Nuclear Notes is produced by the Breakthrough Institute, and covers the policy, regulation, technology, and business of advanced nuclear energy in the United States and beyond.
Every Thursday, we track what’s really going on in the field: the new rules moving through the Nuclear Regulatory Commission, the advanced reactors moving from design to deployment, the companies going public and the utilities merging, and the executive orders and acts of Congress reshaping how the country builds. We explain not just what happened, but why it matters and how the pieces fit together.
A New Calculation of Nuclear Costs Makes Reactors Cheaper Than Gas Plants
Analysts who estimate the cost of future nuclear reactors based on Georgia Power’s Vogtle 3 & 4 are using the wrong yardstick, according to a new paper by Charles Komanoff, a longtime nuclear skeptic turned advocate, and James Boucher, a climate policy analyst.
Vogtle’s two new Westinghouse AP1000 reactors took 14 years to build and cost about double what they were supposed to, and are obvious reference points because they are the most recent builds. But they are outliers, the researchers argue, because, although they were twins, they were hit by the COVID pandemic, the Westinghouse bankruptcy, the fact that they were first- and second-of-a-kind, and other factors.
Far better insight comes from the 47 light-water reactors completed in this country in the 1980s and early 1990s, the researchers said. They tease apart the large cost variations among those reactors to distill insights into how to minimize the cost of new ones.
However, of those 47 reactors, many of them were also late and over budget, and hit by higher interest rates than prevail now, which made delays even more costly. Other factors are similar: supply chain constraints, including both hardware and skilled craft labor, lack of experience in the management of complex projects, and the details of new designs. But the authors’ contention is that there are lessons in the variation among those construction projects.
Some of the arguments are familiar: standardization and building twins or triplets cut costs, as does using an “orderbook approach” (i.e., doing the same job over and over). The Breakthrough Institute has previously endorsed the orderbook idea.
Among their findings:
Projects with mostly standardized designs had costs 29 percent lower than projects that didn’t
Plants with twin or triplet reactors at a single site had an average cost approximately 20 percent lower than sites with one reactor
Projects built with an “orderbook” approach, with multiple projects and a standardized design, drove down costs
Reactors were easier to build in periods where regulations were stable, a condition likely to be achieved now in the United States
New reactors can be built for an average cost of $8,200 a kilowatt or less, the researchers found. Vogtle cost $16,350 a kilowatt. Both figures are “overnight costs,” not counting the expense of interest during construction. Those interest costs appear poised to rise, because rates are rising, but the new research also gives clues about how to hold them to a minimum, by speeding up construction.
The paper’s cost estimates are about 30 percent lower than the estimates in a recent joint report by the New York State Energy Research and Development Authority and the Department of Public Service. And it notes that reactors built in the South were generally less expensive than those built elsewhere, but says it cannot explain why. It also casts doubt on the metric that is often used to compare the costs of competing technology: levelized cost of energy (LCOE). LCOE is calculated by adding up the capital, fuel, and operating costs, and dividing by the amount of energy produced. But that measure does not help evaluate the ability of technologies to contribute to another goal, “running an entire system that has to be exquisitely reliable at all times.”
The report proposes another measure for the value of a plant: comparison with the social cost of carbon, which is an estimate of the economic damage caused by burning fossil fuels, per ton of CO2. During the Biden administration, the Environmental Protection Agency put that at $480 per ton (The Trump administration has rejected the whole concept.) A megawatt-hour of electricity from an efficient gas-fired plant thus has a social cost of just under $200. But a new reactor can produce electricity for $110 to $190 a megawatt hour, they found.
How much these calculations will be incorporated into construction decisions is not clear, but they do provide a theoretical basis for a more optimistic estimate about new nuclear reactors. And they put numbers on factors that are generally recognized as helpful, like building multiple copies of the same reactor, at fewer locations.
One Giant Nuclear IPO Hits Pause but Another Marches On
With growing talk on Wall Street of an artificial intelligence bubble, Holtec has delayed an initial public offering that could have valued the company at $10 billion, citing “uncertainty of data center development.” But Westinghouse appears to be pressing forward and seeking a valuation of more than $50 billion, a tidy sum for a company that declared bankruptcy in March 2017. Brookfield Business Partners, a Canadian turnaround firm, and some partners bought Westinghouse for about $4.6 billion. Six years later, Brookfield sold 49 percent of the business to Cameco, a Canadian uranium mining firm, in a transaction that valued the company at about $8 billion.
Brookfield’s share was valued at $2.3 billion, Cameco’s at $2.2 billion, and the companies assumed another $3.4 billion in Westinghouse’s debt. Elements of the turnaround include the completion of Georgia Power’s Vogtle 3 & 4, the twin AP1000 project that bankrupted Westinghouse, and those reactors’ successful start-up. And the Department of Energy has conditionally decided to loan $17.5 billion for the fabrication of major components for ten AP1000 units.
But among the potential beneficiaries of a $50 billion valuation is the U.S. government, which, in exchange for past help to the company, could take 20 percent of cash distributions made by Westinghouse to its investors over $17.5 billion. Or the government could end up owning stock in the company.
Westinghouse has always been valuable because it has a profitable business serving reactors it has sold over the years. But the much higher valuation it is now seeking is based on the prospect of sales of numerous additional AP1000 models here and in Europe.
But on Wall Street, where the motivations are often described as alternating between greed and fear, there is anxiety at the moment.
As reported last week, Holtec pulled back amid a general retreat of companies related to data centers that were seeking to go public. The New York Times recently reported that SB Energy, a subsidiary of SoftBank, which had proposed building the biggest data center in the world in Ohio, has delayed what was expected to be a $50 billion IPO. So has Aggreko, a company that provides electricity to data centers, the Times reported. Anthropic, the AI company, has delayed a major filing that is needed to go public, at least briefly.
Meanwhile, X-energy went public on April 26, at $23 per share and surged to over $30, but this week was trading at under $17. X-energy plans to build a high-temperature reactor at a Dow chemical plant in Texas, to provide steam that is now made by burning natural gas. NuScale Power, which has a small modular reactor design licensed by the Nuclear Regulatory Commission, went public in May, 2022, at about $10 a share. A year ago it was selling for more than $57 a share, but it is now trading under $9.
TerraPower and Bechtel Will “Move Forward Separately”
TerraPower, which is building a sodium fast reactor with a molten-salt energy storage system, has parted ways with Bechtel, the veteran engineering and procurement contractor, at the Natrium project, the two companies said.
Neither said why.
Bechtel assisted TerraPower through several key steps, including getting a construction permit from the NRC. TerraPower said it did not expect the decision to delay construction.
Deep Underground, an Unusual Reactor Design
Deep Fission, the company that plans to lower a small nuclear reactor into a well a mile deep and use the earth itself as the containment, has just taken two unusual steps: it has publicly released its Nuclear Safety Design Agreement, the document it submitted to the Department of Energy to win approval for testing on DOE property, and it has proposed a reactor without control rods.
The company has previously described its design as including four standard pressurized water reactor fuel bundles. But in a traditional reactor, control rods are lowered among the fuel assemblies to choke off the flow of neutrons. Deep Fission’s safety document describes a “soluble neutron absorber in a concentration high enough to rapidly shut down the reactor.” Among the questions for safety regulators may be how to promptly deliver the liquid from the surface, a mile above.
Kairos Gets an Investor/Supplier
Samsung Construction & Trading Corporation, an engineering firm that is part of the Korean manufacturing conglomerate, will take an equity position in Kairos Power and provide engineering services, with a total value of up to $100 million, the two companies said. Samsung will help build Kairos’ 50-megawatt demonstration reactor, which is supposed to start supplying electricity to Google in 2030.
Kairos is using TRISO fuel, but instead of a gas coolant, its design calls for fluoride salt, which can absorb large amounts of heat at low pressure.
“The companies plan to collaborate on selected engineering and construction activities to strengthen technical and execution readiness, initially focused on power-generation systems and balance-of-plant scope,” according to the announcement.
Thanks for reading!
In costing out any form of electricity generation, there ought to be a calculation of the "social cost of blackouts" associated with the reliability of that supply. One sustained widespread blackout due to unreliable electricity generation (northern climes in January) will be enormously expensive and deadly. Nuclear reactors and coal, will have the lowest SCoB because these two forms of generation store their fuel on-site and do not depend on the weather, season, or day-night situation. Solar, wind, and inverters will have a high SCoB as their penetration into a grid advances.
Building a plant costs X. Operating it costs Y. I don't see consideration of operating costs for SMR nuclear plants in your assessment. NuScale was to build a dozen or so of their 70MW electric reactors in Idaho but the project collapsed because, as I understand it, NuScale could not produce electricity at a cost acceptable to the local utilities. Perhaps it was operating cost that did them in. One 1000MW electric plant needs ~4 operating crews, 3/day plus coverage for weekends, holidays and such. How many operating crews would a dozen 1/12 size reactors require or from a different direction, how many reactors can one crew safely operate, a figure the NRC will doubtlessly establish. Perhaps 2? If so, there are now 24 operating crews - $$$. Also, how many steam turbine plants is each reactor driving - one per reactor or one large steam turbine driven by as many reactors as are operating? I'm not a turbine guy but I suspect that the most efficient turbine is one matched to the steam supply and operating a turbine at part load is inefficient of energy. More $$$. And maintenance - with a bunch of SMRs you get a bunch of individual maintenance projects. Perhaps nuclear notes would like to lay out for us a thorough cost assessment of the SMR craze so we can better understand what it all will likely be.