In addition to what the others have said, the specs were often a generation or two of what was available in a generic Intel box. When it was introduced in 1993, the Indy, SGI's lowest-specced workstation, could handle 256 MB of RAM and was clocked at 100 MHz, which was way beyond anything you could get for a PC, and that's before even mentioning the dedicated graphics hardware for 3D and video workloads. If money was no/little object, then the workstation vendors were happy to take your hand. (And your wallet.)

As efficiencies in cutting-edge hardware improved, the gap closed. Intel and AMD leapfrogged the smaller design firms running these companies, and more and more vendors threw in the towel on hardware design, switching over to standard x86 hardware. By the early 2000s, distinctive OSes like Solaris and NEXTSTEP were just legacy GUIs that could be installed on commodity PCs, although many flavors were discontinued outright in favor of Linux, leaving these companies (several of which were swallowed by HP) without any moat or vendor lock-in. (Notably it happened to NEXTSTEP twice, once in 1995 and again a decade later when Mac OS X 10.4 was officially released for Intel CPUs.)

Microsoft itself was a leading Unix vendor in the 1980s with Xenix.

Every Microsoft developer had a Xenix workstation for things like email, access to network disks, running a decent C compiler, and debugging.

DOS was practically a single-program environment with no memory protection and no networking. Unix offered much better productivity for software developers.

Engineering in general was a field that used Unix workstations heavily. Microsoft didn’t become competitive until Windows NT in 1993.

Multi-tasking that didn't suck.

Same as we use it now, to be frank. Unix workstations as an interaction model have persisted so long because it works just great.

I was writing a lot of Unix software in that period - database apps, business logic, and so on. For me, using an MSDOS-based system was a compromise, which I enhanced by using Desqview to get multi-tasking - it allowed multiple MSDOS instances on a single machine, in which I ran terminal software, compilers (our apps were being ported to MSDOS...), and database admin tasks - just like today.

What we have today in the form of MacOS or Linux workstations is pretty much what we had back then, too. The power is inescapable.

Nitpick: This was the 90s already. 68k based Sun gear was in the 80s but 90s was the RISC era already for Unix boxes. (except for NeXT I, they continued launching new 68k models into the 90s and got left behind)

Amiga OS was a kind of plan B, as they originally were thinking of UNIX, when we listen to history from Commodore employees.

I was really happy they went and did their own thing, classical UNIX was never great at multimedia.

And yet... !

AMIX was actually one of the first gcc targets, as mentioned here: https://www.reddit.com/r/amiga/comments/1ikyw0s/the_fsf_free...

It seems it may have had a pivotal role in the history of the FSF. So, clearly, someone found value in it!

It’s a bit mixed, by 1990 most UNIX vendors were moving to various RISC architectures so a 68k based workstation would appear rather old fashioned for that market. People paid Serious money for a UNIX system to do Serious work, so why cheap out on yesterdays technology?

A/UX didn’t seem to do that well in the market either.

The IIfx wasn't expensive (or fast) because it had a 68030, also it was DOA lol