OPG Delays Darlington BWRX-300 and Raises Cost Estimate | Nuclear Now #77
The Darlington SMR project is real progress for nuclear power, and a serious warning about the economics of building it. Ontario Power Generation has begun the work that can make a GE Hitachi BWRX-300 a functioning reactor, but its first unit will miss the October 2030 commercial-operation target and exceed the previously disclosed C$7.7 billion project estimate. That combination matters more than another announcement of a reactor design, a government pledge, or a venture round.
This is not a reactor that exists only in a presentation. OPG has a nuclear site, an operating utility, an established regulator, a supply chain, and a board willing to absorb the consequences of first-of-a-kind construction. Darlington is one of the few Western SMR projects far enough along to expose the difference between engineering feasibility and bankable delivery. The uncomfortable result is that the BWRX-300 may be an entirely credible reactor and still produce an unattractive first-unit price.
That distinction is where much of the current nuclear debate goes wrong. Critics are using Darlington’s delay and cost pressure to dismiss SMRs altogether. Advocates are treating the project as proof that the technology is ready to scale. Neither reading survives contact with the project schedule. Darlington is valuable precisely because it is making the hidden costs visible: design finalization, procurement, nuclear-grade manufacturing, construction sequencing, licensing interfaces, contingency, and the price of being first. The industry needs this data, but investors should not confuse learning what a reactor costs with proving that it is cheap.
The reported C$7.7 billion figure is for the first of four proposed BWRX-300 units, not a mature fleet price. That is the critical qualifier, but it is not an excuse. A first unit that costs several billion Canadian dollars for roughly 300 megawatts of electrical capacity implies an overnight-cost burden far above the level required for an economically compelling SMR. Even before financing, the arithmetic is harsh: the project is carrying the cost of a large nuclear construction program on a relatively small generating unit.
At a 90 percent capacity factor, a 300 MW reactor would generate about 2.37 terawatt-hours annually. That output is valuable because it is firm, dispatchable electricity, but a C$7.7 billion capital base spread across that production creates a formidable fixed-cost charge. The exact levelized cost depends on financing, operating life, fuel, refurbishment, and construction-period interest. The direction does not. If the first-unit price becomes the enduring price, the BWRX-300 will struggle against new wind, solar, storage, gas, and demand-side flexibility on energy-only economics.
But energy-only comparisons are incomplete. A nuclear reactor produces power through the night, during winter peaks, and when weather-dependent generation underperforms. It also occupies a relatively small physical footprint and can support an industrial load without relying on a massive overbuild of generation and transmission. Those attributes have economic value, particularly in Ontario, where electrification and data-centre demand increase the value of firm capacity. They do not erase capital cost. They determine how much premium a system is willing to pay for reliability.
The project’s next number matters more than the original estimate. OPG has not yet published a replacement cost or schedule, and it expects a revised assessment to be substantially complete by mid-2027. That creates a useful discipline for the market. If the revised estimate rises but the utility can demonstrate a credible path to lower costs on units two through four, Darlington may still validate a fleet strategy. If the first unit becomes a one-off demonstration whose economics depend on public support, the SMR thesis has failed at its most important claim, standardization.
The nuclear industry has spent years saying that factory production will reduce cost. That statement is technically plausible, but a factory does not lower the price of a design that has not yet passed through a first deployment. Standardization begins after the first project has converted drawings into repeatable procurement packages, qualified suppliers, construction procedures, inspection records, trained crews, and a regulator familiar with the configuration. Darlington is paying for that learning curve in public.
The comparison with NuScale is instructive. The company’s planned Idaho project was abandoned in 2023 after projected costs rose sharply, with an estimated levelized cost reported at roughly A$179 per megawatt-hour excluding subsidies. Darlington is further along and uses a different reactor, but the commercial lesson is shared: a smaller reactor does not automatically mean a smaller financial risk. It can mean more units are needed to reproduce the output of one large plant, while each unit still carries nuclear-grade compliance and project-management costs.
Darlington strengthens GE Hitachi’s position as a serious Western SMR vendor because the project is moving through the difficult part of the process, not merely collecting memoranda of understanding. The delay also gives competitors an opening. TerraPower, X-energy, Rolls-Royce SMR, NuScale, and other developers can now sell customers on schedule certainty, but only if they can show more than a design, a license milestone, or a customer announcement. The burden of proof has shifted to costed construction plans and qualified supply chains.
The project should also accelerate a more honest procurement model. Utilities considering SMRs will demand stronger conditions around contingency, owner’s costs, escalation, and the price of the second and third units. They will ask whether a vendor is selling a reactor or an integrated delivery system. They will insist on separating a regulatory achievement from a commercial achievement. A permit can show that a design meets safety requirements. It does not show that the plant can be built on budget.
Canada remains better positioned than most Western countries to learn this lesson because Ontario has both an operating nuclear fleet and a utility with institutional experience. The province’s life-extension program gives it a reason to preserve nuclear capability while new construction proceeds. Darlington’s four-unit plan also offers something the isolated demonstration projects lack, a chance to test whether repetition actually lowers costs. But the second unit must be treated as an economic experiment with published results, not automatically as a political continuation of the first.
The wider nuclear revival will not be decided by whether SMRs can produce electricity. They can. It will be decided by whether utilities can finance a standardized series of reactors at a price that beats the cost of building enough alternative capacity to deliver the same reliability. Darlington has advanced that debate because it has crossed the line from aspiration into exposure.
The industry should welcome the bad news. A late and expensive project is painful for Ontario ratepayers and embarrassing for the SMR sales pitch, but it is more useful than another year of frictionless optimism. The winners will be companies that turn this first-of-a-kind penalty into a measured, documented decline across subsequent units. The losers will be developers that keep asking investors to value manufacturing scale before they have built a reactor. Nuclear is coming back, but the comeback will be paid for one construction schedule at a time.