(I wrote this piece a week or so ago, meant to do a bit more work but haven’t got around to it. Hence slightly dated allusions)
The culmination of Donald Trump’s state visit to the UK was a press conference at which both American and British leaders waved pieces of paper, containing an agreement that US firms would invest billions of dollars in Britain.
The symbolism was appropriate, since a central element of the proposed investment bonanza was the construction of large numbers of nuclear reactors, of a kind which can appropriately be described as “paper reactors”.
The term was coined by US Admiral Hyman Rickover, who directed the original development of nuclear powered submarines.
Rickover described their characteristics as follows:
- It is simple.
-
It is small.
-
It is cheap.
-
It is light.
-
It can be built very quickly.
-
It is very flexible in purpose (“omnibus reactor”)
-
Very little development is required. It will use mostly “off-the-shelf” components.
-
The reactor is in the study phase. It is not being built now.
But these characteristics were needed by Starmer and Trump, whose goal was precisely to have a piece of paper to wave at their meeting.
The actual experience of nuclear power in the US and UK has been an extreme illustration of the difficulties Rickover described with “practical” reactors. These are plants distinguished by the following characteristics:
- It is being built now.
It is behind schedule
It requires an immense amount of development on apparently trivial items. Corrosion, in particular, is a problem.
It is very expensive.
It takes a long time to build because of the engineering development problems.
It is large.
It is heavy.
It is complicated.
The most recent examples of nuclear plants in the US and UK are the Vogtle plant in the US (completed in 2024, seven years behind schedule and way over budget) and the Hinkley C in the UK (still under construction, years after consumers were promised that that they would be using its power to roast their Christmas turkeys in 2017). Before that, the VC Summer project in South Carolina was abandoned, writing off billions of dollars in wasted investment.
The disastrous cost overruns and delays of the Hinkley C project have meant that practical reactor designs have lost their appeal. Future plans for large-scale nuclear in the UK are confined to the proposed Sizewell B project, two 1600 MW reactors that will require massive subsidies if anyone can be found to invest in them at all. In the US, despite bipartisan support for nuclear, no serious proposals for large-scale nuclear plants are currently active. Even suggestions to resume work on the half-finished VC Summer plant have gone nowhere.
Hope has therefore turned to Small Modular Reactors. Despite a proliferation of announcements and proposals, this term is poorly understood.
The first point to observe is that SMRs don’t actually exist. Strictly speaking, the description applies to designs like that of NuScale, a company that proposes to build small reactors with an output less than 100 MW (the modules) in a factory, and ship them to a site where they can be installed in whatever number desired. The hope is that the savings from factory construction and flexibility will offset the loss of size economies inherent in a smaller boiler (all power reactors, like thermal power stations, are essentially heat sources to boil water). Nuscale’s plans to build six such reactors in the US state of Idaho were abandoned due to cost overruns, but the company is still pursuing deals in Europe.
Most of the designs being sold as SMRs are not like this at all. Rather, they are cut-down versions of existing reactor designs, typically reduced from 1000MW to 300 MW. They are modular only in the sense that all modern reactors (including traditional large reactors) seek to produce components off-site. It is these components, rather than the reactors, that are modular. For clarity, I’ll call these smallish semi-modular reactors (SSMRs). Because of the loss of size economies, SMRs are inevitably more expensive per MW of power than the large designs on which they are based.
Over the last couple of years, the UK Department of Energy has run a competition to select a design for funding. The short-list consisted of four SSMR designs, three from US firms, and one from Rolls-Royce offering a 470MW output. A couple of months before Trump’s visit, Rolls-Royce was announced as the winner. This leaves the US bidders out in the cold.
So, where will the big US investments in SMRs for the UK come from? There have been a “raft” of announcements promising that US firms will build SMRs on a variety of sites without any requirement for subsidy. The most ambitious is from Amazon-owned X-energy, which is suggesting up to a dozen “pebble bed” reactors. The “pebbles” are mixtures of graphite (which moderates the nuclear reaction) and TRISO particles (uranium-235 coated in silicon carbon), and the reactor is cooled by a gas such as nitrogen.
Pebble-bed reactor designs have a long and discouraging history dating back to the 1940s. The first demonstration reactor was built in Germany in the 1960s and ran for 21 years, but German engineering skills weren’t enough to produce a commercially viable design. South Africa started a project in 1994 and persevered until 2010, when the idea was abandoned..Some of the employees went on to join the fledgling X-energy, founded in 2009. As of 2025, the company is seeking regulatory approval for a couple of demonstrator projects in the US.
Meanwhile, China completed a 10MW prototype in 2003 and a 250MW demonstration reactor, called HTR-PM in 2021. Although HTR-PM100 is connected to the grid, it has been an operational failure with availability rates below 25%. A 600MW version has been announced, but construction has apparently not started.
When this development process started in the early 20th century, China’s solar power industry was non-existent. China now has more than 1000 Gigawatts of solar power installed. New installations are running at about 300 GW a year, with an equal volume being produced for export. Even allowing for differences in capacity factor, annual solar installations equal of exceed China’s entire existing nuclear capacity (about 60 GW). In this context, the HTR-PM is a mere curiosity.
This contrast deepens the irony of the pieces of paper waved by Trump and Starmer. Like the supposed special relationship between the US and UK, the paper reactors that have supposedly been agreed on are a relic of the past. In the unlikely event that they are built, they will remain a sideshow in an electricity system dominated by wind, solar and battery storage.
{ 29 comments }
Doug Muir 10.03.25 at 11:13 am
If nuclear reactors could indeed be made small, cheap, quickly and simply — or if they could just be made cheap, full stop — nuclear would be a valuable and useful addition to our toolkit. The sun doesn’t always shine, the wind doesn’t always blow, and we’re not close to solving storage at scale. So we’re going to need a lot of baseload. And say what you like about nuclear, it’s about as baseload as you get.
The problem is that nuclear is stupidly expensive, and nobody seems to have a fix for that. I’m on record as saying that I’d oppose a nuclear power plant being constructed in my county, not because I’m afraid of nuclear power — I’m really, really not — but because I don’t want my children and grandchildren to be paying extra-high electricity bills for decades to come. (I’ll note that this is exactly what is happening right now to the unfortunate people of Georgia, USA — they’re paying for the cost overruns on Vogtle, and will be for the foreseeable future.)
Doug M.
Trader Joe 10.03.25 at 11:18 am
Thank you for this.
I’m not a scientist, but I am an investor and many are excited about US companies OKLO and BWX Technologies.
BWX Technology is known for its nuclear reactors for submarines and believes it has a scaled UP version that can be used to power data centers and similar site specific applications (military bases, hospitals other essential infrastructure).
OKLO likewise has a SMR that is essentially a self contained shipping container with a nuclear fission reactor inside of it. They’ve been selected for 3 trial projects in the US and are in the process of getting NRC certification for these units. They are also regarded as ideal modular type power for data centers and other site specific uses.
(I’m likewise familiar with NuScale who you mentioned and discarded, though they have achieved NRC certification.)
I would be curious among the CT knowledge base if there were any views on the technology of these companies, as a lot of investors are very enthusiastic (i.e. valuations are sky high) and there seem to be some very smart companies actively engaged in using these on particularly data center projects they are pursuing.
I’m not myself an investor in any of these and am not recommending them, but Wall Street seems pretty enamored and while its often wrong it isn’t often completely stupid about technology.
MisterMr 10.03.25 at 12:19 pm
Personally, I’m a bit ambivalent about nuclear: here in Italy we had 2 referenda (at some decades distance) that banned nuclear, and in the second one I voted against nuclear(the first one I was still a minor).
However, currently AFAIK Italy is buying energy from France (nuclear, with plants close to the border to Italy), Russia (not a nice idea now), possibly the USA due to fracking (also not a good idea for various reasons, though perhaps less egregious than Russia).
Probably comparatively nuclear would have been the lesser evil, if done some decades ago.
Chances are that, in a few decades, the same argument will be still true, though perhaps for different resons: it all depends if we can churn out enoug green energy, of if for some reason energy demands become smaller (still not very likely IMHO, but who knows).
But if things go on as they are doing now, nuclear isn’t all that bad: maybe it’s true that it’s expensive, but we could produce it here instead of buying from abroad.
Much of Europe is in the same situation.
Comparing USA or Australia to european countries, or other in similar condition, IMHO is a bit misleading because the USA and Australia have an high level of energy self sufficiency.
Michael Cain 10.03.25 at 2:10 pm
Nits… The VC Summer plant is in South Carolina, not North Carolina. The NuScale reactor facility was to be built in Idaho, not Utah, at the Idaho National Lab owned by the Dept of Energy. That site was chosen to avoid the need for any sort of state approval. The commercial high-temperature helium-cooled Fort St. Vrain reactor in Colorado used TRISO particles, but bonded in fuel rods rather than pebbles. The fertile component of the TRISO particles there was thorium. Fort St. Vrain was a technical success, except for the use of water lubrication in a helium recirculation pump. That particular engineering problem has since been solved. Not too many miles from where I live is a storage facility for the remaining live (not spent) Fort St. Vrain fuel.
Charlie W 10.03.25 at 2:39 pm
Slightly ironically, given that you’re citing Hyman Rickover, is that the RR SMR is likely derived from a submarine reactor design. Those do exist, and seem reliable and safe. Whether a SMR is a better idea than, say, PVs plus batteries is another story.
Laban 10.03.25 at 2:59 pm
I confess I fail to understand how lots of small nukes are cheaper than a large one – unless making large (ish) numbers for nuclear submarines means economies of design/scale.
Still hoping the Chinese, who seem to be ahead, crack molten salt reactors – both as reactors which don’t need huge volumes of cooling water and as breeder reactors capable of transmuting long-lived nuclear waste into something more easy to store – with a long term aim of reactors with zero nuclear waste.
While we’re waiting for them (and fusion, always 10-15 years away since my childhood), the Russian VVER designs are either very reasonably priced, or they’re subsidising this for policy reasons.
https://en.wikipedia.org/wiki/Akkuyu_Nuclear_Power_Plant
“In May 2010, Russia and Turkey signed an agreement that a subsidiary of Rosatom would build, own, and operate a power plant in Akkuyu comprising four 1,200 MWe VVER1200 units…Turkish Electricity Trade and Contract Corporation (TETAS) has guaranteed the purchase of 70% power generated from the first two units and 30% from the third and fourth units over a 15-year power purchase agreement. Electricity will be purchased at a price of 12.35 US cents per kW·h and the remaining power will be sold in the open market by the producer. “
John Q 10.03.25 at 6:13 pm
Thanks for corrections. I meant to check my memory and forgot. I wasn’t aware of Fort St Vrain and will look it up
Jimmy 10.03.25 at 6:16 pm
Trader Joe @2 and Charlie W @5:
(Former) US nuclear submarine officer here. As such, I’m generally predisposed towards nuclear and hold fewer safety concerns than the average citizen. That said, I see at least three challenges with adapting submarine reactor designs for civilian use.
(1) Financial constraints: Most relevant to the OP, financial constraints for submarine reactors are very different from those for commercial reactors. Defense budgets and responsibility to taxpayers are not the same thing as capital expenditures and investor expectations.
(2) Enrichment; part of what allows submarine reactors’ small size is that they use much more highly enriched uranium than commercial reactors. (If I recall correctly, the exact numbers are classified so I’m legally prohibited from sharing them. Check Wikipedia.) Expanding the manufacture/access/use of such highly enriched uranium is highly problematic, from a security standpoint.
(3) Public opposition: Whether the public’s concerns about nuclear are justified or not is beside the point. NIMBYism is real, and contributed to the closure of, for example, the Indian Point Nuclear Power plant just outside NYC (where I live). Yet most New Yorkers either don’t know or are unconcerned with the ~15 submarine reactors, housing highly enriched uranium and undergoing much more complicated operations (i.e., regular shutdowns and startups), sitting in nearby Groton, CT.
A common contributing factor to 1-3 is that the Navy takes their nuclear safety record VERY seriously and naval nuclear reactors are subject to intense levels of oversight and regulation from the office of Naval Reactors (a joint office of the Department of Energy and the Department Of Defense, headed by a four star Admiral, the first of whom was Hyman Rickover.) This level of oversight is incredibly expensive, and contributes to the relative lack of public concern/high degree of confidence vis-a-vis commercial nuclear.
John Q 10.03.25 at 6:41 pm
@Laban, Akkuyu is indeed financially problematic, and way behind time. Too opaque to disentangle the impacts of sanctions from the fact that Russia can no longer subsidise projects like this.
The interest rate being charged is apparently 3 per cent, compared to a current Russian central bank rate of 18 per cent, an indication of the opacity of the finances
https://www.paturkey.com/news/2025/7b-funding-delay-hits-russia-led-akkuyu-nuclear-plant-project-in-turkey-21066/
pfb 10.03.25 at 8:05 pm
I don’t really have strong feelings about nuclear power, one way or the other, but it seems relevant to me that France currently gets 70% of its electrical power from nuclear.
I guess I’m wondering why the UK is looking to the US for nuclear advice when their neighbor on the other side of the chunnel seems to have figured it out.
Laban 10.04.25 at 1:06 pm
pfb 10
“their neighbor on the other side of the chunnel seems to have figured it out”
Alas, they haven’t quite figured it out, in that most of the generation in France is from oldish reactors, built 30 or 40 years ago. The Brits (Tony Blair having banned new nuclear build in 1998, thus losing a generation of nuclear engineers) signed up for a new generation of plants which either hadn’t gone through actual construction, or were having major issues. One of which being that the company making the steel forgings had been fiddling their safety data since the mid-1960s.
https://en.wikipedia.org/wiki/EPR_(nuclear_reactor)
The first EPR unit to start construction, at Olkiluoto in Finland, originally intended to be commissioned in 2009, started commercial operation in 2023, a delay of fourteen years. The second EPR unit to start construction, at Flamanville in France, also suffered a more than decade-long delay in its commissioning (from 2012 to 2024). Two units at Hinkley Point in the United Kingdom received final approval in September 2016; the first unit was expected to begin operating in 2027, but was subsequently delayed to around 2030. EDF has acknowledged severe difficulties in building the EPR design.
oldster 10.04.25 at 1:25 pm
You refer to Hyman Rickover by his given name because you and he were close friends?
John Q 10.04.25 at 7:13 pm
Oldster @12 Oops! As I said, I meant to do more work on this, but ran out of time to do it before the link to Trump-Starmer slipped out of memory. Fixed now
John Q 10.04.25 at 7:17 pm
@10, @11 I wrote in 2014 about the reasons for the French success of the 1970s and later failure. Headline (not chosen by me) more definitive on China than I would have liked, but the article stands up fairly well
https://nationalinterest.org/feature/china-can-make-nuclear-power-work-9815
Laban 10.04.25 at 7:34 pm
John Q 14 – the success of China in actually getting the EPR going certainly supports your thesis. OTOH I’d have thought that most of your four elements were still present in France – certainly EDF is still government controlled – have the other factors changed much there?
Paul L 10.04.25 at 9:59 pm
This applies to every other generation technology I can think of: scaling up reduces the cost. Cost is usually not a barrier for military technology. I’m always suspicious of “this time it’s different” arguments.
Aside from cost, one of the biggest downsides of moving to nuclear is the Russian dominance of the fuel supply chain. Swapping Russian fossil fuels for Russian nuclear fuels would help the climate but doesn’t remove the geopolitical dependence. This documentary on Arte (German/French with English subtitles) is well worth watching.
Thomas Jørgensen 10.05.25 at 12:35 am
The real reason everyone keeps designing reactors in the 40-100 mw output range is very simple: “Ocean Freight”. The large freighters that move everything physical around are currently stuck burning bunker oil.
This is Very Expensive. No. really. More expensive than that! 40-50 million euros /year just on fuel.
A civilian knockoff of the French k15 naval reactor would be a whole lot cheaper. Would it require very highly paid crew to manage? Yes. Not, however, 50 million euros /year worth of crew.
This is also basically certain to become even more of a problem – bunker oil is a by-product of gasoline refining. Everyone switches to EV transport ? No bunker oil. It would, of course, free up raw oil for other uses, but the cost of fuel for ships would still go way up, because the remaining refineries would now have to make actual money on supplying them, not just take what they can get to get rid of a waste product.
So there is a large market for ship propulsion units that don’t burn oil. And this market will happily pay a good chunk more than the grid will, at least initially.
There is also an argument to be made for factory produced reactors for the grid.. but since all the same factors that gave us the EPR (Larger reactors are cheaper to operate per mwh produced) are still in full effect anyone serious about that is going to make their units as big as the possibly can and still be moved from the factory to the customer. 300 MWs seem to be the practical limit here.
That’s Rolls Royce. Which means any US reactor vendor going for the UK market is going to get their lunch eaten for sure and certain sure. Either the large conventional reactors win in the end, and EDF keeps it’s iron strangle hold on the UK grid, or SMR’s win.. and the domestic UK offering is just better.
Moz of Yarramulla 10.05.25 at 3:57 am
In terms of investments, remember that the SMR companies are all startups. They don’t have a product, they have the concept of a product. Typically the expectation is that a startup, if it succeeds, will dominate the market it creates or radically reduce costs in the existing market it’s entering. Commodity electricity runs 2-10c/kWh and falling, so slashing the cost of that means coming in well below 1c ($10/MWh in market terms).
Or they need to target some other market, like aluminium smelters that need very reliable electricity in large quantites, the AI money mines that don’t have grid electricity in the quantity required, or cargo ships where running the grid to them isn’t practical. That’s a different market that pays a premium, but generally has other requirements.
Ai already has problems with public acceptance, saying “we have the first generation of SMRs on site too” isn’t going to help. The Savannah showed how popular civilian nuclear ships are, but that could be addressed by targeting specific routes – China to USA, for example, since both countries are nuclear powers already. Big industrial users in remote areas like smelters could be a valuable market, but they’re going to want a lease and high availability with savage penalties (the cost of re-heating a pot of should-be-molten aluminium is high).
Viz, there’s definitely a market but it’s either very cheap or very fussy.
Fake Dave 10.05.25 at 4:05 am
Tbe question of why countries would sign these deals and greenlight dubious projects is complicated by the military nuclear programs in pretty much all these countries. The OP and several commentators have already noted the importance of this technology in submarines, and that has implications for nuclear deterence, especially Britain and France, which don’t have ICBMs. The ability to build and innovate reactor designs in-country is crucial to the long-term reliability of that deterent. That presents an opportunity for for-profit contractors to sell countries on civilian projects that promise to preserve and expand nuclear engineering competency even if they make little commercial sense. The general souring of East-West relations over the last few years seems to track pretty well with this Starmer-Trump deal, AUKUS, and the revived talk of a nuclear “rennaissance.” What it doesn’t track with is the rise of renewables and batteries over the same period. 20 years ago it was possible to imagine nuclear was the key to quickly decarbonizing the economy, but now it just looks like deliberately pdriving in the slow lane.
jone 10.05.25 at 8:32 am
AI spam deleted – thanks to oldster for alert
oldster 10.05.25 at 12:22 pm
I’m pretty sure that “Jone” (author of two spammy comments in this thread) is an AI spambot that plants links to its paying advertisers into comment threads, embedded in a comment that has been made plausibly on-topic by the miracle of LLM’s.
We must build more generating capacity to support the brilliant future made possible by artificial intelligence!
Laban 10.05.25 at 6:53 pm
oldster – yes, in my blogging days the adverts were manually composed and pasted into the comments, now the miracle of AI means they can spam 100 blogs a time!
Tm 10.06.25 at 8:18 am
Thomas 17: Otoh if we stop burning oil and coal, almost half of global shipping would become obsolete!
https://www.forbes.com/sites/michaelbarnard/2023/12/05/how-will-climate-action-change-the-face-of-global-shipping/
Thomas Jørgensen 10.06.25 at 2:04 pm
If, as that article supposes, we switch ore processing to non-carbon processes, that isn’t going to reduce bulk shipping.
Far and away the cheapest place to do bulk electrolysis is going to be coastal equatorial deserts. Hydrogen is a pain in the neck to ship.. but those coasts either have, or can quickly build, ports to handle large quantities of both ore and refined metal. So less coal being shipped around, but new routes running iron ore to Casablanca and sponge iron everywhere from there.
Michael Cain 10.06.25 at 2:49 pm
Bill Gates and Warren Buffett have broken ground for an eventual sodium-cooled reactor in Wyoming. Gates owns TerraPower, the company doing the design work. Buffett owns PacifiCorp, a large vertically integrated utility that owns the site where the reactor will be built. They have hired GE-Hitachi, who has been through the process of getting a reactor design approved, to provide additional engineering expertise. The selected site is adjacent to a PacifiCorp-owned coal-fired power plant (in the process of switching to natural gas) so will have immediate access to the Western Interconnect transmission network.
An interesting feature of the design is that the secondary steam loop that actually generates electricity will be entirely air-cooled. This is a big selling point in the semi-arid western US.
Zamfir 10.08.25 at 9:44 am
@laban, one thing that changed, is that France moves to much more independent nuclear regulation (both on paper, and in reality by giving hard political support to the regulator in confilcts). The ASN is involved in most delaying problems in the Flamanville project, and also in the big Le Creusot problem with France’s older reactors. It predecessor (the DSIN) was a bit notorious in Europe for being uncomfortably tight with the other parts of France’ nuclear sector.
The implication (which is hard to prove) is that in the glory days, the French regulator would quietly accept engineering and production deviations that would have become showstoppers in the modern climate. The safety record of those DSIN-era reactors is pretty good, so perhaps they were pragmatic in a reasonable way -accepting small safety lapses where a fix would be extremely costly and delaying. Then again, a similar ‘pragmatic’ deal with the Japanese regulator resulted in the Fukushima accident.
K Kurtz 10.08.25 at 4:26 pm
Who was it who decided it was better to call Small Modular Reactors “SMRs” instead of “Smores” ?
Tm 10.14.25 at 8:41 am
Big news (only 30 years too late):
Solar and wind outpaced demand growth in the first half of 2025
Global electricity demand grew by 2.6% (+369 TWh) in the first half of 2025. This increase was more than met by increases in solar (+306 TWh, +31%) and wind (+97 TWh, +7.7%) generation, with solar alone covering 83% of the rise. Hydro fell significantly while bioenergy output dipped slightly, and nuclear rose modestly, while overall fossil generation fell marginally (-0.3%).
https://ember-energy.org/latest-insights/global-electricity-mid-year-insights-2025/
Note, these figures are actual generated electricity, not nominal power. China seems to have finally reached peak fossil, by aggressively building up wind and solar. They build nuclear reactors too, but they still play only a marginal role.
Globally, the increase in wind and solar generation (+403 TWh) far outpaces nuclear (+33 TWh). Only a fool can look at these figures and conclude that we should divert investment from fantastically successful renewables to expensive and unreliable nuclear. (Btw the Flamanville reactor, completed in 2024 after >20 years of work and 24 billion in cost, has been shut down since June.)
Tm 10.16.25 at 7:32 am
OT, there are still comments waiting on the “campus debate” thread.
Comments on this entry are closed.