The Nuclear Counterfactual

The most expensive mistake in energy history

Between 1965 and 1975 the United States was building nuclear reactors faster than any country before or since. The blueprint was the one France would later follow to a nearly carbon-free grid: standardized designs, rapid replication, mass electrification. Then the country walked away. Plants were cancelled, supply chains withered, permitting timelines stretched from years into decades, and nuclear went from the central pillar of American energy ambition to an object of suspicion. Nothing about the physics changed. What changed was the story people told about the physics.

I want to do two things with that turn. First, diagnose it — name the error precisely, because it is a specific and instructive kind of error, not just a bad bet. Second, price it. The diagnosis without a number is a complaint; the number without a diagnosis is a provocation. Together they make a claim you can argue with, which is the only kind worth making.

The Shape of the Mistake

The anti-nuclear turn was fear-driven, and the fear was real but miscalibrated. Three Mile Island released no measurable harm to the public and killed no one; it terrified a nation anyway. The word “nuclear” carried freight the reactors did not — mushroom clouds, invisible poison, waste that lasts ten thousand years. A movement that called itself environmentalism organized around opposing the technology, and that opposition reshaped policy, investment, and public sentiment for half a century.

Germany ran the cleanest natural experiment. After Fukushima — an accident whose radiation killed essentially no one, while the panicked evacuation killed many — Germany accelerated the shutdown of its nuclear fleet. The gap did not fill with wind and sun alone. It filled substantially with coal and natural gas. A country that professed to treat carbon emissions as an emergency shut down its largest source of zero-carbon electricity and burned lignite to replace it. Emissions that should have fallen plateaued or rose. This was not a fringe outcome; it was the predictable result of the policy, and it played out in softer forms across the developed world.

That points at the deepest fault, the one that makes this more than a technical misjudgment. It is a credibility incoherence. The same movement insisted, correctly, that climate change is a civilizational threat demanding urgent decarbonization — and then fought the most scalable, most reliable, lowest-carbon source of firm power available. Those two positions cannot both be held by a person reasoning coherently. If climate change is an existential emergency, you do not spend decades obstructing the one proven technology that delivers gigawatts of always-on power at near-zero emissions. The stance is self-refuting: it disproves, by its own conduct, the urgency it proclaims. Either the emergency is real, in which case nuclear is indispensable, or nuclear can be sacrificed to fear, in which case the emergency is being treated as negotiable. You do not get both.

None of this requires imagining a utopian all-nuclear grid or flawless execution. The counterfactual is modest: suppose the environmental movement had campaigned for clean air, climate stability, and energy independence by championing nuclear power instead of fighting it. Political support in place of political obstruction. Nothing more heroic than that. Then ask what the obstruction cost.

Five Ledgers

Here is where this chapter earns its place in a book about markets, money, and prosperity: a mistake this size can be measured, and measuring it forces the argument out of the realm of vibes. I will build the estimate in five ledgers, keep every assumption on the surface, and tell you exactly which numbers to attack. The Bayes-in-the-wild standard applies here as much as it does to virus origins: a number you can argue with beats a conviction you cannot. If you dislike my total, the honest move is not to wave it away but to change an input and rerun the arithmetic.

The physical baseline for all five: across five decades, U.S. fossil generation ran to roughly 125,000 TWh. A nuclear-first trajectory — France’s, not fantasy’s — would have displaced on the order of 30 percent of it, call it 37,500 TWh generated by fission instead of combustion. Every ledger below flows from that single substitution.

Mortality. Burning coal and gas is not an abstraction; it is particulate matter in lungs, strokes, heart attacks, premature births, chronic disease. The epidemiology of fossil air pollution is mature and grim. Displacing 37,500 TWh of fossil generation with nuclear would, conservatively, have avoided about 1.5 million premature deaths in the United States over the period. To turn lives into dollars — a move that discomforts people, but one that regulators make every day to compare policies at all — I use the value of a statistical life at $10 million, the figure U.S. agencies actually use in rulemaking. That yields:

\[1.5 \text{ million lives} \times \$10\text{M} \approx \$15 \text{ trillion}\]

This is the largest entry, and the one I hold most firmly, because both of its inputs come straight from the mainstream literature that the anti-nuclear movement claims to respect.

Carbon. Coal and gas emit roughly 0.7 tons of CO₂ per MWh; nuclear is near zero. Displace a third of fossil generation for fifty years and you avoid about 26 gigatons of CO₂. To price it, the social cost of carbon — the present value of the future damage each ton does — runs anywhere from $50 to $200 per ton depending on the discount rate and the damage function you accept. That range is itself the honest output; I will not pretend to a point estimate the science does not support:

\[26 \text{ Gt} \times \$50\text{–}200/\text{ton} \approx \$1.3\text{–}5.2 \text{ trillion}\]

Midpoint, about $2.6 trillion — and this captures only the monetized slice, nothing for ecosystem loss, ocean acidification, or climate-driven instability.

Electricity cost. A nuclear-heavy grid front-loads capital but delivers low, stable marginal costs and far less exposure to gas price shocks. On U.S. annual electricity spending of roughly $450 billion, a sustained saving of even 10 percent compounds:

\[\$450\text{B} \times 10\% \times 50 \text{ years} \approx \$2.25 \text{ trillion}\]

Geopolitics. Energy scarcity distorts foreign policy, inflates defense budgets, and chains a country to volatile petrostates. A more electrified, less gas-and-oil-dependent United States would have spent less blood and treasure stabilizing markets it did not need to depend on. This ledger is the softest — the causal chains are long — so I keep it deliberately conservative: $1–3 trillion, midpoint $2 trillion, in avoided strategic and military cost.

The lost industry. Had the U.S. sustained continuous buildout, it would today own the dominant share of reactor exports, the fuel-cycle supply chain, and the engineering ecosystem for next-generation designs — molten-salt, breeder, modular reactors that answer the old safety and waste objections. Instead Russia, China, and South Korea became the world’s reactor builders. The foregone export industry, IP, and manufacturing base: $2–4 trillion, midpoint $3 trillion.

The Total, and How to Move It

Add the conservative midpoints:

Twenty-five trillion dollars is the conservative figure. A realistic range — pushing each input toward the upper end of what the literature plausibly supports — runs closer to $35–50 trillion. Either way, this is among the most expensive ideological mistakes in modern history, and the mortality ledger alone, the 1.5 million lives, is the entry that should keep anyone honest awake.

Now the part that makes the number trustworthy rather than merely large: it is built to be attacked, and I will show you the handles. Halve the VSL to $5 million — some economists prefer it — and the mortality ledger drops from $15 trillion to $7.5 trillion, taking the total under $18 trillion. Take the low end of the social cost of carbon and the carbon ledger nearly halves. Think the displacement fraction should be 20 percent rather than 30, because France’s grid is not America’s and execution is never clean? Scale every physical quantity down by a third and the whole estimate shrinks with it. Doubt the geopolitics ledger entirely? Strike it; the total barely notices, because it was the smallest and softest to begin with. Push the other way — a higher VSL, a higher carbon price, a larger displacement — and you climb toward the $50 trillion end.

That elasticity is not a weakness of the estimate. It is the estimate, stated honestly. Changing the inputs changes the number, and I am inviting you to change them. What I will insist on is that whatever inputs you choose, you apply them consistently, and that the mortality figure — the one anchored in ordinary epidemiology and the ordinary regulatory value of a life — does not vanish under any defensible choice. You can argue the total down to fifteen trillion or up to fifty. You cannot argue it to zero without denying that burning coal kills people, and nobody serious does.

The Doomer Symmetry

Step back and the shape of the error is familiar. It is the shape Lessons From Peak Oil diagnoses: a catastrophist reasoning failure — a vivid worst case, statically imagined, its probability inflated by dread, its costs of avoidance ignored. Peak-oil doomers fixated on running out of oil and never priced in substitution and adaptation. Anti-nuclear campaigners fixated on meltdowns and never priced in the coal smoke they were choosing instead. Both are catastrophism. The peak-oil version predicted a disaster that adaptation prevented; the nuclear version predicted a disaster so rare it barely registers in the fatality statistics, and in fighting it caused a slower, larger disaster measured in the fossil pollution ledger above.

This is the symmetry worth holding onto: the anti-nuclear movement was itself a doomer movement. It failed the same field test peak oil failed — it took a low-probability tail risk, treated it as the dominant consideration, and discounted the systemic cost of the alternative it was steering toward. The five questions that catch peak-oil catastrophism catch this one too. What is the probability, really, weighed against the base rate? What does the feared outcome actually cost when it occurs, versus the outcome you get by avoiding it? What adapts? What does the price system already handle? Run nuclear fear through that filter and it fails, exactly as the oil scare failed, but with a body count.

And that is what makes this the darkest entry in a volume otherwise about great progress. The story of the modern world is child mortality falling from one in three to under one in a hundred, life expectancy climbing decade on decade, hot showers no pharaoh could command. That progress is real and it is the default, but it is not automatic. It runs on cheap, abundant, reliable energy, and it can be slowed by a society whose conceptual filters amplify fear instead of coherence. The nuclear counterfactual is progress foregone — a half-century where the curve could have risen faster, the air been cleaner, the carbon lower, and did not, because a movement mistook the ticket for the coat and its own fear for a fact.

History rarely hands us a counterfactual this cleanly evaluable. The lesson is not that nuclear power is perfect; no technology is, and the honest ledger prices its real costs, not a fantasy. The lesson is that evaluating a technology demands reasoning rather than fear — and that when a society’s story about a technology diverges from the arithmetic, it is the arithmetic that eventually sends the bill.