The instruction is smaller than “mov eax, 0” (because it includes the 32-bit zero) and the result is the same. Notice that smaller may mean more instructions in cache for a tight loop, so it may also result in better performace, too.

Fun fact:

nop is also just xchg eax, eax!