Code was never the bottleneck
The scarce resource on an engineering team has never been typing speed. It is judgment.
For a long time the story was that engineering teams move slowly because writing code takes time. That story is getting harder to believe. Code generation is fast now, and getting faster, and a team that can produce ten times as much code does not automatically ship ten times as much value.
What actually slows a team down is deciding what to build, what to cut, and which shortcuts are safe to take today and which ones turn into a rewrite in six months. Those are judgment calls, not typing exercises, and they do not get faster just because the tools generating code got faster.
I think this is the real shift worth paying attention to. As it gets cheaper to produce code, including code nobody asked for and code that quietly does the wrong thing very confidently, the bottleneck moves upstream, to the person or the process deciding what gets built at all. Reviewing decisions is going to matter more than reviewing diffs.