100% nuclear is unfeasible, because nuclear can't be used to respond quickly to demand changes. Nuclear plus hydro plus some fossil (or recently alive biomass) would work. A very quick calculation shows that US produces approximately 19% of power using 100 reactors, so roughly another 250-300 reactors, depending on how much wind and solar (and other) power is utilized. If you want to move as much of traffic to electric power as possible you need to approximately double that to some 600 reactors or so, at a cost of some $10 billion each. You can cut down the costs somewhat by building multiple reactors in one power plant, but we're talking about an investment in the general vicinity of 20-30% of US GDP spread over several years (up to a decade). Roughly the entire US military spending over two presidential terms would be in the correct order of magnitude. Expensive and difficult, but actually within the realm of possibility.
However ... that's the easy and cheap part. The much more difficult part is replacing all the established infrastructure, building new railroads for freight where needed, etc, while fighting a massive political battle against oil lobby, against states with oil production, against Greenpeace, against Russian and Saudi money, against car manufacturers that spend good money on engines and so on and on.
It would be extremely difficult to do without this last bit, I'd say "maybe possible", but no more. This last bit is more difficult than the rest combined.
McHrozni