Expensive yes; unburned is a choice (and less important given the energy density); the long term danger is both true and misleading because while it's a long time the actual volume of material is tiny.

For me, the problem with nuclear is the safety profile: although the mean deaths per TWh is really really low, the observed upper bound on damage that can happen when it does go wrong is sufficient to bankrupt a superpower.

Low mean/high variance is exactly the kind of scenario where I expect people to cut corners because they think it will never affect them.