Unlikely. Some people really just are better at low level than high level (and vice versa.) Further, I would say anyone who started after say... '97 never needed to have that layer, and '87 may be true too. The true pressure hasn't existed in over 25 years and yet we still have low level expertise because embedded will continue to be a use case.

The most powerful abstractions are typically not free. The rate of performance improvement from improvements in hardware has slowed down and might continue slowing down. Thus, the pressure to learn what happens behind various abstractions might actually increase, if advantageous performance improvements can primarily be obtained by peeling back the layers of abstraction at the cost of some development efficiency, since you can't just wait a couple of years for hardware to develop to a point where it can handle your inefficient abstraction anymore.

> Or are things like Rust and embedded AI going to bring in a bunch of new blood?

AI product dev hat on for this comment. My take is AI will continue to give solid first drafts, and expertise, creativity, and curiosity are required to take it the rest of the way. Maybe, just maybe, we get beyond the GPT architecture and embed a reasonable state of the world for wider use cases similar to a CoPilot.

We are getting more abstractions, not necessarily better ones. I much prefer working with people who figure out how things work under abstractions rather than relying on them blindly.