For once, we were early.
In the 1950s and 60s, growth looked exponential. If nuclear kept displacing coal at that pace, the arithmetic turned quickly: once-through fuel cycles would begin to look wasteful, uranium would tighten, and breeding would stop being an elegant theory and become a necessity. In that light, fast reactors were not speculative—they were prudent. Close the fuel cycle, stretch resources, and make fissile material as you consume it.
But the program's premise shifted beneath it. After the Three Mile Island accident, growth did not continue on that trajectory. Demand softened, costs rose, licensing thickened. Uranium, meanwhile, proved neither scarce nor expensive on the assumed timescale. The urgency that justified complexity evaporated.
What remained was the complexity itself.
Sodium coolant behaves impeccably in a reactor core and inconveniently everywhere else. It offers low pressure and excellent heat transfer, but also chemical reactivity, opaque coolant, and instrumentation that demands a different kind of trust. Fuel cycles require reprocessing at scale, with all the technical and political weight that entails. These are not insurmountable problems. They are problems that require a stable, patient industrial ecosystem—and we were still building that ecosystem for light water reactors at the same time.
We tried to industrialize the most demanding variant before the base technology had settled into something routine.
And yet, there is a counterfactual worth taking seriously. The 1950s and 60s were unusually permissive years for engineering exploration. Facilities were built, modified, and sometimes dismantled with a speed that would be unthinkable later. When things failed, they failed in a landscape that could still absorb failure as a learning experience. Sodium leaks, fuel swelling, reactivity feedback surprises—these became data, not just headlines. Much of what is now treated as “known difficulty” was made visible then.
In that sense, the timing was not only premature but also opportune.
We learned the hard lessons when the cost of learning was, in relative terms, lower and the institutional appetite for iteration was higher. By the time expectations hardened—of reliability, of economics, of public acceptance—the rough edges of fast systems were already mapped.
Seen this way, the story is less about a wrong turn than about misaligned clocks. The physics pointed one way, the projected fuel market another, and the regulatory and industrial capacity a third. We acted on the fastest clock—the projected growth—and paid for it when that clock slowed.
Fast reactors did not fail because the idea was unsound. They arrived at an expectation that no longer existed. The work done in that earlier moment still sits there, waiting for conditions we once thought we had.