It should be pretty straightforward to build a PC for audio these days, but it might be worth doing some research, especially if you’re considering more unusual options.
For example, I’d steer clear of the higher end AMD CPU options (Threadripper) at this point, since they’re NUMA architecture. (Essentially multiple separate machines with separate RAM banks, and “high speed” serial interconnect.) They annihilate all competition in video rendering and the like, but the memory management causes problems for any non-NUMA-aware low latency audio applications, which is, AFAIK, still all of them. (Might be possible to work around with virtualization and VEP, but why build a single machine in the first place, then?)
Windows 10 also has a problem with multithreaded realtime applications if there are more than 28 or so logical cores. Don’t remember the exact core count now, but my 14 core i9 avoids that issue. The 16 core version does not, so unless M$ has finally fixed that issue, you’ll need to disable cores beyond 14, or (in the case of Cubase) ask Steinberg for the “secret” fix. I don’t know how this applies to other hosts, but AFAIK, the root cause is in Windows 10; not Cubase.
But, generally speaking, I’ve had pretty good experiences running Cubase, Sonar, and Pro Tools on various PCs, including some relatively inexpensive laptops these last few years. Using a consumer soundcard (as in, no native ASIO driver) is probably still a bad idea, but other than that, I think the days of IRQ conflicts, nasty GPU drivers and whatnot rendering the typical PC unusable for low latency audio are pretty much over now.