Presumably these computer models are mostly useful for creating new designs (since the old designs were proven by real tests). Would such new designs be convinc

Presumably these computer models are mostly useful for creating new designs (since the old designs were proven by real tests). Would such new designs be convincing enough to the adversary to fulfill its role as a strategic deterrence? When (in XX years?) almost all US nukes are only simulated on computers and not actually tested, the Russians may start wondering if the US aresnal actually works, no? That would be a horrible outcome, since it means the Russians would be taking somewhat greater risks in their decision-making. Wouldn't far outweigh any opertaional or financial benefits the newer designs offer? I suppose one could argue that if the loss of confidence in strategic weapons matched the actual loss in reliability, it might be a "no op" (although even this is arguable). But if the Russians think the US simulations suck, while the US is actually building really good simulations, the loss of confidence would be greater than the actual loss in reliability. In the extreme case, the nukes work great, but everyone thinks they are scrap metal. Of course, the same happens in reverse: if the Russians are upgrading their weapons to untested designs, the US may start underestimating the risk.