Other than using COTS parts (incl. the CPU), the BIOS, while proprietary, was in a way the weakest link as far as cloning, since it established a ROM-based standardized hardware interface that isolated the OS from the hardware.

Companies like Compaq, and later Phoenix and AMI, were able to get around the proprietary nature of the BIOS by building clean-room BIOS clones that withstood IBM's legal challenges.

However, given the willingness of Microsoft (apparently with little IBM could do about it) to sell MS-DOS variants to others like Compaq, and later the emergence of MS-DOS clones like DR-DOS, it's not obvious that clones might not still have taken off without the unintentional assist of the standardized BIOS interface.

I wouldn't say the BIOS was the weakest link. It was really the only obstacle, albeit a weak one. Surely, the BIOS was clean-room reverse engineered very soon and after that, the PC-clone market just exploded.

However, if there were no BIOS, the thin hardware abstraction layer that the BIOS provided would be part of MS-DOS. I see only two historical alternatives from that:

1. Microsoft would have had an even greater upper hand in controlling the PC market.

2. IBM could have kept the BIOS proprietary (even though as a part of MS DOS), which prevented Microsoft from selling MS-DOS independently with an IBM PC abstraction layer.

However, even if option 2 prevailed, Microsoft could have created its own BIOS to ensure that software written for MS-DOS would be compatible across the PC clone market.

There’s a lot, lot more to a PC being standard than the BIOS API. Attempts like the Tandy 2000 showed that hardware that deviated too much couldn’t run the same software and failed.

> standardized hardware interface that isolated the OS from the hardware

Exactly the point of having a BIOS. CP/M had one as well.