If a nuclear plant's life is 30+ years, making it disaster prone in "once in 100 years" circumstances is pretty reckless. If a nuclear plant vendor aims to make 1000 plants, they should aim for negligible risk that any of their reactors would crap itself due to weather in 150 years or something like that. You pretty quickly find yourself mitigating risk of once-in-a-million-years events from a single plant's POV in this kind of math.