The relevant question is the degree to which memory errors remain a significant source of bugs in modern C++ code. Are they now of mainly theoretical importance? I can't answer that definitively, but I can say it has been many years since I had a bug in C++ code that using Rust, instead, would have prevented. My bugs are specification bugs. Memory bugs just are not on the radar. But coding Rust takes me enough longer than C++ that I can't justify it.

Perhaps the driving force demanding memory safety isn't for one's own use, but rather the hope that the code one depends upon, written by other people, will get better. I.e., I write good code, and you write good code, but God forbid we have to rely on Microsoft, Google, Mozilla, and Red Hat monkeys to write good code! That's a legitimate desire, but it doesn't drive much personal adoption. Meanwhile, the PITA factor -- borrow checker not happy! "&" required! "&" forbidden! -- interferes with personal adoption of Rust.

And C++ keeps adding features. Not to fall increasingly behind, Rust needs to multiply its adoption rate. I don't see how that could happen. C++ is already good enough, so Rust can't cannibalize significantly from C++ coders. Rust is enough better than Go, Java, and C# to be worth a significant effort to switch, if users of those languages could perceive it. The greatest benefit to the world derives from driving people off C -- to Rust or modern C++, it matters little which.