Well, that's not an innaccurate observation - That is what I am doing... And it was a choice, except z80/CP/M was missing from the choice, so the PC only ever got compared to things like the Amiga and ST and Mac most of the time... And maybe the odd workstation.
CP/M wasn’t part of the conversation because it had already lost well before the "PC vs. Mac vs. Amiga vs. Whatever" conversation had started.
I mean, it really is that simple. For all the reasons I laid out the CP/M-80 platform was essentially obsolete by 1983 at the very latest. (Picking that as the date because it's when CP/M Plus/3.0 came out, but the limited adoption of that release underlines that it really was already effectively a dead product.) A thing you need to understand, though, about how things were back then verses today, is that in the first decade (and a half?) of personal computing a "dead" platform could still enjoy a very long and healthy zombie existence. To be clear, I'm not just talking about die-hard users clinging like mad to their
discontinued orphans; there were plenty of companies that were perfectly happy to keep churning out new units of computers they'd mostly lost technical interest in for as long as people were willing to buy them. (In fact they'd often turn into lucrative cash cows the companies would milk vigorously to fund their successors; the Apple II is the poster child for this. The Macintosh might have gotten the bulk of the "mind share" when it came out in 1984, but there were reports that 85% of the company's revenue was still coming from the Apple II division in late 1985, and Macintosh sales didn't pass II revenue until 1988.) CP/M was technically a thing that was still "out there", but it didn't mean it was actually relevent in any future-looking discussion.
(And as I noted, that platform fragmentation issue it suffered from effectively made it even less worthy of discussion than the Apple II or a few other still-popular legacy 8-bit platforms, because by and large outside a few scattered markets there was never really a sufficiently dominant CP/M-centric platform to be worthy of specific consideration. Sure, in the US you still had plenty of Kaypros still kicking around doing perfectly good work as word processors or whatever, but there was nothing as compelling as, say, the Apple II's huge base of educational and game software, that would make it worth considering buying a Kaypro as your *next* computer despite its obsolesence.)
Sure, if you throw enough glitter around it you can make a Z80 computer that superficially can rival the capabilities of a mid-80's PC, or even a Macintosh or Atari ST... that is clearly the thing that's making this seem like a reasonable question to you. But it doesn't mean it was ever really a smart thing to do. Sure, if you already have committed to the Z80 for backwards compatibility or whatever then it might make some sense to keep evolving without changing that fundamental (that's why things like MSX2 exist), but the reason the 68000 CPU was the darling of the first generation of GUI-equipped computers wasn't because it was needlessly new and shiny; it was because its support for 32 bit direct addressing (no screwing around with segment registers or page flipping) made it uniquely well equipped to deal with pushing pixels around large framebuffers. Throw enough hardware acelleration and DMA at it a Z80 can work great acting as the conductor/traffic cop in an elaborate game system, no problem; it doesn't take that much CPU oomph to instruct a VDP chip to shove sprites around and respond to collision detection interrupts. But this doesn't work great for a "general purpose" PC; sprites don't help you a whole lot rendering WYSIWYG text displays, and general purpose 2D accelerators that *would* help with this task were very much in their infancy. (And usually cost more than your main CPU, and might not actually be a lot faster than it...) The Z80 is simply the wrong tool for this job. The 8086 *also* wasn't the ideal CPU for this, but it was still a lot more suited to big memory tasks than the Z80, which, again, is why it had already won that war.
At this point, sure, I guess you might ask "why if the 68000 was so much better than the 8088 why didn't it immediately render the PC as moot as it had the CP/M machine, especially given that the low-end 68000s were very price-competitive with PC clones?". And you're not going to like this answer, but it's in large part because of a reason you
mistakenly attached to CP/M:
installed base.
By the mid-80's the PC platform, because of its wild, unhinged growth, was
the de-facto standard for software and data interchangability;
a thing the industry didn't really have before now. CP/M gave manufacturers a common least-denominator thing they could customize for their hardware to get a basic computer out the door, but all the end user ever got out of it was mediocrity and massive fragmentation. By contrast, if you had a PC clone you had access to a huge library of software you could just pick and run off the shelf, you could bring work home, you could easily pirate all the good software you
didn't want to pay for, there was a slew of hardware expansions and options available... In short, there were all kinds of great reasons to buy a PC if you didn't already have one, and if you'd had one for a few years you'd have a lot of good reasons to look at getting a faster/more powerful one (which were now coming out) as your next one to preserve the investment. As I said, CP/M's installed base, IE, the base of *every CP/M machine sold, by every manufacturer*, wasn't really any bigger than the number of any single other 8-bit machine's when the IBM PC came out, (maybe half a million or so, total?) so there really weren't that many of them out there. And, let's be frank, software was
smaller back then. There certainly had been plenty of programs written for CP/M, but when programs were only a few kilobytes in size it's not like it was that much to ask to port the stuff worth porting to another machine. So... yeah, it really wasn't that huge of a deal for most people to just move on from CP/M when better options showed up. By contrast, IBM PC clones were selling on the order of
five or six million units a year by 1985, and because these machines had their larger memories and more elaborate capabilites this software was comparatively harder to port than Wordstar. There's a lot of value to consider here, even if technically an Atari ST might offer a little better hardware bang for the buck than an XT clone for the brief period under consideration.
Discussions around the value trade-offs chosing one platform over the other were way more complex in 1985 than they were in 1982, and when you figure in how the PC platform continued to evolve as Intel introduced faster/more capable CPUs (including jumping up to 32 bit with the 386, nullifying a lot of the 68000's architectural advantages), well... yeah. CPM-80 had no place in these discussions anymore. It had already been religated to legacy applications, toys, glorified game consoles, and a few die-hard cheapskates. Which are perfectly valid niches, no shame, but yeah, not a "serious" computer platform anymore.