> But breeders themselves are uncompetitive

Hard to say. Hasn't really been commercialized to the point one could make any definite conclusion. There's nothing inherent in a breeder reactor that would make it substantially more expensive.

> and fast breeders are inherently dangerous, since they require much higher concentrations of fissionable material, and could potentially experience fast prompt supercriticality in severe accidents.)

No. In a fast reactor the prompt neutron multiplication time is an order or magnitude faster than in a thermal reactor, but still several orders of magnitude slower than in a weapon. Further, the enrichment is much lower than in a weapon (recent-ish designs are made to work below the 20% HEU limit for obvious reasons), so it's quite unclear if it's even possible to get all the material into a suitable geometry before it would blow itself apart.

Of course it's possible to have a criticality excursion that would destroy the reactor, but those are possible in thermal reactors as well.

> If nuclear is not competitive with cheap uranium and burner reactors, it's not going to be competitive with breeders and reprocessing.

The fuel cycle costs are basically fuel + enrichment + disposal for a once-through thermal cycle, and fuel + reprocessing + disposal for the breeding cycle. Fuel and disposal costs would be lower for the breeding cycle, although with current prices and maturity level of the technology you're correct that the once-through cycle is cheaper. Not written in the stars that it will remain so forevermore, though, and breeder + reprocessing tech is waiting in the drawer for that day, in case there ever will be a need. No hurry.

All the attempts at building breeders have experienced serious problems, and estimates of their cost are higher than for thermal reactors.

> No. In a fast reactor the prompt neutron multiplication time is an order or magnitude faster than in a thermal reactor, but still several orders of magnitude slower than in a weapon.

One could still have prompt supercriticality. So, even if the neutrons are slower than in a weapon, they are still much faster than in a thermal reactor, so the doubling time could be much shorter. There's also the possibility that, because the core contains tons of plutonium, the rearrangement driven by the supercriticality could cause the arrange to become MORE reactive, not less (this may be unlikely, but it's very hard to rule out in any possible accident.) Edward Teller famously warned about this issue back in the 1960s.

For this reason I suspect that nothing but a fast MSR could be licensed in the west.

You will note that the French have now abandoned Astrid, which was to be their next attempt at a prototype fast reactor. It's so distant now that it's not worth keeping the expertise around.

> All the attempts at building breeders have experienced serious problems, and estimates of their cost are higher than for thermal reactors.

Nope. EBR-II back in the day had no big problems over it's lifetime. In Russia there are a couple of relatively big ones producing power, with no big issues.

But crucially, so far all breeder reactors have been one-off things done for R&D purposes. They haven't been optimized for producing low cost power, and there has been no learning curve either since all have been bespoke designs.

Then again, in the west we have failed to build even traditional LWR's on time and budget, as well as other large projects.

> So, even if the neutrons are slower than in a weapon, they are still much faster than in a thermal reactor, so the doubling time could be much shorter.

Yes, but not short enough that a non-trivial fraction of the fuel would fission before the assembly would disintegrate.

> There's also the possibility that, because the core contains tons of plutonium, the rearrangement driven by the supercriticality could cause the arrange to become MORE reactive, not less

Now you're just concern-trolling. If part of the core becomes supercritical due to melting, in the absence of cooling that would vaporize the supercritical part of the fuel, reducing its density and pushing everything else outwards, thus reducing the neutron multiplication. Melted fuel flowing is just a much too slow process to rapidly cause a big bunch of fuel to become prompt supercritical.

Nuclear weapons are very Very VERY carefully engineered so that a substantial fraction of the fuel fissions before the assembly blows apart. Nuclear reactors OTOH, including fast reactors, are very Very VERY carefully engineered so this won't happen even in the most wildly imaginary accident scenario.

> For this reason I suspect that nothing but a fast MSR could be licensed in the west.

Oklo Inc. is in the process of licensing their fast reactor having submitted the application this year. Presumably the believe they have a decent chance of having their license application approved.

There's also the Versatile Test Reactor(?) which the US government is planning to build. Essentially a PRISM reactor (evolved from EBR-II / IFR) modified for materials testing rather than power production.

> You will note that the French have now abandoned Astrid, which was to be their next attempt at a prototype fast reactor.

Yes. I was never a big fan of ASTRID, AFAICS there wasn't anything really innovative there that would have brought any major new insights. Still disappointing though.

No, not concern trolling. Designs where rearrange of the assembly increases k were explored for weapons. They were not pursued because they were large and required lots of fissionable material. But a fast reactor core has orders of magnitude more material than a bomb. The regulators will require proof that this scenario is very unlikely, not just an assumption. And that's going to be difficult to provide.