IBM didn't make a "decision" about the BIOS. They didn't even come up with the name, they copied it from CP/M which Gary Kildall split into the "BIOS" (basic input/output system) and "BDOS" (basic disk operating system) at the request of Imsai when they wanted it ported from the altair to their S/100 clone. The split was explicitly to make porting to new hardware platforms easier.
The PC was _designed_ as a 16 bit sequel to the CP/M boxes, the 8086 was a 16-bit upgrade to the 8080, the OS was a 16 bit port of CP/M (either CP/M itself from Digital Research or Tim Patterson's QDOS clone rebranded by Micro-dash-soft), the ISA bus was the S100 bus with unused wires removed (there was even an adapter to fit the big cards in the small slot, the voltage and timing of signals was all the same you just had to connect up the appropriate pins). The PC had a BIOS so it could run CP/M (DOS 1.0 was a straight CP/M clone, then Paul Allen extended it by adding a bunch of Unix features during 2.0 as a transition path to Xenix. Until he came down with Hodgkins Lymphoma in 1983, Microsoft's planned successor to DOS was Xenix. Messing around with OS/2 and Windows happened after Paul Allen left and Gates started following blindly after IBM and Apple because Allen had been the ideas guy, Gates was always about cashing in on _other_ people's ideas.)
So when Compaq cloned IBM's PC bios, it was not exactly a new idea. Despite IBM's best efforts: that's what it was FOR.
I mirrored an old geocities site on CP/M history way back when that covers some of this: