Right. In "-f bin" mode, Nasm merely moves "section .data" after "section .text" and "section .bss" after that - they are not actually in distinct "segments". In a .com file, your stack is above that, at the top of our one-and-only segment - 64000 bytes in ".bss" is going to come awfully close to that, depending on how much code/data is present. This "could" cause mysterious problems as your code grows...
FWIW, if you want all those zeros in your "on-disk file"...
; in "section .text" or "section .data"
times 64000 db 0
... or simply ignore the warning...
Attempting to put initialized code or data in "section .bss" is "ignored" - probably an actual error!
Best,
Frank