If RISC-V is a success, there is a high probability that assembly will come back hard: "write 64bits RISC-V code paths once, run everywhere..."
Not to mention that it is not that hard to port assembly code between modern ISAs. For instance, porting from x86_64 to 64bits RISC-V should not be that hard, very probably brutal... but really not that hard.
Reminds me of one of those Emacs memes that say, 'Your text editor runs in a browser, but my browser runs inside my text editor. We're not the same.'