The rate and cost of failures directly relate to insurance costs. A 1 in 100,000 chance per year to cause a 500 billion dollar accident represents a ~5 million per year insurance cost to offset that risk before considering the risk premium associated with insurance. And that’s on top of the normal risks for large complexes that have little to do with nuclear just high voltage equipment etc. Unsubsidized insurance costs are something like 0.2c/kWh which is quite significant for these projects.
In the end you see a lot of people talking nonsense around nuclear costs using wildly optimistic numbers, but there hasn’t been a power plant built and operated in the last 20 years that come even close to these numbers. Let alone when you start to compare predictions for decommissioning costs with actual decommissioning costs.