Yes, it's because it keeps track on object lifetimes and data access when sharing objects, even across threads. It means that once things compiles a whole category of common and often difficult to debug errors are gone. It means much less time debugging and fewer issues once in the hands of the end user. There can still be bugs but it's more about logical errors than difficult memory issues.
As a C++ dev for 20 years, I love Rust. Humans are fallible, even if endeavouring to use safe patterns. Might as well just let the compiler use some CPU cycles on that.
I have Intel iGPU and Nvidia dGPU on my laptop, it works perfectly with Prime offload. I never need to switch, it uses Intel for desktop/VA-API and offload for games. No issues, at all on my distribution.
Anyway, every thread have your kind of unhelpful comment. The thing is some people have Nvidia, some have AMD and AMD also have bugs. Let's try to make everyone happy, not everyone have piles of money to throw after new hardware.