Nuclear fission power

technology

The controlled fission of heavy nuclei to raise steam, and the most important counter-example to the learning curve in the energy system. Fission supplies roughly nine per cent of world electricity from a fleet of some four hundred and forty reactors, down from a peak share of around seventeen or eighteen per cent in the mid-1990s — a decline driven by flat construction alongside rising total generation rather than by closures alone. Its cost history runs the wrong way. Lovering, Yip and Nordhaus assembled overnight construction costs for most of the world fleet and found that costs rose over time in the United States and France even as cumulative capacity grew, the opposite of what a learning curve predicts; other national programmes show flatter or mixed trends. The structural explanation is the one this Journey turns on: a reactor is constructed on a site by a coalition assembled for that site, not manufactured repeatedly on a line, so the units of experience are few, separated by decades, and not obviously transferable. Recent Western projects are consistent with that reading — the two Vogtle AP1000 units in Georgia entered service in 2023 and 2024 at a total project cost several times the original estimate, and European first-of-a-kind EPR projects ran long by comparable factors. Fission nevertheless supplies something intermittent generation does not: dispatchable, weather-independent output with very high capacity factors, which is why the argument about it is not really an argument about carbon.

A technology is not a place. Drawing it on a map would assert something about the world that no stored fact supports.