Nuclear power works. It produces low-carbon electricity, avoids the health burden of fossil-fuel combustion, provides firm generation and can operate for many decades. My skepticism about building large new nuclear fleets has never depended on pretending otherwise. The more interesting question is why countries that have built nuclear successfully have generally done so through national industrial programs, while attempts to turn reactors into ordinary merchant-market investments keep running into the same financing problem. When I first wrote about this in 2023, I used the deliberately provocative headline that nuclear energy and free-market capitalism were incompatible. That was shorthand, not a claim that private companies cannot manufacture equipment, construct plants, operate reactors or invest in nuclear businesses. They obviously can. The narrower point has held up well: new nuclear is a poor fit for hands-off merchant project finance. The programs that scale best are usually ones in which governments constrain technology choices, organize financing and regulation, develop supply chains and workforces, and commit to enough reactors over enough time for learning to carry from one project to the next. The financing characteristics explain why. A gigawatt-scale reactor commits billions of dollars years before it produces electricity. Every delay stretches the period during which financing costs accumulate without operating revenue. Engineering problems, regulatory changes and supply-chain failures are concentrated inside one very large asset rather than spread across thousands of smaller projects. That makes nuclear economically different from modular technologies such as solar, batteries and wind, where individual project failures do not threaten the economics of an entire national buildout. Britain’s recent experience makes the pattern unusually clear. Hinkley Point C has continued to accumulate cost and schedule problems despite already having long-term revenue support. The response at Sizewell C was not to discover a pool of investors willing to carry Hinkley-style construction risk more cheaply. It was to restructure the economics so government equity, regulated cost recovery during construction and explicit public support absorb more of the risk before the plant begins generating. Sizewell may still be a rational strategic choice for Britain, but it is hard to describe that structure as a merchant generator simply attracting private capital because wholesale electricity revenues offer an attractive risk-adjusted return. Czechia is approaching the same problem with a different mechanism. The proposed Dukovany expansion combines a state loan intended to cover construction costs, long-term revenue stabilization and a project company that is majority state-owned. Britain’s small modular reactor program also points in the same direction: government selected the technology, funded development and is attempting to create the initial order book and industrial repetition on which the eventual economics depend. Different mechanisms, but the direction of travel is consistent. New nuclear becomes investable when governments reshape the risk profile. Finance is only half the story. Nuclear also needs a kind of standardization that competitive technology markets do not naturally produce. The strongest construction programs settle on a proven design or tightly constrained family, resist continuous customization, retain experienced construction teams and suppliers, and build enough units for repetition to matter. Nuclear’s integrated plant is itself a major unit of learning. Change the reactor rating, safety architecture, major systems or supply chain too much and part of the accumulated learning disappears. China is revealing because it removes many excuses commonly offered for poor Western nuclear performance. It has patient state capital, state-controlled utilities, deep engineering capability, national infrastructure planning and enormous electricity demand. Yet even China has struggled to impose the degree of standardization that nuclear economics reward. Its current program has rationalized around Hualong One and CAP-derived families, but variants remain, safety architectures differ, supply chains remain partly separate and new derivatives continue to appear. China is gaining repetition, but it has not frozen one national product line and simply manufactured it repeatedly. That matters because China also demonstrates what genuine industrial throughput looks like elsewhere in its electricity system. Adjusting annual additions for representative capacity factors, TFIE’s analysis finds that China’s 2025 wind and solar additions represent roughly 696 TWh of additional annual generation capability, compared with about 14 TWh from that year’s nuclear additions. The comparison is not pretending that a gigawatt of solar is equivalent to a gigawatt of nuclear. It is measuring how much annual generation capability each industrial ecosystem is adding. The policy conclusion is not that countries should never build nuclear. It is that they should stop pretending new nuclear is just another power technology waiting for markets to discover it. A serious nuclear program requires patient capital, standardized designs, controlled change, repeat construction, durable supply chains, trained workforces and an explicit decision about who pays when schedules and costs move. Those are the characteristics of national industrial policy, not spontaneous merchant-market deployment. A country may reasonably decide that firm low-carbon electricity is worth those commitments. But the commitments should be visible, their opportunity costs should be counted, and nuclear should be compared with alternatives on that full basis. For the deeper comparison of Sizewell C, Dukovany, Britain’s SMR program, China’s reactor standardization and the economics of repetition, read New Nuclear Needs The State To Make The Economics Work at TFIE Strategy Briefing.
+38 others
