Lecture 8: What Investors Need to Know About Small-Modular and Microreactors

Main Ideas

1. A benchmark for evaluating new nuclear designs

The lecture begins by revisiting a wind-and-solar system designed to address weather variability through storage and continent-scale transmission. Using 43 years of weather data, the cited analysis estimated 99.9% grid reliability at approximately $131 per megawatt-hour.

The instructor presents this as a demanding, non-optimized benchmark for carbon-free electricity. New nuclear proposals should be assessed against it rather than assumed to be economical. A second reference point is the projected capital cost of an AP-1000: about $7,700 per kilowatt of capacity. Because nuclear costs are largely upfront capital costs, the lecture treats this as a useful comparison.

2. Smaller reactors do not automatically cost less

Historical light-water-reactor data showed that cost per unit of capacity tends to fall as reactor size increases—roughly a 20% reduction for each doubling of size in the dataset discussed.

On that basis, a single NuScale module was estimated to cost about 2.35 times more per kilowatt than an AP-1000. Going smaller can therefore make the economics worse, not better. The exact historical relationship may not apply to very small or non-light-water designs, but the underlying issue remains: many costs do not shrink in proportion to a reactor’s output.

Modularity alone is not a breakthrough. The AP-1000 also uses factory-built modules, and problems fitting them together led to rework rather than savings. Any potential economic advantage must come from something more substantial, such as a genuinely different technology or a manufacturing approach that supports high-volume factory production.

3. Examples of small-reactor designs and their trade-offs

• NuScale: A relatively mature small pressurized-water design developed over roughly 20 years. Its proposed plant groups six 77-megawatt modules in one facility, sharing infrastructure such as the containment, cooling pool, turbine, site, and licensing. The instructor argues that this shared plant is economically important—and that the overall system is closer to a conventional-sized plant split into modules than to a truly tiny reactor.
• Holtec: A smaller, conventional pressurized-water-reactor concept that, according to the lecture, has made limited progress.
• Last Energy: A 20-megawatt concept that replaces concrete containment with a thick steel vault. It is air-cooled and buried underground. The proposed safety features may appeal to customers willing to pay more, but the instructor doubts the design is moving toward lower costs.
• Deep Fission: A proposal to place a 15-megawatt reactor deep underground. The lecture questions whether a borehole can be made reliably straight and round, whether the reactor would actually be isolated as proposed, and how it could be serviced or retrieved.
• High-temperature gas reactors (HTGRs): These use coated fuel particles in graphite and helium coolant. The designs are presented as relatively simple in some respects and potentially capable of passive cooling. However, their low power density means substantial reactor material is required for comparatively little output.
  • Ultra Safe Nuclear and its successor, Valor, are associated with this general approach.
  • X-energy’s pebble-bed design circulates fuel pebbles through the reactor; a prismatic version uses blocks instead. The lecture notes historical concerns about pebble abrasion and dust, as well as the difficulty of sorting and managing large numbers of pebbles.
  • The instructor emphasizes that HTGRs are not entirely new: earlier versions were built and operated. The examples cited had limited lifetimes, low capacity factors, high costs, or contamination and maintenance problems. The lecture therefore urges skepticism toward claims that current versions have already solved these difficulties.
• Molten-salt-cooled designs: Replacing helium with molten salt could increase power density. However, the lecture describes unresolved issues involving salt purity, corrosion, tritium production in the proposed fluoride-lithium-beryllium salt, and short vessel life at operating temperatures. The possible reduction in reactor size may be offset by replacement and corrosion costs.
• Molten-salt-fueled reactors: Dissolving fuel in the salt creates additional maintenance and safety challenges. Fission products can plate onto surfaces, while radioactive gases require separate systems for cooling and safe storage. Historical examples discussed in the lecture had poor operating performance and very high estimated costs.
• Heat-pipe or “nuclear battery” concepts: These aim to move heat from small reactors through heat pipes instead of a conventional coolant loop. The lecture treats their economics as uncertain because the concepts were not yet fully designed.

Overall, the lecture argues that many advanced concepts move away from the established economics and operating experience of water reactors while relying on unresolved engineering questions.

4. Cost estimates need to be treated cautiously

Published estimates for microreactors vary widely. Some are based on early, abstract concepts; “first-of-a-kind” and “nth-of-a-kind” estimates may not be grounded in completed designs or observed manufacturing costs.

The lecture argues that comparisons should account for realistic operating patterns. Many estimates assume a reactor runs nearly all the time. If a reactor must load-follow, its cost per unit of delivered electricity would rise. The instructor estimates an illustrative adjustment of about 1.4× for a capacity factor of roughly 70%.

Examples discussed include:

• CARM: A small Brazilian pressurized-water reactor under construction that is not a factory-made battery design. The lecture’s calculation from its historical costs puts its electricity price at roughly $370/MWh.
• A proposed battery-style reactor: A particularly optimistic published estimate gives an “nth-of-a-kind” range that, after the capacity-factor adjustment, becomes roughly $120–$500/MWh.
• Nuclear Energy Institute (NEI): Estimates that the instructor says provide little transparent basis.
• Idaho National Laboratory (INL): A design estimate of roughly $2,174/MWh at a 95% capacity factor. The lecture says its eventual “nth-of-a-kind” estimate relies on expert opinion.
• Project Pele: A U.S. Department of Defense effort for which the lecture says public design and cost details are limited.

The key point is not that these estimates are definitive, but that the lecture finds no credible claim that tiny battery-style reactors are already cheaper than conventional nuclear power. Their case would depend on markets other than ordinary grid electricity, such as remote sites or industrial heat.

5. Design and construction costs often rise as projects mature

Early reactor estimates commonly increase as engineering reveals additional equipment, manufacturing constraints, and other requirements. The lecture gives examples of estimates rising substantially for NuScale, X-energy, and GE Hitachi designs.

The AP-1000 is used as a warning: its early estimate was about $3,300/kW, while the completed projects cited cost about $20,000/kW. A Congressional Budget Office review of 75 U.S. nuclear plants is also cited as finding that completed costs averaged 207% above original estimates.

The instructor’s rough rule of thumb is to expect:

• Around a threefold increase from an early concept to a finished design.
• A further two- to threefold increase from design to completed construction.

The lecture attributes cost-estimate failures to a mix of genuine forecasting difficulty and incentives to present optimistic numbers, particularly in startup fundraising. It also notes that major infrastructure projects in other sectors experience cost overruns.

6. Factory production could help, but production rate matters

Historical reactor data cited in the lecture show only about a 1% cost reduction per doubling in the number of reactors built under conventional construction.

A claim from a Soviet naval-reactor factory manager suggested about a 5% cost reduction per doubling of factory-produced reactors. The instructor notes that U.S. naval-reactor cost data are difficult to obtain, limiting comparison.

The lecture distinguishes between:

• Cumulative production: how many units have ever been built.
• Production rate: how many units are made per year.

Producing a large number quickly may create different economies of scale from producing the same total over many decades.

The instructor questions whether automation by itself guarantees lower costs. A comparison with jet-engine manufacturing suggests that advanced machining and computer control may substitute capital for labor without necessarily reducing unit costs.

A study of stationary fuel cells is offered as a possible analogy: it found substantial cost reductions as annual production rose. The instructor treats this as a hypothesis worth investigating, not as proven evidence for nuclear reactors.

The lecture also raises the possibility of standardizing components, not just whole reactor designs. Repeated production of particular parts could help, but nuclear-grade materials, welds, certification, and quality controls make such manufacturing specialized and expensive.

The broader lesson is that credible cost projections require understanding the minimum efficient production scale—both how many units are made and how quickly.

7. Likely timelines and future markets

The lecture uses NuScale and the AP-1000 to illustrate that reactor design, engineering, licensing, and construction can take decades. It estimates roughly 20 years from concept to licensing for the examples discussed, followed by another five to ten years for construction.

On that timeline, the instructor argues that reactors proposed today are unlikely to arrive in time to supply data centers that need power in the near term.

If they arrive decades from now, they will compete in a much more renewable-heavy electricity system. They may therefore need to load-follow or serve non-grid uses, such as industrial process heat for activities like paper or iron production.

Proposed niche markets include disaster relief, remote Arctic communities, mines, islands, and military installations. The lecture questions whether these markets are large enough to support high-volume production. It also raises a security concern about locating reactors in war zones.

8. The developing-world market: an illustrative siting analysis

The instructor describes a geographic analysis intended to estimate whether small reactors could support electrification in underserved areas:

1. Use satellite observations of nighttime lights as a proxy for existing electricity access.
2. Combine the data with population, roads, water or rail access, terrain, and other geographic information.
3. Identify areas with low electrification but enough population and power demand to potentially support reactors.
4. Exclude locations that are physically unsuitable, difficult to reach, or in war zones.
5. Apply a governance and regulatory threshold, using Ukraine’s inherited nuclear system as the benchmark described in the lecture.

The analysis identified substantial theoretical potential, but the instructor cautions that this result does not establish economic viability. Many countries already subsidize electricity, and expanding service to additional populations would require funding. The lecture warns against assuming poorer communities can finance the scale of reactor manufacturing needed to make reactors cheaper.

Staffing is another constraint: a very large fleet would require either extensive automation or a major expansion of nuclear training and operational personnel.

Overall Conclusion

Small and microreactors are not automatically cheaper because they are small, modular, or factory-built. Their economic prospects depend on specific engineering advantages, realistic cost estimates, repeatable manufacturing, and a sufficiently large market.

The lecture’s most cautious long-term scenario is a standardized design manufactured at high volume for future grid flexibility or specialized non-grid uses—not a near-term, low-cost solution for data centers or general decarbonization.

Speakers and Sources Featured

Speakers

• R. Scott Kemp, the lecturer.
• Unidentified students, who ask questions and offer comments during the lecture.

Sources, organizations, and examples cited

• A weather-data analysis attributed in the subtitles to “Shainer”; the spelling is unclear in the auto-generated text.
• The U.S. Energy Information Administration (EIA), for an AP-1000 capital-cost projection.
• NuScale, Holtec, Last Energy, Deep Fission, Ultra Safe Nuclear, Valor, and X-energy.
• Historical high-temperature gas reactors, including Fort St. Vrain, AVR, and China’s HTR-10 and HTR-PM.
• The Congressional Budget Office, for its review of U.S. nuclear-plant cost overruns.
• The Nuclear Energy Institute (NEI) and Idaho National Laboratory (INL), for cost estimates discussed in the lecture.
• Project Pele, a U.S. Department of Defense reactor effort.
• A Soviet naval-reactor factory manager’s claim about factory learning rates.
• A stationary-fuel-cell manufacturing study attributed in the subtitles to “ENL”; the acronym is unclear in the auto-generated text.
• A departmental reactor-cost analysis attributed in the subtitles to “Yakobo Bondrono”; the name may be mistranscribed.
• CARM in Brazil and Westinghouse’s eVinci concept.
• Historical jet-engine manufacturing and automobile-industry production-scale comparisons.