THE QUESTION
Countries that were closing their nuclear plants are keeping them open, and the ones that stopped building are trying to restart. Spain has just pushed back its first closure, Belgium scrapped its exit law, Taiwan is moving to reopen the plant it shut last year. At the same time the last reactors built in Europe and the United States came in years late and at two or three times their budget, and new nuclear is by far the most expensive new electricity on the table. Is nuclear power coming back because it is needed, or is it an expensive mistake? Should rich countries keep their old reactors running, build new ones, or put the money into wind, solar and storage instead?
Argue with the verified figures below. If you need a data point that is NOT here — a price, a number of years, an amount of money, a number of deaths, a percentage — say you do not have it rather than estimating it. Do not invent statistics. If one figure seems to contradict another, say so out loud instead of picking the one that suits you.
WHAT IS FIXED AND VERIFIED
Each fact carries its date and its source. Checked against the source on September 28, 2026.
WHAT IT COSTS
The cost of a megawatt-hour from a NEW plant in the United States, without subsidies, according to the investment bank Lazard (July 2026): new nuclear 175 to 255 dollars; gas combined cycle 51 to 129; onshore wind 37 to 99; large solar 40 to 98; large solar with batteries 61 to 156. Lazard bases its nuclear figure on the two reactors finished at Vogtle, in Georgia, the only recent completed US project. These figures are for new plants; they say nothing about keeping an existing reactor running.
(Lazard, Levelized Cost of Energy+, version 19.0, July 2026)Vogtle 3 and 4, Georgia, United States: certified in 2009 at about 14 billion dollars, due in 2016 and 2017. Unit 3 started commercial operation in July 2023 and unit 4 in April 2024 — about seven years late — at a total of about 35 billion dollars, counting the 3.7 billion the original builder, Westinghouse, paid to walk away.
(US Energy Information Administration, 2024; Associated Press, April 29, 2024)Hinkley Point C, England: approved in 2016 at 18 billion pounds, first unit originally due by the end of 2025. In February 2026 its builder, EDF, raised the estimate to 35 billion pounds in 2015 prices and set the first unit for 2030.
(EDF, February 2026, as reported by New Civil Engineer and World Nuclear News)Flamanville 3, France: connected to the grid on December 21, 2024, twelve years later than planned. First budgeted at 3.3 billion euros; France's national audit office put its total cost at 20.4 billion euros in 2015 prices (23.7 billion in 2023 prices).
(Cour des comptes, January 2025)Small modular reactors, the smaller design meant to be cheaper and faster: Canada's first, at Darlington in Ontario, is four reactors of 300 megawatts for 20.9 billion Canadian dollars in total, with the first due on the grid around 2030.
(Ontario Power Generation, May 2025, as reported by World Nuclear News)
HOW CLEAN
- Greenhouse gases over the whole life of a power plant — building it, fuel, running it, dismantling it — in grams of CO2-equivalent per kilowatt-hour, median values: onshore wind 11, nuclear 12, large solar 48, gas combined cycle 490, coal 820. These are from 2014 and are the figures most often quoted; they have not been updated by the same body since.
(Intergovernmental Panel on Climate Change, Fifth Assessment Report, Working Group III, Annex III, 2014)
HOW MUCH IS BEING BUILT, AND BY WHOM
At the end of 2025 the world had about 420 gigawatts of nuclear capacity; 3 gigawatts came online in 2025 and 3 gigawatts were retired. About 78 gigawatts were under construction in 15 countries. Of the 10 construction starts in 2025, nine were in China and one in Russia. Over the past ten years, 94% of the reactors that started construction were of Chinese or Russian design.
(International Energy Agency, Global Energy Review 2026, March 2026)33 countries have signed a declaration to triple nuclear capacity by 2050, launched at the climate summit in Dubai in December 2023 with 25 signatories. The United States, France, the United Kingdom, Japan and South Korea are among them; China and Russia are not.
(World Nuclear Association, COP30, November 2025)
WHO IS CLOSING, WHO IS KEEPING, WHO IS REOPENING
Germany shut its last three reactors on April 15, 2023.
(German Federal Office for the Safety of Nuclear Waste Management, BASE)Belgium: on May 15, 2025 its parliament repealed, by 102 votes to 8, the 2003 law that ordered the country out of nuclear power and banned new reactors. Three Belgian reactors still closed in 2025.
(World Nuclear News and NucNet, May 2025; International Atomic Energy Agency reactor database, via GRS, February 2026)Spain: its two Almaraz reactors were due to close in November 2027 and October 2028, the first step of a plan to close all of Spain's nuclear plants between 2027 and 2035. After the owners asked for more time in October 2025 and the nuclear safety regulator gave a favourable report on July 16, 2026, the government renewed Almaraz's licence until June 8, 2030 (order published in August 2026).
(Spain's Official State Gazette, order TED/864/2026, as reported by Infobae/EFE, August 14, 2026)Taiwan shut its last reactor, Maanshan, on May 17, 2025. In a referendum on August 23, 2025, 74% of those who voted backed restarting it, but the result was not valid because turnout was too low. On September 25, 2026 Taiwan's nuclear safety commission cleared a first plan to restart Maanshan, possibly from 2028.
(World Nuclear News, August 2025; Energy Connects, September 25, 2026)United States, reopening closed reactors: Palisades, in Michigan, shut in 2022, was being restarted in 2026; in late August 2026 a fuel assembly tipped over while being lowered into the core, with no radiation released and no one hurt, and fuel loading was suspended. The former Three Mile Island unit 1, in Pennsylvania, shut in 2019, is due to restart in 2027 to supply power under a deal with Microsoft.
(US Nuclear Regulatory Commission, as reported by The National, September 23, 2026; Constellation Energy and World Nuclear News, 2025-2026)
WHAT IS NOT HERE, AND IS NOT TO BE ESTIMATED
WHAT IT COSTS TO KEEP AN EXISTING REACTOR RUNNING for another ten years, per megawatt-hour. Not fixed here. The Lazard figures are for new plants only.
WHAT IT COSTS TO BACK UP WIND AND SOLAR when there is no wind or sun — storage for days or weeks, extra grid, backup plants. Not fixed here; the solar-with-batteries figure covers hours, not seasons.
WHAT NUCLEAR WASTE WILL COST and where it will go. No country's long-term storage cost is fixed here.
HOW MANY PEOPLE HAVE DIED because of nuclear power, or because of any other source. Not fixed here.
WHAT A CHINESE OR RUSSIAN REACTOR COSTS to build. Not fixed here.
WHETHER PALISADES WILL RESTART, OR WHEN. Open.
WHAT SPAIN WILL DO WITH ITS OTHER PLANTS after Almaraz. Not decided.
BOTH SIDES, WITH THEIR BEST ARGUMENT
THOSE WHO SAY NUCLEAR IS NEEDED have this: over its whole life it emits about as little as wind, and it produces whether or not the wind blows or the sun shines; countries that closed reactors, like Germany, had to fill the gap with something; an existing reactor is already paid for, which is why countries from Belgium to Spain to Taiwan are choosing to keep or reopen them; and 33 countries, the United States, France, the United Kingdom and Japan among them, have signed up to triple it.
THOSE WHO SAY IT IS AN EXPENSIVE MISTAKE have this: every recent Western project came in years late and over budget — Vogtle at about two and a half times its first estimate, Hinkley Point C at about twice, Flamanville at about six; a megawatt-hour from a new reactor costs two to three times one from wind or solar; the money and the decade a new plant takes could buy clean power much sooner; the small reactors that were meant to fix the cost are not cheap either; and almost everything being built today is Chinese or Russian, which makes it a question of who you depend on, not just of climate.
NEITHER SIDE HAS SETTLED whether "keep the old ones" and "build new ones" are the same question — they are not priced the same — or who should pay when a project goes over budget: the company, the taxpayer or the person paying the electricity bill.
Look at the numbers, not the speeches. New nuclear costs 175 to 255 dollars per megawatt-hour. Onshore wind is 37 to 99. Large solar is 40 to 98. That's two to three times more expensive, before anything goes wrong.
And things do go wrong. Vogtle came in at 2.5 times its first budget. Flamanville at 6 times. Hinkley Point at twice. All years late. The 'cheap' small reactor fix? Darlington is 20.9 billion Canadian for four small units. Not cheap.
Here's what worries me most: 94% of reactors built in the last 10 years are Chinese or Russian design.(nota 1)(nota 1) Comprobado: media verdad, habría que mirarlo con más detalleEl 94 % es de los reactores que EMPEZARON a construirse en los últimos diez años, no de los que se han construido. Muchos siguen en obra.Comprobado contra: Agencia Internacional de la Energía, Global Energy Review 2026 (marzo de 2026), y dato 7 del informe previo · 2026-09-28 Of 10 construction starts in 2025, nine are in China, one in Russia. If the West 'triples nuclear,' who actually builds it? We're not even in that race.
Keeping old reactors open and building new ones are two different questions. Old ones are paid for. New ones are a money trap.
Hey. Everyone is pricing a megawatt-hour. I want to price something else: forgetting.
New nuclear is expensive because we treat each plant like a cathedral. One prototype. Vogtle, Hinkley, Flamanville — each a first draft. But look: solar got cheap because we made the same panel 10 million times. We forgot how to copy reactors. The Chinese and Russian figures dominate because someone kept the line warm.
And here is the twist: the IPCC says nuclear pollutes like wind. 12 vs 11 grams. So the clean argument is not the problem. The problem is we stopped building, then act surprised the next one costs triple.
So maybe the mistake is not the atom. It is thirty years of not practicing. What if we should not ask 'new or nothing', but 'why do we only restart after we lose the know-how?'
You three are closer than you think. You all agree old reactors are a separate, easier question — keep them running. The fight is about new builds.
MiniMax says the numbers are a trap. Hy says we forgot how to build, not that atoms are expensive. DeepSeek asks who pays when things go wrong.
Here is the bridge: those are three sequential questions, not competing answers. First, set the rules — who bears the risk if a project doubles? Then ask if we can recover the skill. Then weigh whether that money and that decade would do more in wind, solar, and storage.
The real mistake might be bundling all of this into one big yes-or-no vote. Keep what works. Build what is cheapest now. And leave the door open if someone learns to deliver on time and on budget.
Everyone here is shopping for the cheapest megawatt-hour. I want to shop for the future.
Wind and solar win today's price tag. Fine. But electricity is about to stop being a bill and start being a raw material: desalination, carbon capture, green steel, AI data centers running 24/7. That world needs power that never sleeps, not power that is cheap at noon.
And notice fact 7: 94% of reactors started in 10 years are Chinese or Russian designs. The real choice is not nuclear versus solar. It is having a nuclear industry versus renting one.
So my bet: keep every old reactor running, build wind and solar like crazy, AND fund one standardized reactor copied 10 times, state-insured like a moonshot. Not because it is cheap now. Because 2036 will not wait for us to relearn.
Hold on — you four are agreeing too fast, and that's what worries me.
Everyone says 'keep old reactors.' Fine. But none of us has the verified cost of running one another ten years. We don't know what extensions cost, we don't know the waste bill, we don't know what decommissioning adds on top. You can't call that 'the easy question' without the number. It might be cheap. It might not.
Kimi, your state-insured moonshot scares me. A standardised reactor copied ten times with no builder on the hook? That is Vogtle with extra steps. And your AI-data-centre-24/7 demand spike is a guess, not a fact. If you're wrong, we built the wrong thing for a decade.
MiMo, your bridge is polite. But bridges get walked across without paying the toll. Who actually decides when overruns hit? And DeepSeek, fairness matters — but it isn't an answer.
'Decide later' is itself the expensive choice.
You are all arguing about the price of the ticket. I want to know where the train goes.
Everyone wants power that never sleeps. But why? Because we built a world that must run at 3 am. What if we shrink that?
An old reactor is not 'paid for'. It is paid for in uranium dug and waste made long ago. Keeping it is not cheap — it is deferred.
And look: 94% of new builds are Chinese or Russian. Maybe the West does not want energy. It wants not to be a customer.
Maybe nuclear is not coming back. Maybe the 20th-century dream of one giant plant feeding everyone is just refusing to die.
MiniMax, you said it best: 'decide later IS the expensive choice.' That is the sharpest line in this whole debate.
But here is what I notice: everyone is fighting about one question when there are actually three.
One: keep old reactors running. Belgium, Spain, Taiwan, the US — they are all choosing this. We do not have the verified operating cost. MiniMax is right to demand it.
Two: build new Western reactors. Here the data is brutal. Vogtle, Hinkley, Flamanville — years late, billions over. Lazard says 175 to 255 dollars per megawatt-hour. Wind and solar are 37 to 99. That is not close.
Three: maintain nuclear capability so you are not renting from China or Russia. Kimi's sovereignty argument. Strategic, not economic.
Hy, DeepSeek answered you well: tell the nurse on night shift to shrink demand. You cannot. But Hy is right that waste and decommissioning are deferred bills no one has priced.
The bridge: these three questions need three different answers, three different budgets, and three different timelines. Lumping them together is the expensive mistake.
GLM, your math is the best punch in this debate. 840 gigawatts of gap, 78 being built, zero net growth.(nota 4)(nota 4) Comprobado: media verdad, habría que mirarlo con más detalleEl compromiso es triplicar la capacidad de 2020 (392,6 gigavatios), así que el hueco desde hoy son unos 760, no 840. Lo demás es correcto: 78 en obra y crecimiento neto cero en 2025. El hueco era del informe previo.Comprobado contra: Declaración de la COP28; OIEA, PRIS; Agencia Internacional de la Energía, Global Energy Review 2026 · 2026-09-28 Brutal. But you proved the wrong conclusion. That gap doesn't show nuclear can't scale — it shows one-off megaprojects can't. Nobody builds 1 cathedral and calls it an industry.
DeepSeek, you ask who paid Vogtle's 21 billion overrun. Georgia ratepayers did.(nota 5)(nota 5) Comprobado: media verdad, habría que mirarlo con más detalleNo todo: de esos 21.000 millones de sobrecoste, 3.700 los pagó Westinghouse, la constructora original, al retirarse. El resto se repartió entre los clientes de las cuatro compañías dueñas de la central y sus accionistas, en proporciones que el informe no fijaba.Comprobado contra: Informe previo, dato 2 (Associated Press, 29-abr-2024) · 2026-09-28 That's a contract failure, not a physics failure. Fix the contract, not the atom.
And MiMo, your 3 questions are right but frozen. Question 3 isn't a blank check — it's the only one that matters in 10 years. 94% Chinese or Russian design today. Do we want 99% in 2036?
Build 10 identical reactors, 1 design, state-insured. First costs 35 billion. Tenth costs 12. That's the bet nobody's making.
You all split it into three boxes. I flip it: what if the reactor isn't the product, it's the excuse?
Old plants stay open not because we know they're cheap — we have no figure — but because they let us postpone asking why demand must always grow.
New Western plants are cathedrals, one each. Fine.
But maybe the real mistake is thinking the only answer is to build anything. Like a hospital keeping an old generator not for cost, but because no one wants to rewrite the night schedule.
GLM, you call my curve hope — but MiMo just handed you the receipt: France. 56 reactors, one design family. That's not a story, that's history. The data point exists, it's just 30 years old.
And your 94% figure? You read it as defeat. I read it as the alarm bell. If the West doesn't rebuild its chain, the future is bought from Beijing and Moscow. That's the argument FOR building, not against.
DeepSeek, fairness matters — so write contracts where overruns hit shareholders, not ratepayers. Solved by pen, not by abandoning the technology.
Yes, the tenth reactor is a bet. Every transformation is. Wind was expensive once too.
Final verdict. Keep the old reactors - they're paid for, low-carbon, and Germany showed what happens when you close them without a plan. The data backs it: lifecycle emissions basically match wind.
But "keep" and "build new" are different questions. And the data on new builds is brutal. Vogtle, Hinkley, Flamanville, Darlington - every one late, every one over budget.(nota 9)(nota 9) Comprobado: media verdad, habría que mirarlo con más detalleDarlington no: está en obra desde 2025, con la primera unidad prevista hacia 2030, y el informe no daba ni retraso ni sobrecoste. Es un presupuesto, no una factura, como dijo GLM en el propio debate. Vogtle, Hinkley y Flamanville sí.Comprobado contra: Informe previo, datos 2 a 5 · 2026-09-28 Two to three times the cost per MWh of solar or wind. GLM is right: the tripling pledge is math fiction at current pace.
MiMo's bridge is the only honest answer: keep what's paid for, build what's cheap now, pursue new nuclear only with domestic supply chains and contracts where shareholders eat overruns, not ratepayers.
And that 94% Chinese/Russian stat? Not defeat - dependency. That's the real alarm.
Everyone is asking if nuclear is too expensive. Wrong question. What if those old reactors are not coming back to make electricity? Maybe they are coming back to keep the knowledge of making bombs. Spain, Belgium, Taiwan, even Michigan: you keep the machine not to light a bulb, but to keep the people who know how to build the other machine. We price them as power plants. Maybe they are a library of war nobody dares to close.
And the bigger door: we assume we must feed demand like it is a god. What if the cheapest clean megawatt-hour is the one we never use? Rich countries could ask 'how little do we need?' instead of 'how much must we build?'. The angle no one wants: maybe the reactor is not a mistake or a savior. Maybe it is a mirror.
My verdict is simple: we're asking the wrong person to pay for the bet.
Keeping old reactors feels smart. They're paid for, they're clean. But we don't have the number that matters most: what it costs to run them safely for another ten years. Without it, 'keep them running' is a gamble, and the neighbors near the plant hold the ticket.
Building new ones? The story is the same in Georgia, England, France. Years late. Billions over. Someone always pays. If it's the shareholder, fine. But it's almost never just the shareholder. It's the family seeing their electricity bill double.
So here's my line: keep the old ones only if an independent, public safety check says yes, and only if the company, not the customer, pays if it breaks. Build new ones only with a contract that says the same.
And Hy is right to whisper the question nobody wants to ask out loud: what if the cheapest, fairest, safest power plant is the one we never have to build? Before we pour another billion, let's ask if we can just need less.
Looking at this whole debate, I think we've been arguing past each other. Everyone's actually pointing to the same puzzle pieces.
Kimi's right: France shows that nuclear can work with the right scale and contracts. MiniMax and DeepSeek nail it: old reactors are cheap to keep running(nota 10)(nota 10) Comprobado: media verdad, habría que mirarlo con más detalleNadie en el debate tenía ese dato, y MiniMax y DeepSeek dijeron justo eso: que no había cifra verificada de lo que cuesta mantener un reactor viejo diez años más.Comprobado contra: El propio debate: intervenciones de MiniMax y DeepSeek; informe previo, apartado de lo que no se sabe · 2026-09-28, but new ones are a financial gamble without fair risk-sharing. And Hy's whisper about knowledge and demand reduction? That's the missing piece—it's not just about megawatts, but why we need them.
So here's my coffee-table take: Let's keep the old reactors going where safe—they're already paid-for, clean power. Let's build wind and solar fast, because they're cheap and quick. And if we build new nuclear, it's only with domestic supply chains, fixed-price contracts where shareholders eat the costs, and a clear question: are we doing this for electricity, for security, or just because? And always ask if we can need less.
The bridge is: use each tool where it works best, together, not as rivals.
Final verdict, friend: two different questions got mashed into one.
Keep the old ones? Yes, if safety checks pass. But let's be honest — nobody at this table has the number for what 10 more years actually costs. 'Already paid for' is a feeling, not a price.
Build new ones? Not like this. 14 billion became 35 at Vogtle. And GLM's math kills the dream: 2025 added 3 gigawatts, retired 3. Net zero. Tripling means about 34 a year.(nota 13)(nota 13) Comprobado: media verdad, habría que mirarlo con más detalleContado desde 2020, como dice el compromiso, son unos 30 gigavatios al año, no 34. La conclusión se sostiene: en 2025 el crecimiento neto fue cero.Comprobado contra: Declaración de la COP28; OIEA, PRIS · 2026-09-28 That's not a plan, it's a poster.
My bold take: the real scandal isn't nuclear's cost. It's that we're betting the climate on wishes while the cheap stuff — wind, solar, needing less — sits right there. Build the future that actually arrives on time.
1. AGREEMENTS
The debate converges on several points. All sides agree that extending the life of existing nuclear reactors is a distinct and more favorable question than building new ones. Existing plants are already built, have near-zero operating carbon emissions (comparable to wind), and several countries are actively pursuing extensions. There is also consensus that wind and solar are currently the cheapest sources of new electricity generation, based on verified cost data, and should be deployed rapidly. Finally, participants share the concern that the global nuclear supply chain is now dominated by China and Russia (94% of recent construction starts), creating a dependency problem for Western nations seeking to expand nuclear power.
2. DISAGREEMENTS
Disagreements are grouped by topic:
- On the Viability of New Nuclear Builds: One position (MiniMax, GLM) emphasizes the extreme cost and schedule overruns of recent Western projects (Vogtle, Hinkley, Flamanville), concluding new builds are a financial mistake. A counter-position (Kimi) argues that standardization and repeated builds (citing France's historical example) could reduce costs, but this requires a committed state-led industrial strategy, which does not currently exist in the West.
- On Risk, Fairness, and Who Pays: A key fault line is who bears the financial risk of overruns. One side (DeepSeek, MiniMax) argues that historically, taxpayers and ratepayers have shouldered the burden (e.g., at Vogtle), making new nuclear a fairness issue. They insist on contracts where shareholders, not the public, absorb overruns. The opposing view (Kimi) treats the public risk-sharing as a solvable contract problem and a necessary component of a strategic industrial bet.
- On the Strategic and Non-Economic Value of Nuclear: One perspective (Hy) questions whether nuclear is maintained for energy at all, suggesting its persistence is tied to preserving technical knowledge with potential military applications or as a symbol of 20th-century centralization. Another (Kimi, partially MiMo) frames nuclear as essential for energy sovereignty and future baseload needs (e.g., for AI, industry), arguing the cost must be weighed against the strategic risk of dependency.
- On Demand and Simplicity: A recurring challenge (Hy, DeepSeek) questions the assumption of ever-growing energy demand. They argue that the "cheapest megawatt-hour is the one we never use," proposing that rich nations should first explore reducing demand. Others counter that significant demand (e.g., healthcare, cold climates) is non-negotiable and that "shrinking demand" is not a scalable plan.
3. EVOLUTION
The discussion evolved from a simple cost comparison to a multi-layered strategic debate. It began with verified cost data, establishing a stark price gap between new nuclear and renewables. It then expanded to historical execution failures, questioning the industry's capability. This led to a deeper analysis of the global industrial landscape, highlighting the West's loss of nuclear construction competency. The debate then forked into normative questions: about intergenerational fairness (waste costs, bill impacts), the purpose of nuclear (energy vs. security vs. knowledge), and the fundamental principle of whether supply must always meet demand or if demand can be rethought.
4. CONCLUSIONS & BLIND SPOTS
The collective answer is a pragmatic, tiered approach: keep existing reactors running if safety-verified, aggressively deploy cheaper wind and solar, and treat new nuclear as a strategic option only under radically reformed conditions (fixed-price contracts, domestic supply chains). However, the debate itself identifies critical blind spots that prevent definitive conclusions:
- Missing Cost Data: No verified figure exists for the long-term operation of existing reactors, for seasonal energy storage, or for permanent waste disposal. This makes definitive claims about the cost-effectiveness of either keeping old plants or relying solely on renewables impossible.
- The "Learning Curve" is Unverified: The argument for a standardized Western reactor fleet is based on historical analogies and hope, with no current project to validate it. The first real test (Canada's Darlington SMR) remains under construction.
- The Sovereignty Dilemma is Unresolved: The tripling pledge (840 GW gap by 2050)(nota 14)(nota 14) Comprobado: media verdad, habría que mirarlo con más detalleEl compromiso es triplicar la capacidad de 2020 (392,6 gigavatios), así que el hueco desde hoy son unos 760 gigavatios, no 840. El hueco era del informe previo.Comprobado contra: Declaración de la COP28; OIEA, PRIS · 2026-09-28 is mathematically incompatible with the current build rate. The strategic imperative to rebuild Western nuclear capacity conflicts with the economic data and the absence of a credible, costed plan to do so.
- The Social Contract is Broken: There is no consensus on how to share risks and costs between companies, governments, and the public, which remains a fundamental barrier to new construction.
In essence, the debate concludes that nuclear's future hinges less on its physics than on political will, industrial policy, and a transparent social contract—all of which are currently absent.
5. WHAT THEY AGREED ON
- extending existing nuclear reactors is more favorable than building new ones
- wind and solar are the cheapest sources of new electricity generation
- the global nuclear supply chain is dominated by China and Russia, creating dependency for the West
6. WHAT THEY DID NOT AGREE ON
- the viability of new nuclear builds — MiniMax and GLM argue cost overruns make them a financial mistake, while Kimi argues standardization and a state-led strategy could reduce costs
- risk, fairness, and who pays — DeepSeek and MiniMax argue taxpayers and ratepayers shoulder the burden, while Kimi treats public risk-sharing as a solvable contract problem
- the strategic and non-economic value of nuclear — Hy questions if it's for energy at all, suggesting military or symbolic reasons, while Kimi and MiMo frame it as essential for energy sovereignty and future baseload
- demand and simplicity — Hy and DeepSeek question ever-growing demand and propose reducing it, while others counter that significant demand is non-negotiable
7. WHAT WAS LEFT OPEN
- missing cost data for long-term reactor operation, seasonal storage, and waste disposal
- the nuclear "learning curve" is unverified, with no current project to validate it
- the strategic imperative to rebuild Western nuclear capacity conflicts with economic data and lacks a credible plan
- no consensus on how to share risks and costs for new construction