You're discounting the network effect of compatibility. I don't even mean "network" as in networking, but in the ability to save a Word document or whatever on a floppy on virtually any PC and load it back on another. This was long, long before every device could render a web page (and you could take for granted that it would have a TCP/IP connection!)

There was huge real value in creating "the standard", not least because it enabled some truly phenomenal economies of scale. I mean, the 386 had the hardware for protected memory, virtual memory and preemptive multitasking. Or you could pay 10x as much for a contemporary SPARC or MIPS processor.

In 1994 I was at college and we were doing some numerical stuff, metal fatigue IIRC. You could run your code on a SPARC 5 or drop into DOS on a crappy PC and give 100% of the CPU to your code. Guess which was actually quicker...

IBM mistakenly asked Microsoft for an 8088 version of CP/M (the predecessor to DOS). Gates then bought a clone called 86-DOS and slapped the MS label on it.

If MS didn't exist, someone else could have easily replaced them. It was also mainly due to Intel, who created such a cheap processor, that we have cheap computers today. Not Microsoft.

I was able to save a wordstar document on any CP/M (including CP/M 86) machine and transfer it to any other, including my Apple II running CP/M through a coprocessor. Most other word processors of the time could exchange files in one format or another. I never had any problems with it.

And you shouldn't be using a SPARC 5 on 1994 anyway.

That is true, but it is also true that CP/M was never dominant in the same way that Windows became. At the same time as CP/M there were all the micros (BBC, Commodore, Sinclair, Amstrad), Japan had the MSX "standard" etc etc.

Whereas in the Windows era you were likely to fnd machines running the same applications everywhere, even at home.

CP/M was very important in the small-business microcomputer space, offering a measure of compatibility other platforms couldn't. Minis lived above that space and didn't compete directly with micros in the small-business.

The home computer space was much more fragmented and was where Apple, Commodore, Atari, BBC, Amstrad and Sinclair competed. BBC, Amstrad and Sinclair were important only in Europe, further fragmenting that market. It later became dominated by the same IBM-PC clones that took over small businesses.