWh, so first-of-a-kind SMR power does not win on price. The history is a warning. NuScale’s Idaho project saw its target price climb from $58/MWh to $89/MWh before cancellation in November 2023. Subscription risk killed it, not engineering.
Check your own utility’s next resource plan for the SMR price it assumes, and whether that price includes financing. If it does not, treat it as an aspiration.
One more point is easy to miss. SMRs lower the generation line of a bill, not transmission or distribution charges. Any claim that reactors cut bills outright is incomplete. Everything we cover, from capacity to counterparties to regulatory stage, is logged and searchable in SMRbrief Pro, which is where to check a project’s status before trusting a headline cost.
What would change the outlook by 2035
The honest answer is that most households will not see an SMR in their power mix by 2035. Darlington units are the realistic source of cost data by then, with Clinch River next if TVA starts building soon. Wylfa’s 2029 decision date leaves roughly six years to deliver, which looks tight for a first unit. This is analysis, not a reported schedule.
Three things move the outlook. Units two through four at Darlington have to come in well below the first, and OPG’s own refined estimate will show whether they do. TVA has to publish a cost for Clinch River. And buyers need long-term contracts that let a developer finance the second plant.
Lower bills in the early 2030s are plausible only where a utility with cheap capital and a repeat design pays for the learning. If the second Darlington unit costs materially less than the first, does that finally make the NOAK case, or does it just prove learning works for one owner at one site?