It’s not some design issue that you can fix with small modular reactors or whatever. Every time anything significant goes wrong people kept adding systems to prevent it happening again. Sometimes that’s something physical like a wall or redundant system, and other times it’s longer procedures and more paperwork. This has been great for avoiding another SL-1, 3 mile island etc, but mitigating every possible risks isn’t cheap. https://en.wikipedia.org/wiki/SL-1
For example there was once an expensive issue where foreign material fell into the spent fuel pool where fixing it significantly extended an outage at a single reactor so now everyone working on anything at any reactor takes significant steps to mitigate that risk. Great things working as intended except now something that might have cost 100k now costs significantly more because of that mitigation effort.
Extend that to any significant issue across hundreds of reactors and 50+ years and suddenly triple checking to avoid every fuck up means doing anything is really expensive. Worse very little of it is wasted effort, there’s just far more ways to fuck up in practice than in theory.
PS: I do believe it’s possible to have safe and cheap nuclear power, it’s just going to look very different than what we’re doing today.