I guess the answer depends on which country you're talking about. The U.S. government pioneered nuclear power but has done very little with it over the past several decades, and has impeded private efforts. Canada has a regulatory system friendlier to reactor startups than the U.S., and at least one molten salt reactor company (Terrestrial Energy) is making good progress there. China has an aggressive government program developing every type of GenIV reactor, and Terrapower moved there after giving up on the U.S.

This article mainly covers the MIT fusion effort. For years they struggled to keep their government funding, despite their tokamak having the most powerful magnetic field of any in the world. They finally lost that battle, and now private investors are stepping in. It doesn't appear that there was an alternative.

Government can obviously throw more money but it tends to be more conservative about what it funds. And private funding can go pretty far; it was interesting to see that TAE (Tri Alpha Energy, an aneutronic fusion effort) is up to $800 million in funding. They started work in 1999, so long-term fundamental research does seem to be achievable in a startup setting.

Besides, comparing the efficacy or ability of government's in 1940s wartime vs the modern nation states is a terrible idea and is very weakly related. In terms of pure scale, administrative top heaviness, legislative/regulatory oversight has expanded 100x in the past century, the degree of economic intervention, access to talent in peacetime, etc are all vastly different.

Even NASA in the peacetime 1950-60s which was so famously efficient and effective was largely the result of talent quickly absorbed from private industry and academia into one organization. It was a newish organization which held many of the benefits of non-gov/private organizations, as they were not yet fully affected by the heavy bureaucratic and political load which hits every government agency over time.

The type of stuff that scares away the raw talent and creatives and shifts to a system which values people who play politics and shifts power to administrative roles over the producers. https://www.jerrypournelle.com/reports/jerryp/iron.html

If a self-sustaining nuclear fission reaction had not been previously demonstrated, could you build a proof-of-concept in 2018 on a startup budget? I would say yes. It is conceivable that you could build a modern equivalent of Chicago Pile-1 on a startup budget, assuming that you didn't implement any more safety features than the original CP-1. And if nuclear fission had never been demonstrated before this hypothetical startup probably wouldn't have much in the way of safety regulations to consider.

The more difficult question: how would a startup come to experiment with ways to build a self-sustaining fission reaction if the basic concept hadn't been demonstrated before? Something would have to be very different about history. As an improbable hand-wave, let's say that academic scientists had only ever experimented with fast-neutron fission, and didn't realize that natural uranium could sustain a chain reaction if the neutrons were moderated. We're probably deep into alt-history "World War II never happened, also the world has been surprisingly peaceful" fiction with that hand-wave.

Finally, if you really mean current nuclear power technology, e.g. one of the actual Generation III reactor designs currently operating, plus supporting infrastructure -- no, a startup could not invent all of that.

https://en.wikipedia.org/wiki/Generation_III_reactor#Generat...

You could easily blow through 100 million dollars just trying to prototype the enrichment process for the uranium that your reactor needs as fuel. Never mind actually sourcing enough natural uranium in a world where uranium has previously been a niche material like rubidium, building a uranium enrichment plant, fabricating fuel elements, forging a huge pressure vessel, developing the whole field of nuclear engineering...

Actually uranium at various enrichments is cheaper than ever. SWU cost last year was $125.43/SWU https://www.eia.gov/uranium/marketing/

Right, that's true in the actual world. I'm trying to imagine a world where a startup is developing nuclear power for the very first time, which includes developing/building any enrichment technologies their reactor might depend upon.

I don't think that a reactor requiring enriched fuel is a good Very First Reactor design -- and the actual first reactors did not require enriched fuel -- but enrichment came up in the course of answering whether a startup could invent a modern reactor in the absence of an existing nuclear industry.

Possibly a nuclear industry kicked off by startups would use different technology. It may be no coincidence that we used a gigantic price-insensitive entity to develop the early technology, and ended up with expensive reactors.

Molten salt reactors, for example, were known back in the 1950s, and appear to have a number of major cost advantages.

Molten salt reactors might be the second generation of reactors in an alternative history with nuclear-by-startups. They require a higher fissile material concentration than found in natural uranium. To start operating a MSR you need some reactors fueled with natural uranium and moderated by graphite or heavy water in order to breed fissile plutonium from uranium, or uranium 233 from thorium, to get concentrated fissile material for starting molten salt reactors.

Or you could start a MSR with U-235 enriched from natural uranium, but that would require developing complex and expensive enrichment technology before you get your first watt of nuclear power generation.

True, and maybe a CANDU would be best at first if enrichment is a huge barrier. But MSRs don't necessarily need especially high enrichment. The IMSR for example only needs 5% U235, which is the upper end of what light water reactors use. https://en.wikipedia.org/wiki/Integral_Molten_Salt_Reactor