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.
They say the same company worked in a Z80 version but I haven't been able to find any info onbthat, does anybody know? Especially anything related to a modern version of the old Amstrads.
As foretold in Matrix (1999)