Yves here. Nuclear reactors are back in vogue due to the expectation of ginormous AI driven energy demand requiring all sorts of sources being brought to bear. Consider these search results:
Some advocates have touted small nuclear reactors as one solution. Some of our commentors have thrown cold water on that idea. This post reinforces their views with its longer-form treatment.
By Leon Stille, who has background in energy sciences (MSc and BSc) and is pursuing a PhD in energy policy. He currently runs his own company, New Energy Institute, as an independent energy expert and is co-owner and director of Hovyu BV. He holds several teaching positions at universities of applied sciences and international business schools. Originally published at OilPrice
- SMR’s are being hailed as the perfect solution for large industrial power consumers.
- SMRs are currently being marketed like they’re the iPhone of nuclear energy: smarter, smaller, cheaper, scalable.
- Despite the hype, there are currently no SMR’s operating on a commercial scale.
You can feel the buzz: nuclear is back. Or so we’re told.
From Brussels to Washington, a new wave of enthusiasm for so-called Small Modular Reactors (SMRs) is sweeping through policy circles, think tanks, and energy startups. These compact, supposedly plug-and-play nuclear units are being hailed as the perfect solution to power data centers, feed artificial intelligence’s growing hunger, and backstop our energy transition with clean, stable electricity.
There’s just one problem. Actually, there are many. None of them small.
The Hype Cycle Is in Full Spin
SMRs are currently being marketed like they’re the iPhone of nuclear energy: smarter, smaller, cheaper, scalable. A miracle solution for everything from remote grids to decarbonizing heavy industry and AI’s server farms. Countries like the U.S., Canada, and the UK have announced ambitious deployment plans. Major developers, including NuScale, Rolls-Royce SMR, GE Hitachi, and TerraPower, have painted glossy timelines with glowing promises.
Except the fine print tells a different story.
There are currently no operational commercial SMRs anywhere in the world. Not one. NuScale, the U.S. frontrunner, recently cancelled its flagship Utah project after costs ballooned to over $9,000 per kilowatt and no investors could be found. Even their CEO admitted no deployment would happen before 2030. Meanwhile, Rolls-Royce’s much-hyped SMR factory hasn’t produced a single bolt of steel yet.
So, we’re betting on a technology that doesn’t yet exist at commercial scale, won’t arrive in meaningful numbers before the 2030s, and would require thousands of units to significantly contribute to global energy demand. That’s not a strategy. That’s science fiction.
Big Nuclear Hasn’t Exactly Inspired Confidence Either
Even the large-scale projects that SMRs claim to “improve upon” are struggling. Take the UK’s Hinkley Point C, once heralded as the future of nuclear energy in Europe. It’s now twice as expensive as originally planned (over £46 billion), at least five years late, and facing ongoing construction delays. The French-backed EPR reactor design it’s based on has already been plagued with similar issues in Flamanville (France) and Olkiluoto (Finland), where completion took over a decade longer than promised and costs ballooned dramatically.
Let’s be honest: if any other energy technology was this unreliable on delivery, we’d laugh it out of the room.
Price Floors for Nuclear, and Price Ceilings for Reason
In France and Finland, authorities have now agreed to guaranteed minimum prices for new nuclear power, effectively writing blank checks to ensure profitability for operators. In Finland, the recent deal sets the floor above €90/MWh for 20 years. Meanwhile, solar and wind regularly clear wholesale power auctions across Europe at €30–50/MWh, with even lower marginal costs.
Why, exactly, are we locking in decades of higher prices for a supposedly “market-based” energy future? It’s hard to see how this helps consumers, industries, or climate targets. Especially when these same nuclear plants will also require major grid upgrades, just like renewables, because any large-scale generator needs robust transmission capacity. So no efficiency win there either.
The SMR Promise: Too Small, Too Late
Back to SMRs. Let’s suppose the best-case scenario plays out. A couple of designs clear regulatory approval by 2027–2028, construction starts in the early 2030s, and the first commercial units are online before 2035. Even then, the world would need to build and connect thousands of these small reactors within 10–15 years to displace a meaningful share of fossil generation. That’s a logistics nightmare, and we haven’t even discussed public acceptance, licensing bottlenecks, uranium supply, or waste management.
For perspective: in the time it takes to build a single SMR, solar, wind, and battery storage could be deployed 10 to 20 times over, for less money, with shorter lead times, and with no radioactive legacy.
And unlike nuclear, these technologies are modular today. They’re scalable now. They’ve proven themselves everywhere from the Australian outback to German rooftops and Californian substations.
The Elephant in the Reactor Room: Waste and Risk
Nuclear fans love to stress how “safe” modern designs are. And yes, statistically speaking, nuclear energy is relatively safe per kilowatt-hour. But it’s also the only energy source with a non-zero risk of catastrophic failure and waste that stays toxic for thousands of years.
Why, exactly, would we take that risk when we have multiple clean energy options with zero risk of explosion and waste streams that are either recyclable or inert?
You don’t need to be a nuclear physicist to ask this: how is betting on high-cost, slow-deploying, risk-bearing, politically toxic infrastructure a better idea than wind, solar, and storage?
A Footnote in the Transition, Not the Headline
Let’s be clear: nuclear power will likely continue to play a role in some countries’ energy mixes. France and Sweden have legacy fleets. New projects may go ahead in China or South Korea, where costs are contained and planning is centralized. But for the majority of the world, especially countries trying to decarbonize fast, new nuclear is not the answer.
SMRs, despite their branding, will not save the day. They will be at best a niche, possibly a small contributor in specific applications like remote mines, military bases, or industrial clusters where no other solution works. That’s fine. But let’s stop pretending they’re some kind of energy silver bullet.
Final Thoughts
We are in the decisive decade for climate action. Every euro, dollar, and yuan we invest must yield maximum emissions reduction per unit of time and cost. By that standard, SMRs fall flat. Nuclear power, small or large, is simply too expensive, too slow, too risky, and too narrow in its use case to lead the energy transition.
So let’s cool the reactor hype. Let’s focus instead on the technologies that are already winning: wind, solar, batteries, heat pumps, grid flexibility, green hydrogen. These are not dreams. They’re deploying by the gigawatt, today. SMRs are fascinating, yes. But when it comes to decarbonization, we need workhorses, not unicorns.
Why is it that whenever there is a discussion about nuclear power stations there is rarely if ever any mention of the largest builder/operators of these plant? Rusatom are clearly world leaders in this field and their fast breeder technology is now proven and is in the process of being rolled out across Russia ,China and India. Just because ‘the west’ is myopic and blinkered doesn’t mean that solutions don’t exist that are increasingly being employed in the RoW.
Aye. The Russians, Chinese, and South Koreans have all demonstrated the ability to build nuclear power stations for less than $2500/kW (about 15% of what Hinkley Point C will cost). It’s clearly possible to deploy cost-effective nuclear. The West has forgotten how. [And I mean that fairly literally. When you don’t build reactors for 40+ years, the people who have the most experience end up retiring and passing away, and the people trying to build them today have to figure out some hard-learned lessons all over again. I knew a guy who helped design and build TVA’s nuclear fleet, and he would have been a good person to bring in as a consultant. Alas, he passed away last year at the age of 88.]
Thank you as always, Yves. Looking forward to a very interesting discussion about this.
I thought that new nuclear might solve the intermittency problem in the grid.
I am not an expert but what I have been hearing lately was that total renewables are not feasible for industrial power?
I am a Greenie by the way but we do need to be scientifically rational.
If we do need baseload power and nuclear cannot provide it do we still need coal? Or, more probably, gas?
Looking forward to NC experts educating me.
Renewables are more suitable for industrial power than domestic power, because most industrial users are less dependent on dispatchable energy – in other words, you can link the demand to projected power outputs more easily. Many industrial uses are in fact vital for the efficient roll out of renewables, because many can be tied into the use of surplus electricity to create products such as green hydrogen or ammonia. The astonishing drop in the price of solar actually has massive potential for revolutionising many different industrial power uses. The main problem though is linking investment cycles – current industrial users are designed for current grid designs – it takes time for industrial capital investment to catch up with changes in electricity networks (and vice versa).
Coal is not ‘needed’, except insofar as any large grid needs a variety of power inputs for stability. Every power source has its own vulnerability (cost, supply of power, safety considerations, vulnerability to weather, etc), so the more sources the better. This is why many coal plants are partially mothballed and kept available for years or even decades after they cease to be economically viable, and its probably one reason why China are building so many even as they massively build out renewables and nuclear.
Gas is utterly vital these days for both nuclear, coal and renewables. Only gas can provide the peaking power nuclear and coal can’t, and the balancing for renewables down-time (if you don’t have hydroelectric at scale). Natural gas power is in general very expensive, but requires relatively little capital investment relative to other forms (assuming you have adequate gas supply available). The higher the renewables penetration to a given grid market, the more vital gas turbines are for back up, although paradoxically, the more you build of both, the less gas you need (because they become a fall back energy source, not baseload). This is, incidentally, a source of major confusion – generating capacity does not necessarily equate to power use – all grids have massive overcapacity built in – they can’t operate efficiently otherwise. Campaigners getting all excited about a few gas peaking plants are missing the big picture.
Baseload, btw, depends (as such much does so much) on the nature of a particular grid, and all grids have variable characteristics. It also depends on how you define ‘baseload’ – traditionally it meant the very large thermal plants that chugged away in the background providing the basic electricity need, with other sources dealing with emergency, daily, or seasonal peaks, but in reality that’s not how the most modern grids work now. In Europe many national grids have little to do traditional thermal background load, utilising instead a mix of gas, renewables and import/exports. In many developing countries, especially in Africa, they are frequently finding that a focus on baseload is too inefficient – the very low price of solar (in particular) has meant that investing in decentralised grids is proving far more cost effective – they are sacrificing some reliability for low cost and speed of roll-out instead.
Nuclear doesn’t solve the intermittency problem as nuclear reactors are typically run on 100% or they are off. That doesn’t help intermittency as what you need for that is flexibility and nuclear is less flexible than wind, because it can’t increase when already at 100%, and it is worse at spilling.
The grid need as much production as usage at every time, so when production is less then demand you need to increase production or decrease demand. But if production is more then demand, you need to decrease production or increase usage. You can increase usage by turning on pumping to storage, or making the price low (or even negative) to induce demand. You can decrease production by paying producers to decreasing production by spilling energy. Ever seen a group of wind mills and every third is still? They are spilling energy because not all of it is needed at that point of time. Nuclear plants in general are bad at spilling energy as you don’t want to turn them on and off to much because it can cause strains in reactor parts. French nuclear plants can safely spill some of the energy, but wind is better at it.
Nuclear also has a problem in not actually being “on demand” when it has problems. It needs to be able to be taken offline at any risk, if it is to be run safely, which means that it can go offline at short notice. For example half the french nuclear plants had to be taken offline half a year in 2022, which together with gas supply problems and margin price electricity pricing, spiked electricity prices across a large portion of Europe.
Intermittent energy like wind and solar and baseload like nuclear contributes the same service to the grid at large, it saves more flexible top energy – like hydro or gas – for another day. New nuclear thus competes directly with new wind, and it can’t compete on price.
SMRs in turn creates new problems for nuclear electricity production and as the article points out, it is unclear if they really solve any.
“Nuclear doesn’t solve the intermittency problem as nuclear reactors are typically run on 100% or they are off.”
That may be how they’re operated, but it’s not a hard technical requirement. Even back in 2011, the French were routinely running their reactors between 50% and 100% (per https://www.oecd-nea.org/nea-news/2011/29-2/nea-news-29-2-load-following-e.pdf), and newer reactors are designed to be able to swing between 30% and 100% pretty quickly. This is adequate to follow demand variation of the course of the day.
Or, if you wish to build fewer reactors, you can deploy energy storage stations to cover the demand swings. To completely eliminate the need to power swings at nuclear stations, you’d need storage capable of providing about 25% of grid capacity for about 4 hours. This would be ~1000 GWh for the US. In contrast, keeping an all-renewable grid energized over an extended period of unfavorable weather would probably require storage capable of providing 70% of of grid capacity for two weeks. This would be ~250000 GWh for the US. 250X as much.
And how are we doing at deploying storage? Well, in California, where they’re pushing hard for renewables, they’ve deployed a “whopping” ~30 GWh of storage. They still “spilling” (i.e., curtailing) lots of renewable energy production during the days of sunnier months (per https://www.caiso.com/about/our-business/managing-the-evolving-grid), and they still burn fracked natural gas every single night to keep their grid up.
So, who writes the liability insurance policies for any nuclear reactor? New, old, large or small?
No private insurer will touch them because of the potential liability.
Result, the only slice of Soviet style Marxist command and control economy in America:
The Price Anderson Act.
SMR’s have been built and used for nearly 70 years now – the first viable one was in USS Nautilus (it was originally intended to be that fashionable thing – a sodium cooled reactor – but was changed to a simpler water cooled system which is still the basis for nearly all small reactors).
All five of the major nuclear powers have spent untold sums trying to get reactors for their naval forces cheaper and more efficient, and have tried many different designs (the Soviets in particular), but nobody has managed to make them any more viable for anything but the most expensive applications (i.e. nuclear subs and occasional aircraft carriers). Notably, the new Chinese supercarriers are diesel powered so far, hardly a vote of confidence in their own SMR investments.
SMRs may have a future role in grid stabilisation, especially due to their potential to provide baseline power in more remote locations, but they are nowhere near competitive yet with conventional large scale nuclear, let alone renewables, which continue to drop dramatically in price year by year.
PK: SMR’s have been built and used for nearly 70 years now – the first viable one was in USS Nautilus (it was originally intended to be that fashionable thing – a sodium cooled reactor – but was changed to a simpler water cooled system which is still the basis for nearly all small reactors).
No.
[1] The Nautilus was intended as LWR-powered from before its construction plans began being drawn up, and was built exactly that way.
[2] The USS Seawolf, the second US nuclear sub, built in 1953 and launched in 1955 —
https://en.wikipedia.org/wiki/USS_Seawolf_(SSN-575)
—did have a liquid metal-cooled (sodium) reactor. But the reactor had such problems while on its maiden voyage Rickover had it pulled out and replaced with a light water system in 1958-60.
The histories and contemporary accounts downplay the scope for disaster in putting a sodium-cooled reactor into the Seawolf. But: –
Firstly, anybody who’s taken Chemistry 101 and seen water brought into contact with sodium knows that it ‘triggers a vigorous, exothermic chemical reaction’ e.g. flames and an explosion;
Secondly, molten-salt-cooled reactors are highly corrosive of their innards, which over the decades has been their main difficulty despite their notional advantages, so putting one in a submarine was asking for trouble. Rickover and the Navy were lucky to get off so lightly.
[3] How did Rickover even get talked into the Seawolf experiment in the first place? The politics, though seventy-five-years-old, continues to reverberate to this day (as the politics of technology often does), so here it is briefly.
The LWR is essentially the Boiling Water Reactor (BWR) of Eugene Wigner and Alvin Weinberg, scaled down and optimized. Here’s Weinberg in his autobiography, THE FIRST NUCLEAR ERA (1994): –
…’Rickover and his officers had come to Oak Ridge to learn about the technology and, if possible, to launch a project to build a nuclear-powered submarine. The coolant for the nuclear-submarine reactor had yet to be determined. Rickover himself seemed to favor high-temperature sodium…I naturally called to their (Rickover’s team’s) attention the ideas we had developed on the use of pressurized water. For a submarine reactor, pressurized water had … advantages: firstly, a reactor based on it would be small enough to fit comfortably in a submarine … because the distance a fission neutron travels in water is about one fifth the distance it travels in graphite…
‘Rickover at first objected to water because its use limited the temperature of the steam to around 250C and this meant lower thermal efficiency. Rickover’s assistants, especially Lieutenant Commander Eli Roth, were impressed with my arguments, particularly … that the simplicity of a pressurized-water power plant far outweighed its lower thermal efficiency, and Roth (convinced) Rickover….
‘Thus was born the pressurized-water reactor–not as a commercialized power plant, and not because it was cheap or inherently safer … but because it was compact and simple and lent itself to naval propulsion. But once (it) was developed by the Navy, this system achieved dominance for central system power.’
Ironically, though, Alvin Weinberg always preferred fluid-fuel power breeders despite having all those BWR patents. So when he controlled Oak Ridge Lab from 1955-73, he pushed development of the Molten Salt Reactor (MSR) and the Liquid Metal Fast Breeder Reactor (LMFBR), with the MSR’s development being underwritten by USAF money, ostensibly as the reactor for Curtis LeMay and SAC’s proposed nuclear bomber aircraft (which the thorium reactor today is a direct outgrowth of.)
Likewise, Weinberg pushed for the second nuclear sub, the Seawolf, to have a MSR and Rickover, interested in the greater potential power, went along with it. A crazy experiment IMO, as the last place you want to put a sodium-cooled reactor is on a submarine, FFS!
Not true. Russia has been operating a couple floating reactors powering Pevek for five years now.
And there are half a dozen RITM-200 reactors being built to expand the fleet.
Plus, of course, the Bilibino NPP reactors, which have been operating since the 1970s, fit the SMR definition too
China has the HTR-PM operating too.
But yes, in the West this is more hype than substance. In principle though it is a viable path for certain applications. For grid-scale power however, large reactors are much more efficient solutions.
Any analysis of energy matters that talks about money goes straight in the trash bin.
The two things that matter here are energy return on energy investment and long-term sustainability, and while the price is not entirely disconnected from the former, it is often distorted by all sorts of factors that have nothing to do with it.
Fossil fuels score very poorly on long-term sustainability, and increasingly poorly on EROEI, as the high-grade resources get depleted.
Wind and solar are very bad on EROEI and also quite catastrophically bad, especially the current windmills, on long-term sustainability, because they can only be made right now thanks to a huge fossil fuel subsidy that drives the machinery.
Nuclear is the only thing that could have bought us enough time to figure out the future.
But it’s largely too late for that anyway.
The SMR criticisms are well directed, but there is no mention of fusion power, which promises to deliver a much more attractive solution for baseline power generation. The growing investment in this sector indicates that design breakthroughs are leading to possible deployed fusion reactors within a decade.
I’m sorry but I have been hearing about a breakthrough for fusion rectors meaning that they will be deployed in a decade’s time since the 1970s. Some scientists have spent their whole, complete entire careers trying to get these reactors to work. At this point in time we are more likely to get warp core reactors first.
And while it may be a triumph to have these seething dragons of plasma generate more energy than they consume for a fraction of a second, to be practical, you have to reliably couple then to an old-tech steam turbine and have them run indefinitely. Maybe it can be done? Maybe in what? 20 years time?
So what keeps the sun in check for billions of years until it runs out of fuel and explodes? It all sounds counterintuitive. If we understand the sun as an energy machine, what is the mechanism that creates stability? And when we engineer our own solar energy facilities what is our earthly mechanism for over heating? Fuse breakers? It seems like harnessing fusion is worth the effort.
So what keeps the sun in check for billions of years until it runs out of fuel and explodes?
Gravity, to be simplistic. Gravity on scales humanly inconceivable, to some extent.
Suns — stars — originate in vast, cold clouds of gas and dust which can span dozens of light-years till gravity takes over and lumps within the cloud coalesce, then start to collapse under their own gravity. As that happens, pressure and temperature rise, the core heats under gravitational compression and a protostar forms. If the protostar accumulates enough mass, it reaches the critical temperature of ~15 million °C and the gravitational pressure needed for nuclear fusion to commence.
Thereupon, hydrogen nuclei fuse and overcome their electrostatic repulsion and combine to form helium. This fusion releases vast amounts of energy as light and heat.
And — to answer your question — equilibrium is achieved. The outward pressure from fusion balances the inward pull of gravity, stabilizing the star and marking its entry into its main sequence phase—its long, stable burning period.During this phase, the fusion reactions in the core produce photons that travel outward, eventually escaping as visible light, ultraviolet radiation, and other forms of electromagnetic energy.
When this phase ends because a star has exhausted its nuclear fuel, it will expand and then collapse back in on itself. It will then undergo one of three fates, depending on its mass. (I’m vastly simplifying here.)
Massive stars go supernova. The gravitational forces in their cores are so vast they then become black holes.
A star whose mass is merely above the Chandrasekhar limit—more than 1.4 times the mass of Earth’s sun—will also explode in a supernova where, around a collapsing core that becomes a neutron star, an expanding shell of debris contains the heavier elements forged by stellar nucleosynthesis.
Finally, a star as relatively small as our Sun will first expand into a red giant, then the remnant will become a white dwarf, before cooling into a black dwarf — which as the Universe is still relatively young remains a theoretical object, as there hasn’t been time for any black dwarfs to exist yet.
You write: It seems like harnessing fusion is worth the effort.
Is it? Think about it. As stability in a star’s fusion process is achieved by gravitational pressure so colossal it creates a minimum temperature of approximately 10 million Kelvin, what sort of forces do we bring into play to replicate that colossal pressure?
The standard approach is to use a tokamak, a doughnut-shaped device that creates powerful magnetic fields to replicate it. But so far energy output from fusion produced in tokamaks has been negligible—not enough to exceed or match the input.
I know I’m dating myself and anyone else who recalls the reference, but this sounds an awful lot like the “energy too cheap to meter” line we grew up with in the 1950’s and ‘60’s.
Why does the growing investment indicate breakthroughs?
Correlation is not causation
What many of these articles don’t address are the implications and reactions to the combination of rising power prices globally — a new phenomenon since the dawn of the industrial revolution — power shortages as industrial users (including hyperscaling data centers) compete with consumers, and a out-dated and beyond creaky transmission and distribution system that no one wants to spend money on?
And this ignores the risks posed by climate change; hello CA, and other, wildfires, the fact that much generation requires large quantities of water, etc.
There isn’t much new here. We’ve known that atomic energy is expensive since Thatcher tried to privatize generation and couldn’t find buyers for the nukes. So what? If replacing one infrastructure tech (fossil) with another that slows the climate catastrophe then it’s perverse to additionally demand it make the banks a pot of profits.
And yes, there have been some spectacularly disappointing generator projects in the West. What causes these failures? Is the problem intrinsic to atomic energy (let’s compare with the rest of the world, as some commenters already did above) or is it another instance of the Neoliberal West discarding its capacity to get strategic projects done. I still sometimes follow the news from my home country of Scotland. Relatively simple projects such as a parliament building or ferry upgrades beclown us. Atomic power plants? pifft. NC frequently focuses on the loss of organization knowledge and skills in complex operations, how easy it is to lose, and how it takes a generation to rebuild even under the best of conditions that generally don’t exist today. I actually don’t know the truth of the matter here: are nukes too difficult in and of themselves or just too difficult for the Neoliberal West?
They’re just too difficult for the West. Russia, China, and South Korea have all recently built nuclear power stations faster and cheaper than we manage to do in the West these days. If you look at https://www.worldnuclearreport.org/reactors.html, you’ll see that the West’s nuclear efforts pretty much ended before 1990, but the rest of the world has kept chugging along. Indeed, things actually started ramping up in 2005, and China currently has 31 reactors currently under construction (with only 2 of them being considered “behind schedule”).
.Tom: If replacing one infrastructure tech (fossil) with another that slows the climate catastrophe then it’s perverse to additionally demand it make the banks a pot of profits.
There it is.
Our Duke Energy is pushing the SMR idea and wants rate payers to help pay the research bill according to an activist site up in North Carolina. If they get approvals such a reactor would not operate until 2035. Here’s Duke’s press release
https://news.duke-energy.com/releases/to-accelerate-the-exploration-of-new-nuclear-technologies-duke-energy-joins-industry-group-vying-for-doe-grant
As I’ve talked about before one of Duke’s failed nuclear projects is a mere fifteen miles away in Gaffney, SC where only the containment building was completed and found it’s only use years ago as a filming tank for James Cameron’s The Abyss.
One might even suggest there’s a nuclear/political industrial complex to go with that other one although various boondoggles have set it back in recent years. Obama was greatly supported by Exelon.
As a lover of scifi, “That’s not a strategy. That’s science fiction.” is offensive. I suggest “That’s fantasy.”
Seriously, given that the proponents are not idiots, I suspect its a misdirection (inducing shiny object syndrome) – the AI demands will require massive investment in the grid and power that is NOT happening (US power generation has been flat for decades, but steeply decreasing on a per capita basis).
My SWAG is that the highest ROI for the oligarchs (on an acceptable timeline – building out (and replacing) the existant power system ROI is on decadal timescales) is forced reductions in residential electric energy usage (by pricing out – passed back to the public as the cost of AI).
Holding the latest shiny “wonderwaffen” in front of a largely uninformed public ensures that it will be too late to build out the national power supply to meet demand, ensuring public impoverishment is the only option.
Failure to plan is planning to fail.
Also, the absence of a Western fuel supply, except for waivers on sanctioned Russian fuel, is a rather unattractive investment point.
Nuclear is, to quote, “the only energy source with a non-zero risk of catastrophic failure and waste that stays toxic for thousands of years.
Why, exactly, would we take that risk when we have multiple clean energy options with zero risk of explosion and waste streams that are either recyclable or inert?”
I’m not sure what Leon Stille means by “risk of explosion”. If he means a massive explosion like an atomic bomb, then he’s perpetuating a myth that has never been true of nuclear power stations. If he means a more mundane hydrogen explosion (like happened at Fukushima and Chernobyl), then it’s true, there is some risk. But it’s worth noting that nobody is building exact duplicates of the reactors at Fukushima or Chernobyl, as many lessons have been learned in the 50+ years that have elapsed since they were built. The newer reactors are safer.
And renewable energy has some explosion risks of its own. Probably the biggest example we’ve seen thus far was the big battery station fire at Moss Landing back in January (https://www.technologyreview.com/2025/02/13/1111843/battery-fire-moss-landing-power-plant/), where ~900 MWh of battery burned for several days and required the evacuation of thousands of people. I still haven’t seen an evaluation of how nearby soils were contaminated, but given the substantial lithium (and other metal) content of the fire, I expect it to be significant.
And if we go all-in on renewable energy, we’re going to need hundred of thousands of stations of this size.
New designs are safer, but catastrophic risk and the liability for it still exist. There are questions of money and organization that will have to be settled as well.
Insurer testimony leading to the adoption of the Price-Anderson Act during the 50s established that insurers would not sign up to insure the plants. So Congress passed the law to cap utility liability for any accident. Currently, the utility has to carry $500 million of commercial insurance per site, and damages above that amount are paid by assessments on all other nuclear plants, but the total cap is about $15 billion. A Forbes article put the cost to Fukushima refugees at $60 billion, and, as another example, the Camp Fire in California caused $15 billion in damage. Congress periodically resets the liability limits, but does not even keep up with inflation.
As readers here will be aware, insurers during the aftermath of the large California fires have demonstrated skill in delaying and minimizing payouts, which turn out to be inadequate. In the case of the Camp Fire, the judge flat out stated that the victims would not be made whole. Right now we’re living in a time where airplane makers build planes that kill people, in order to fund stock buybacks — can a corporation under strict financial market discipline safely operate a nuclear power plant?
If we’re going to assume the risk of nuclear technology, we need a system that gives incentives for safety and can make all victims whole — otherwise, any money made by plant operators will be just a proactive transfer from the damaged public to the stockholders of the enterprise.
> “nobody is building exact duplicates of the reactors at Fukushima or Chernobyl, as many lessons have been learned in the 50+ years that have elapsed since they were built. The newer reactors are safer.”
Yes, but safety in this case is another outcome of successful strategic infrastructure engineering efforts. So “They’re just too difficult for the West” applies to safety too. The root cause a lot of us arrived at for the 737 MAX-800 crashes was because the strategic purpose of the Boeing corporation, its reason to exist, changed from making great airplanes to maximizing shareholder value. (Because markets; go die.)
I’m in favor of atomic energy not instead of anything else but together with everything else including renewables and aggressive demand reduction (which I believe has to mean shrinking energy-intensive economies). Atomic energy is not more risky than what we know is coming without it.
I wonder how far along this effort has gotten (this is from 2021):
Billionaires Back Building of Small Modular Reactors at Abandoned Fossil Site
https://jpt.spe.org/billionaires-back-building-of-small-modular-reactors-at-abandoned-fossil-site
Broke ground for the plant in 2024:
TerraPower Begins Construction on Advanced Nuclear Project in Wyoming
https://www.terrapower.com/terrapower-begins-construction-in-wyoming
This is from 2022 also:
China Powers Up the World’s First Commercial Onshore Small Modular Nuclear Reactor
https://interestingengineering.com/science/the-worlds-first-small-modular-nuclear-reactor-is-sending-power-to-the-grid
Hm, China managed to do this without billionaires, that’s a novel approach.
The US Army has experience with this sort of thing: SL-1.
And then there was this one-
https://theconversation.com/the-us-army-tried-portable-nuclear-power-at-remote-bases-60-years-ago-it-didnt-go-well-164138
James Mahaffey’s book “Atomic Awakening” is a goldmine of information about the history of nuclear power, and covers a whole bunch of stuff, including both those reactors. His others – Accidents and Adventures are also good, but don’t eclipse Awakening.
In Awakening he describes how, as a neophyte operator, he and his buddies would engage in reactor racing – seeing who could bring a reactor from cold to operating power in the shortest tim.
For folks interested in excruciating details about driving a PWR, check out How to Drive a Nuclear Reactor by Colin Tucker.
Ah! Thanks for these.
What about China’s progress with thorium molten salt reactors? I understand that China has succeeded in re-fueling an experimental operating thorium reactor which was a major milestone. Thorium, apparently, is abundant and much safer than other nuclear reactor fuel and cannot be weaponized.
Apparently, the US was looking into this in the ’60s and decided against it, probably because it did not give them nuclear waste which could then be weaponized . Now, China is making it work and it offers the possibilities for small in-town (population centers) or even home power generation.
I was surprised that there was no mention of this in the article. Do I have this completely wrong..?
Here is the article by Hua Bin:
hXXps://wentworthreport.com/2025/04/28/china-leads-the-world-nuclear-breakthrough/
mgr: Thorium, apparently, is abundant and much safer than other nuclear reactor fuel and cannot be weaponized.
No. The thorium reactor is a civilian reactor technology that extremely efficiently converts to a high-grade weapons material breeder technology, with minor reconfigurations in the plant layout. If one is building such a reactor, then one can build in a breeder cycle section and it becomes the fastest easiest path to large quantities of high grade fissile U-233. With a little more work, one can also get weapons-grade plutonium out of it.
See–
U-232 and the Proliferation-Resistance of U-233 in Spent Fuel (2001) by Kang and von Hippel.
Moreover, once you set up a thorium reactor in the requisite way, it’s more compact and not as energy input intensive as centrifuges, and doesn’t need large input stream/regular fuel turnovers and fuel reprocessing.
Link –
https://fissilematerials.org/library/sgs09kang.pdf
For ‘U-232 and the ProliferationResistance of U-233 in Spent Fuel’
The articles I skimmed suggest it will be fully online in 2030, so it will be unavailable (assuming no unforeseen technical hurdles) to provide even minimal power until the mid-2030s, and possibly could play a significant role in the 2040s – way longer than the AI rollout timeline for power demands.
Thanks everyone for all the feedback and links!
I may be worth reading this Substack, and it’s follow up, to understand why constructing nuclear plant in the West is so expensive:
Did you forget a link? I’m curious to read these articles on Substack.
I’ll try again:
It’s a pretty good short review. I like the 5steps.
Probably the biggest one is the difficulty in changes learned during construction for build 2,3,4 etc.
In today’s world with 3d building programs most of that should be eliminated, but not all.
And it’s why the proponents of 300mw reactors think it’s the better model vs fewer 1200mw. Repetition greatly speeds up construction and reduces cost.
But we can’t leave out the utilities. There are no 2 sites that have the same build which greatly increases costs, which increases profits for investors.
And we also can’t leave out the builders whom would cut every corner they could to make more money, which does explain some of the regulatory hurdles.
Lots of blame to go around.
Ah, I found it, thank you. Along with part two:
Part 1: https://jackdevanney.substack.com/p/the-auto-genocidal-tragedy-of-us
Part 2: https://jackdevanney.substack.com/p/the-auto-genocidal-tragedy-of-us-9e9
And before you responded, I found a different but similarly-themed article on Substack:
https://schlanj.substack.com/p/how-the-us-can-make-nuclear-energy
Both articles contain some depressing reading about how nuclear power stations got ever more costly to build over time. It didn’t have to be that way.
Great stuff. I just scanned though them; I’ll download and read slowly tomorrow.
@ Bill. Thanks
Nice article bill
It took me a while to figure out that
I’ll try again was an active link
The article fails to even discuss what does SMR mean? How do you define it.
Here you go. It comes in two parts. Part 1 is size, roughly 1-300MW is sort of the generally accepted range.
Part 2 is technology. There is no specific technology for a SMR. It can be Gen 4 of any type or Gen 3 PWR.
The only data point given is the NuScale which is Gen4 molten salt/pebble bed. However there are others in the approval process of 300MW Gen3 PWR reactors.
What most people hear or think is very small made in a factory delivered on a truck and you hook it up to the grid. yes those are being worked on and roughly 5-50MW.
There are large discussions of engineers and the economics of 4 x 300MW vs 1 x 1200MW ( like the AP1000).
The idea that nuclear is expensive is incorrect, as others have mentioned above, China is doing it for roughly $2000 KW with roughly 4 year start to finish times, these are actual numbers of actual plants having been built. Some 25 under or starting construction and 150 by year 2035.
Like solar, the US has to decide how you want to do things. They could actually do what the Chinese did and still do: have foreign companies come in and build stuff, watch and learn how to do it, then improve on it and then make it yourself. Nah.
Can nuclear be a part of the low carbon ecosystem, absolutely. And like all of other options, each has its best locations and reasons for it.
There has been discussion that people are not going to put up with data center added electricity costs.
Check out this video of the gov of Pennsylvania saying exactly that.
Things are moving pretty quickly
https://m.youtube.com/watch?v=NyJ8W0T6ZvA