First-of-a-kind small reactor costs are published, the learning-rate case fails its first test, and the order pipeline thins
The final capital cost of the first unit and the contracted cost of the second and third become public through mandatory utility-commission reporting,…
Claude · 2031 · plausible
Prior state
Construction of the first Western grid-scale small modular reactor began in Ontario in the mid-2020s under a multi-unit programme, with first power targeted around the turn of the decade and subsequent units following. Sodium-cooled and high-temperature gas demonstration plants were under construction in Wyoming and Texas against their own commissioning schedules. Between 2024 and 2027 a large number of memoranda and development agreements were signed, many with data centre operators, most of them non-binding and carrying decision gates later in the decade. The entire commercial case rested on the proposition that factory fabrication and multi-unit repetition would produce a steep cost decline that conventional large nuclear never achieved.
Material change
The final capital cost of the first unit and the contracted cost of the second and third become public through mandatory utility-commission reporting, providing the first empirical observation of the learning rate. The observed decline is real but far shallower than the commercial case required. The consequence within the year is a sorting: one or two designs with demonstrated multi-unit repetition and a utility willing to order fleets receive firm orders, while the majority of announced projects are cancelled or deferred past their decision gates, and the expectation of hundreds of units deployed by the mid-2030s is publicly retired. Procurement shifts from bespoke first units toward repeat fleet contracting by a small number of utilities.
Why now
The disclosure is the event, and it is calendared. The first unit's in-service and cost-reconciliation reporting to the provincial regulator falls at the turn of the decade, and the contracted cost of the following units is disclosed in the same proceedings — this is the first year in which a second-unit cost exists to compare against a first. The demonstration reactors in Wyoming and Texas reach commissioning or cost-disclosure milestones in the same window. And a substantial share of the mid-2020s development agreements carry decision gates dated to 2030 and 2031, so the counterparties must convert or walk away within the year. Before 2031 there is no second-unit number; after 2031 the decisions have already been taken.
Mechanism and resistance
The mechanism is that capital markets and utility commissions price the observed number rather than the projected one, and an offtaker with a decision gate and a real number makes a real decision. Resistance is genuine and could preserve the pipeline. Electricity demand from data centres is price-insensitive by the standards of the sector and may tolerate costs that no ratepayer-funded utility would. Policy support — loan guarantees, production credits, and public equity — can carry projects past commercial logic for years. And for the non-light-water designs the binding constraint is fuel rather than cost: high-assay low-enriched uranium supply remains the gating item, so their schedules may slip for reasons that leave the cost question unanswered rather than answered badly.
Consequences
The winners are the incumbent heavy nuclear supply chains and the industrial clusters that host repeat orders, along with the Korean and Chinese vendors offering conventional large units at known prices to buyers who had been waiting for something better. The losers are the venture-financed design startups, the jurisdictions that wrote small reactors into decarbonisation plans as a load-following resource, and the industrial offtakers who deferred alternative procurement while waiting. The most consequential distributional effect is for low- and middle-income electricity systems, which had been told that small reactors would be their route to firm low-carbon power without grid-scale investment: the disappointment returns them to solar with storage plus gas, which is cheaper and faster but leaves the industrial heat and firm-capacity problem unsolved. The broader significance is a rare, well-documented public test of the claim that modularity defeats the cost pathology of nuclear construction, and the answer within this interval is a qualified no.
End state
A small-reactor sector reduced to a small number of real programmes with firm fleet orders and known costs, a cancelled majority, and a public expectation recalibrated from mass deployment to niche industrial and grid-support application.
Observable test
The reported overnight capital cost per kilowatt of the first Ontario unit and the contracted cost of the second or third in regulator filings; the count of firm orders worldwide in 2031 against the letter-of-intent pipeline of 2027; and the number of announced projects formally cancelled or deferred within the year.
Disconfirming sign
Second-unit costs come in far below the first, and firm orders in 2031 exceed the 2027 pipeline in number and committed capacity.
Themes
Energy & resources, Business & industry, Infrastructure & transport