In the early 80's it was Bill Gates who made strategic decisions for MS. That was even before they went public. My wonder is if Gates had ever used Unix. He (personally) developed BASIC for a CPM (I think) machine. I am unaware of any system level skills in his experience.
I can say I have personally seen him do so ;-) He and Bob Greenberg were at Ricki's Hyatt at the infamous meeting with AT&T in late 1979/early 1980 [I have forgotten the precise date it was early/mid-winter IIRC -- I was there as ½ of Tektronix's reps - this meeting would lead the Sys III license]. Bob and Bill had some sort of PC with them running a pre-Xenix of some type as an example. Please remember that he and his team ran the Seventh edition of UNIX on a PDP-11/70 at Microsoft (called 'kermit' IIRC) and also TOPS-10 on a KL -- I believe that both of these systems are at the LCM in Seattle these days (which is currently in mothballs due to CV-19 and funding which is a real shame).
If he had knowledge of or used Unix or XENIX (for which he had a master license from AT&T), why on earth would anyone go down the bazaar path of DOS with lettered drives, tortuous IO interfaces, and assembly language source code?
To answer this I have a few educated >>guesses<< which are based on the history of the times and practical reality. Gates (and Paul Allen) had personally grown up TOPS-10 and RSTS in HS and in his 2 semesters at Harvard; so the DEC disk naming scheme for having the system written using assembler was natural to him since DEC did that too. And second, DOS was purchased from Seattle Computer Products (SCP - story told elsewhere and not UNIX history) and it has been written to be modeled after CP/M (which had been modeled at RT/11 and DOS/11 - the last two again using DEC style naming conventions). Interestingly enough, CP/M had been written in PL/M which was Kidall's simplified PL/360 style language for the 8080 that he eventual sold to Intel. I was under the impression SCP used 8086 assembler language for the development of their DOS86 system which was the direct parent to MS/DOS - but they might have used PL/M. So you can add, that an issue at the time was that Intel's PL/M tools were not very portable and since the primary development systems at Microsoft were TOPS-10 and RSTS (and Bob was trying to replace RSTS with UNIX at the time), I don't think there were PL/M tools that ran on them.