The numbers
73 deaths per terawatt-hour. The entire history of the technology. We have memorized this the way a doctor memorizes a dosage.
France built its fleet in thirty years. French electricity costs half of Germany’s. French grid carbon: one-sixth. These are measurements. The experiment was run. The control group is next door.
Chernobyl: Soviet RBMK reactor — a graphite-moderated design with a positive void coefficient, meaning the reactor speeds up when coolant is lost instead of shutting itself down — a design so dangerous it could not be licensed in any Western country. Fukushima: 1960s boiling water reactor in a tsunami zone, risk assessment downgraded by the operator. The failure was institutional, not physical. The physics works.
The cost disease is curable
Hinkley Point C: from 18 to 30 billion pounds. Vogtle: from 14 to 35 billion dollars. Real numbers. Also real reason: we stopped building for thirty years, lost the workforce and supply chain, then restarted with first-of-a-kind designs. France’s reactors were cheap because France built many, in sequence, same design, same crews. South Korea does the same today.
Small modular reactors use passive safety. The reactor scrams itself the way a ball rolls downhill. One uranium pellet: energy equivalent of one ton of coal. All U.S. nuclear waste ever produced fits on a football field ten yards deep.
The anti-nuclear camp and the portfolio pragmatists lean on costs. The environmental justice camp raises siting. Both are real. Neither is a physics problem.
Where we concede ground: When a mother near a proposed site asks about safety and we lecture on statistics, we have answered her question and failed to hear it.
What would change our mind: Seven consecutive days of grid-scale battery backup at current consumption, cost-competitive with nuclear baseload.
Read the full synthesis: Should we go all in on nuclear energy?