It's true that light water reactors are not a good way to make plutonium for weapons. I disagree that designs running on natural uranium fuel (moderated by graphite or heavy water) are more efficient. Even countries that already have nuclear weapons and are beyond non-proliferation fears, like China, Russia, and the UK, don't build new power reactors moderated with heavy water or graphite.
Reactors using unenriched uranium can be more fuel-efficient in terms of electrical output per ton of natural uranium consumed, but they're not more cost-efficient. High project cost is the number one problem that impedes nuclear construction while fuel cost is almost negligible. Keep in mind that Canada's newest CANDU, Darlington 4, started construction in 1986; if you're comparing its costs to a light water reactor then you should also use mid-1980s costs for those and not e.g. the high costs for more recent EPR or AP1000 reactors. Building a new CANDU in Canada today would also cost more than it did in the 1980s.
My point was more specifically about push to use only light water moderated reactors, and yes for various reasons those were generally done even among nuclear armed states (because you want to keep certain things close and because an LWR can be sold elsewhere).
However I'd argue that in terms of total lifespan costs the increased waste production of LWRs is considerable portion of perceived costs - essentially the long-term storage fears vs burn down in fast reactors etc. Plus the focus on LWRs ensured that research in more capable designs was stiffled.
Do not take this "rah rah nuclear best renewables suck" though. Though I'd be very interested in how costs would go if we taxed/penalized emissions do the max
- essentially, I consider every new fossil fuel plant, including gas backup for renewables, to be a policy failure. My personal dream is overprovisioning with nuclear and renewables and instead of curtailment try to funnel extra power into other, previously uneconomical things (replacement of coal in smelting with actual green hydrogen, stuff like that)
Fast reactors can theoretically reduce the amount of waste that needs to go to long term disposal, but those are even rarer than graphite or heavy water moderated reactors. They're also even more expensive. I believe that there are currently only two fast power reactors in operation, both Russian (BN-600 and BN-800).
I agree with you that it would have been (still would be) better to have emissions taxes penalizing pollution from fossil fuels instead of technology-specific incentives and subsidies for the various kinds of fossil alternatives. I live near a nuclear power plant and I have no problem with its track record of safety or emissions. I do think nuclear power has a major cost and scheduling problem for new projects and I'm not sure how that can be overcome. SMR projects like this one that Utah just canceled were the latest great hope for taming those issues.
Reactors using unenriched uranium can be more fuel-efficient in terms of electrical output per ton of natural uranium consumed, but they're not more cost-efficient. High project cost is the number one problem that impedes nuclear construction while fuel cost is almost negligible. Keep in mind that Canada's newest CANDU, Darlington 4, started construction in 1986; if you're comparing its costs to a light water reactor then you should also use mid-1980s costs for those and not e.g. the high costs for more recent EPR or AP1000 reactors. Building a new CANDU in Canada today would also cost more than it did in the 1980s.