The next one if they make it should be way better from the off. Going from a Witcher setting to a Cyberpunk one was never going to be easy especially with the added scale of the city, but if we get a sequel they'll already have a foundation from this (after they've got the bugs and other stuff out of the way) and won't have to deal with last gen consoles either which should make things a lot easier.
"Setting" of any kind is irrelevant.
Bethesda, RockStar, and many other dev studios regularly produce multiple games across multiple subgenre settings.
What is relevant however, is WHAT they were working with. As I understand things, The Witcher 3 was built on a vastly different (and somewhat inferior) version of RedEngine, while Cyberpunk 2077 was developed on the new RedEngine 4.0, which was apparently only in production while the game itself was still basically in concept. In
2015*. The game was announced 3 years prior in 2012. Which means the game was still mostly (maybe completely?) concept until the engine was complete.
* 2015 is important here, because it stands to reason that nothing was "presentable" from a gameplay perspective as production on the engine was barely under way by then. So even under the most generous of circumstances (engine finished sometime in 2016), the game saw, at best, between 2 and 4 years of actual dev time. For a game of this magnitude, that's little to nothing.
Any work they may have done in RedEngine 3 (assuming there was any), would then need to be updated or completely retooled for RedEngine 4. And that's not accounting for any RedEngine 3 systems that may not be compatible with RedEngine 4. RedEngine 4 touts all new "advanced animation and AI" systems. Neither of which are on display in the current iteration of the game in any obvious "advanced" capacity.
Simply put:
Somebody bit off more than they could chew in the timeframe they agreed to deliver on.