The Fire We Forgot

Anima Mundi · On Fire and Civilization

here is a man most of you have never heard of. His name was Alvin Weinberg. He was, by most accounts, one of the great scientific minds of the twentieth century — the man who co-invented the light water reactor that powers roughly 70% of all nuclear plants operating on earth today. He spent three decades running Oak Ridge National Laboratory in Tennessee. He was brilliant, careful, and unusually honest about what he was doing and why.

In 1972, he was fired.

The official reason was insubordination. The real reason was that Weinberg kept insisting, publicly and persistently, that there was a better reactor design — one that was safer, cleaner, more efficient, and that produced waste lasting hundreds of years instead of hundreds of thousands. He had built a working prototype. It ran for four years without incident. He wanted to develop it. His superiors wanted him to stop talking about it.

The better design was the molten salt reactor. The technology that killed it was not physics, or engineering, or economics. It was institutional inertia rooted in the weapons complex — the simple fact that the uranium fuel cycle was compatible with making bombs, and the thorium fuel cycle mostly wasn’t.

Weinberg died in 2006. His reactor sat in the archives. The world kept building pressurized water reactors and accumulating spent fuel and debating whether nuclear was safe, using a reactor design optimized for submarines and bomb-grade material production.

I’ve spent the last several weeks going deep on this — deeper than I expected to. I want to take you through what I found, because I think it’s one of the most important stories nobody is seriously telling. It touches everything we’ve been discussing in this newsletter: resource wars, civilizational entropy debt, the gap between what is known and what is discussed, and the question of who gets to build the infrastructure that shapes the next century.

It also has a direct geopolitical consequence unfolding right now, largely invisible to Western publics.

Let’s start with the physics. I’ll keep it brief, and I promise it matters.

I. What the Reactor Actually Is

In every nuclear power plant you’ve ever seen a photograph of, the fuel is solid. Uranium pellets are pressed into ceramic, sealed in metal rods, bundled into assemblies, submerged in water. The water cools them. The heat produces steam. The steam turns turbines. The spent fuel — still intensely radioactive — gets pulled out and put in pools and eventually into dry casks and then, theoretically, into underground repositories that don’t exist yet.

The molten salt reactor works differently. In a molten salt reactor, the fuel is dissolved directly into a liquid salt mixture that flows through the core. The salt is both the fuel carrier and the coolant. There are no fuel rods. There is no cladding. There is no solid material that can melt.

The salt mixture typically operates at around 700 degrees Celsius at near-atmospheric pressure. For comparison, a pressurized water reactor operates at roughly 155 atmospheres of pressure — the engineering equivalent of storing everything in a bomb casing. The molten salt reactor is, in this sense, fundamentally more relaxed. If something goes wrong, you don’t have pressure fighting you.

There are no fuel rods. There is no cladding. There is no solid material that can melt. The engineering equivalent of Fukushima is physically impossible by design.

The thorium fuel cycle adds another layer. Thorium-232 itself doesn’t split — it’s what physicists call fertile rather than fissile. But when it absorbs a neutron, it slowly transforms into uranium-233, which does split, and does so with remarkable efficiency. The reactor breeds its own fuel as it operates. You start with a small amount of fissile material to get things going; the thorium feeds the reaction; you extract the ash and keep running. The waste products have half-lives measured in centuries rather than geological epochs.

The passive safety mechanism is the most beautiful piece of engineering in the whole system. At the bottom of the reactor vessel sits a freeze plug — a section of salt kept frozen solid by active cooling. If power is lost, the cooling stops, the plug melts, and the fuel drains by gravity into subcritical storage tanks below. The reactor shuts itself down. No pumps required. No operator action required. No backup power required. The default state of the system, without any energy input, is safe.

Chernobyl required a positive feedback loop between void formation and reactivity — physics impossible in this design. Fukushima required solid fuel that continued generating decay heat after shutdown, with no coolant to remove it — impossible here because there is no solid fuel. Three Mile Island required a partial core melt — impossible here because there is no core in the conventional sense.

These aren’t safer versions of the same accidents. They are a different category of machine.

II. Why Uranium Won

The story of why the thorium molten salt reactor was never developed at scale is one of those historical hinge moments that looks obvious in retrospect and was almost invisible as it happened.

In the early 1950s, the United States was simultaneously building a weapons complex and trying to create a civilian nuclear industry. The weapons complex needed enriched uranium and plutonium. The civilian industry, almost by accident, got built on the same infrastructure. Light water reactors — the pressurized water design Weinberg helped develop for submarine propulsion — were already understood, already built, already supported by a nascent industrial base.

Thorium doesn’t produce plutonium. The uranium-233 produced in the thorium cycle is always contaminated with uranium-232, whose decay products emit such intense gamma radiation that anyone trying to use it for a weapon would be irradiated through the process. This makes it proliferation-resistant — a genuine arms control advantage — but it also made it strategically unattractive in the 1950s and 60s, when the weapons complex and the civilian industry were intimately intertwined.

The Atomic Energy Commission, which ran both, had little institutional interest in developing a fuel cycle that didn’t feed into its primary mission. Weinberg’s molten salt program at Oak Ridge was a parallel track that kept producing good results and kept being de-prioritized. When he became too vocal about safety concerns with light water reactors — publicly questioning whether the existing reactor fleet was being built with adequate margin — he was removed from his position and his program was terminated.

The better technology was suppressed not by market forces but by institutional path dependency rooted in weapons infrastructure. A line of history turning on an administrative decision most people have never heard of.

The world that resulted from that decision has been living with the consequences ever since. Roughly 440 nuclear reactors operating globally, almost all light water designs, producing roughly 2,600 tonnes of spent nuclear fuel per year. In the United States alone, approximately 90,000 tonnes of spent fuel sit in temporary storage. The Yucca Mountain repository — intended to be the permanent solution — has been politically blocked for decades. Finland is building the world’s first operational permanent repository, designed for 100,000 years of isolation. A hundred thousand years. Modern humans have existed for roughly 300,000. We are making commitments that extend across a third of our species’ entire history, and calling the current situation temporary storage.

This is what I mean by entropy debt. The disorder we externalize doesn’t disappear. It accumulates, and the interest compounds.

III. What China Is Building in the Desert

In 2023, in the Gobi Desert in Gansu Province, China achieved first criticality in the TMSR-LF1 — the world’s first operational thorium molten salt reactor since Oak Ridge shut down its experimental reactor in 1969. The facility was built by the Chinese Academy of Sciences. The project has been running since 2011 with roughly 700 dedicated researchers. The 2 MW experimental unit currently operating is a materials testing platform, not a power plant. The real target is a 373 MW commercial unit planned for the early 2030s.

The choice of the Gobi Desert is not incidental. The site was selected partly for remoteness, partly for geological stability, and partly because the desert has no water. This matters because the TMSR is designed to use supercritical carbon dioxide rather than steam for power generation — a more efficient thermodynamic cycle that doesn’t require a river or a cooling tower. A reactor you can build in a desert, at near-atmospheric pressure, that shuts itself down passively, that runs on a fuel you mine domestically, producing waste that needs management for centuries rather than millennia.

China’s largest rare earth mine, at Bayan Obo in Inner Mongolia, sits on enormous thorium deposits currently classified as waste byproduct. As the TMSR program scales, this waste stream becomes strategic fuel inventory. China is the world’s dominant processor of rare earth elements. It is developing domestic lithium-7 enrichment capacity, breaking a near-monopoly previously held by Russia. It produces roughly 45% of global molybdenum, a key constituent of the Hastelloy-N alloy that the reactor’s structural components require.

The vertical integration here is not accidental. This is a decades-long program designed to control the full materials stack of a technology that could become the dominant energy infrastructure of the late twenty-first century.

If molten salt reactors become as significant as lithium-ion batteries became to transportation, the country that controls the materials stack controls energy geopolitics. China appears to be playing a fifty-year game. Most Western governments are playing a four-year electoral cycle game.

The Belt and Road angle is worth sitting with. China has spent twenty years building infrastructure dependency in energy-poor nations across Africa, Central Asia, and Southeast Asia. A mature TMSR technology, exported to countries with thorium reserves and limited industrial capacity, would create the same dynamic that American reactor exports created in the 1960s and 70s — technology dependency, fuel cycle dependency, service dependency. The Atoms for Peace playbook, one generation ahead.

India sees this clearly. India has the world’s largest thorium reserves — roughly 846,000 tonnes, largely in the monazite beach sands of Kerala and Odisha. India imports roughly a third of its energy needs and runs a persistent current account deficit partly driven by fossil fuel costs. The Indian Department of Atomic Energy has been developing a three-stage nuclear program explicitly designed around thorium since Homi Bhabha in the 1950s — a seventy-year project playing out on exactly the civilizational timescale that Western democracies find almost impossible to sustain.

Brazil sits on the world’s second-largest thorium deposits. Australia has substantial reserves alongside its uranium. Egypt has significant thorium in its western desert. The thorium map is, almost precisely, the inverse of the oil map. The nations that have been structurally disadvantaged by fossil fuel dependency are, in many cases, the same nations sitting on the fuel source of the next energy cycle.

This matters. It matters enormously.

IV. The Silence Ratio

I want to talk about why this story isn’t being told, because the absence is as significant as the story itself.

The public discourse on nuclear energy is almost entirely captured by two frames, both of which are roughly forty years out of date. The anti-nuclear frame — Chernobyl, Fukushima, waste, proliferation — extrapolates from accidents in solid-fuel light water reactors onto a technology that has a completely different failure physics. The pro-nuclear frame defends existing light water reactor fleets because that’s where the industrial base and regulatory frameworks exist, not because the technology is optimal. Both frames are arguing about a previous generation of design.

The actual frontier conversation — thorium fuel cycles, online reprocessing chemistry, passive safety through physics rather than engineering controls, the geopolitical implications of a shift in which nations hold strategic fuel reserves — is happening almost entirely in Chinese state research publications, small specialist conferences, and a handful of startup investor memoranda. It is read by hundreds of people globally when it should be read by millions.

This is what I called the Silence Ratio in an earlier essay: the gap between what is known and what is publicly discussed, and the consequences of that gap for collective decision-making. The Silence Ratio on thorium reactors is one of the largest I’ve encountered. The knowledge is substantial, the implications are significant, and the public conversation is essentially nonexistent outside specialist circles.

Part of this is the nature of nuclear discourse — the topic has been so thoroughly colonized by fear and by advocacy that nuanced technical conversation is nearly impossible in public. Part of it is that the technology is genuinely complex, and complexity is the enemy of media attention. Part of it is that the Western programs developing molten salt technology are small, privately funded, and operating against regulatory frameworks designed for different reactor types — they don’t have the institutional voice that state-funded programs have.

And part of it, I think, is that the story is uncomfortable for established interests on multiple sides. For the existing nuclear industry, it implies that the infrastructure they’ve spent decades building is not the best available design. For the fossil fuel industry, it implies the existence of a clean, abundant, base-load energy source that could genuinely replace them. For Western governments, it implies a strategic failure — a decades-long inability to develop and deploy a technology that was invented at a national laboratory, with public funding, and then abandoned.

V. Entropy Debt and the 300-Year Question

I want to come back to the waste question, because I think it reveals something important about how we reason about civilizational time.

The thorium cycle produces waste with half-lives measured in centuries rather than geological epochs. The longest-lived significant byproducts need roughly 300 years of isolation before reaching natural background radiation levels. This is often described as a dramatic improvement over the uranium cycle, which produces transuranic elements — plutonium, americium, neptunium — requiring isolation for tens of thousands of years.

And it is a dramatic improvement. But 300 years is not nothing. Three hundred years ago, Bach was composing the Brandenburg Concertos. The United States did not exist. The Industrial Revolution had not begun. The world that will manage the last of the thorium waste is as unimaginable to us as our world would have been to them.

What the thorium cycle actually offers is not elimination of entropy debt but a restructuring of it. Instead of a geological-scale storage problem — the almost incomprehensible demand we are placing on future civilizations to manage material that will remain dangerous longer than modern humans have existed — we have a civilizational-scale management problem. Shorter timescales, but requiring continuous technical competence across centuries rather than passive containment across millennia.

We trade a geological-scale storage problem for a civilizational-scale management problem. Whether that’s better depends on whether you trust future civilizations to maintain operational competence across centuries. It is, at minimum, a more honest commitment.

There is something almost philosophical in this trade. The uranium fuel cycle externalizes its consequences maximally — push the problem as far forward as possible, make it as passive as possible, hope that the geological formation holds and that nobody drills into it in the year 52,000 AD. The thorium cycle demands ongoing attention. It requires future civilizations to remain technically competent, to maintain operational infrastructure, to actively manage the chemistry of the waste stream.

In the framework I’ve been developing around Heliogenesis — the idea that sustainable civilization means open energy flows and closed material loops — the thorium cycle is closer to the right direction. It doesn’t achieve closed loops, but it shortens them dramatically. The entropy debt is real but it’s denominated in human generations rather than geological epochs.

This is a meaningful distinction. It’s also an honest one. We are not making promises to people 100,000 years from now that we have no ability to keep. We are making demands of people 300 years from now that are at least within the order of magnitude of human planning horizons. That’s different. It’s still a debt. But it’s a debt that future generations can conceivably understand and manage.

VI. What Weinberg Understood

I keep coming back to Weinberg. Not as a symbol of suppressed genius — that framing is too simple, and the history is more complicated than heroes and villains. But as an example of someone who understood something important about the relationship between technology and civilization, and who paid a professional price for saying it out loud.

Weinberg argued, in the later years of his career, that certain technologies carry inherent civilizational requirements. Nuclear energy, he said, requires a priesthood — a continuous, multigenerational technical community capable of managing the infrastructure and the waste. He didn’t mean this as a criticism. He meant it as a recognition that some technologies bind you to a certain kind of social organization. If you choose nuclear, you are committing to maintaining technical institutions across the timescales the waste demands.

He was right. And the question he was implicitly raising — which we still haven’t answered — is whether that commitment is one we can actually keep.

The thorium molten salt reactor doesn’t eliminate this commitment. It shortens it. It makes the failure modes more recoverable. It locates the strategic fuel reserves in different nations than the current uranium cycle does. It produces a different kind of waste on a different timescale. These are real improvements. They’re not magic.

What strikes me most, reading the history and the physics and the geopolitics of this technology, is how much of it was already known. Weinberg knew. Oak Ridge knew. The physics was demonstrated in 1965. The reasons it wasn’t developed were not technical. They were political, institutional, strategic — the ordinary human machinery of path dependency and institutional self-interest and short-term thinking dressed up as technical judgment.

The fire was already lit. We just walked away from it.

VII. What Comes Next

The TMSR-LF1 is running right now in Gansu Province. It is collecting materials data, measuring corrosion rates, characterizing tritium permeation through metal walls at 700 degrees Celsius, testing in-core instrumentation. Every hour of operation feeds the engineering database that the commercial-scale LF2 will be designed from. The timeline for LF2 is the early 2030s. If it works — and the physics and the four-year Oak Ridge prototype suggest it should — the question becomes how quickly a technology can be manufactured and deployed at scale.

Western programs exist. Terrestrial Energy in Canada, Moltex in the UK and Canada, Flibe Energy in the United States, Copenhagen Atomics in Denmark. They are real, technically credible, staffed by serious people. They face structural disadvantages: regulatory frameworks built for different reactor types, capital structures that don’t fit 15-year development timelines, materials qualification processes that take decades, supply chains with critical single points of failure.

The Li-7 enrichment situation alone is worth lingering on. Lithium-7, isotopically enriched to remove the neutron-absorbing lithium-6 fraction, is essential for FLiBe salt. The United States shut down its enrichment capability in 1963 after severe environmental contamination. Western enrichment capacity is effectively zero. Russia has supplied the global market. After 2022, that supply chain became politically untenable. The Department of Energy has commissioned studies. New enrichment processes are being developed. The optimistic timeline for Western Li-7 independence is ten to fifteen years.

China produces its own. China has been producing its own since its TMSR program began.

This is the pattern we’ve seen before. Solar panels were invented in the United States. The manufacturing capacity ended up in China. Lithium-ion batteries were developed largely in Japanese and Korean labs. The dominant manufacturing position ended up in China. Electric vehicles were conceptualized and prototyped across multiple Western countries. China is the world’s largest EV market and the dominant manufacturer.

The mechanism is always the same: the West invents, debates, regulates, delays. State-directed industrial policy builds manufacturing scale while the debate continues. By the time Western consensus forms around deployment, the supply chains and manufacturing expertise are established elsewhere.

I’m not making a simple argument that China’s approach is better or that Western democratic deliberation is a weakness. I think genuine public deliberation about large-scale technologies matters. I think the regulatory caution around nuclear energy, given the history, is understandable. I think the question of who controls the infrastructure that shapes civilizational energy flows deserves exactly the kind of slow, careful deliberation that democratic institutions are supposed to provide.

But I do think we need to be honest about the gap between the deliberation we’re doing and the deliberation the moment requires. We are discussing a previous generation of nuclear technology while a different technology is being built in the Gobi Desert. We are debating the legacy of Chernobyl while the relevant physics has moved on. We are making supply chain decisions on four-year timescales while the relevant competition is playing out on fifty-year timescales.

The fire was already lit in 1965. We walked away from it for reasons that had nothing to do with whether it worked. The question now is whether we can find our way back before someone else has already built the hearth.

Coda: Weinberg’s Last Interview

In 2002, four years before he died, Alvin Weinberg gave an interview in which he was asked whether he was bitter about what had happened to the molten salt reactor program. He said he wasn’t. He said the history of technology is full of better ideas that didn’t get developed, and worse ideas that did, and that this is the ordinary condition of technical progress in a world run by humans with competing interests and limited foresight.

What he was, he said, was hopeful. Because the physics doesn’t change. The thermodynamics don’t change. The half-lives don’t change. The materials science had advanced. The chemistry was better understood. The arguments for the technology were as good as they had ever been, maybe better. All it required was someone willing to build it.

Someone is building it.

Whether it’s the right someone, on terms that distribute the benefits equitably, in a way that doesn’t simply replace one form of energy dependence with another — these are the questions that deserve the serious public deliberation they’re not currently receiving.

Weinberg was asked, at the end of that interview, what he would say to a young scientist considering going into nuclear energy. He said: go. The problems are real, the stakes are civilizational, and the physics is beautiful. There is no better combination.

He was right about that too.

With warmth and fire,

Malte

More on my personal website

https://www.joshwagenbach.com/blog/what-eur-001-kwh-energy-does-to-a-continent

A Note on Sources and Depth

The physics and engineering described in this essay draw on published literature from the Chinese Academy of Sciences TMSR Center, the Oak Ridge National Laboratory archives (particularly the Molten Salt Reactor Experiment reports from 1965–1969), the work of Alvin Weinberg collected in The First Nuclear Era (1994), and technical publications from the Thorium Energy Conference proceedings. The geopolitical analysis of materials supply chains draws on IAEA reports on nuclear fuel supply, US Department of Energy strategic materials assessments, and published analyses of Chinese rare earth and lithium processing capacity. The Weinberg firing and its context are documented in Jack Radcliff’s institutional history of Oak Ridge and in Weinberg’s own memoir.