Not on a regular basis but I have run 7 VM's at once a few times, 5 VMs multiple times and do occasionally run 2-3 VMs.
Obviously it's going to depend on what those VMs are doing but in my experience when doing this the biggest bottleneck often turns out to be less the cpu and more the NVME SSD and chipset, specifically the I/O. Not bandwidth, too many requests for data flooding out the bandwidth.
While it depends on use case but If I was doing high load 7 VM's all the time I'd be grabbing as many cores as I could with a cpu/board/chipset with the highest I/O I could find (AMD), with the highest number of NVME drives all of which would have the highest I/O numbers (average bandwidth is fine), all running in raid.
Still rocking my 8700k, it's still a strong performer and I expect it to remain so.
We've long passed a point where CPUs have largely peaked and software is now the major performance hurdle, it just isn't able to take full advantage of more cores. Once it does though, drive I/O will be an issue, though that's actually a much easier fix. So it often doesn't matter if you have 32 cores and a full speed 4th gen NVME, the I/O and software will still hamstring you to older gen/slower performance under most loads.
If I was to upgrade anything it would be my gpu/screens in order to get higher refresh rate before I upgrade my CPU.
Benchmarks and CPU performance today is a massive lie, sure it's faster, but by how much real world, we have two issues colliding today more than ever.
It takes more and more power to go faster, adding 2 cores to 2 existing was a big jump, but adding 2 more cores to 12, not so much. Worse, a LOT of software has trouble using them (especially higher core counts) so adding a novelty amount isn't really doing much of anything and adding a bunch to a bunch also does very little. Worse still, we're in a march to zero in more ways than one, we can only reduce trace size so much before it becomes atomic scale so we're nearing an end to faster clock speeds but the same is happening to processing itself. Let's use boot and render times as examples, a 10% reduction in boot is significant if your boot time is 3 minutes or your render time is 5 hours, but if your boot time is 8 seconds and your render time is 20 seconds, a 10% reduction translates to less than a 1 second reduction on boot and 2 seconds on render. Keep in mind, 10% takes a massive bump in power to accomplish today.
Worse still, it's only getting worse as time goes on. We need a dramatic shift in software and hardware to really see any sort of significant increase at this point (except edge use cases). We can't keep adding mhz, and more cores are not going to help if software can't even use them and even if they could it's not linear like mhz because not everything is helped with more cores. Cores are like driving a buss around a race track, you get more people around at one time but it's not going to be fast.
Bottom line, in day to day use (including games), a 16 core or 32 core cpu probably isn't going to be a dramatic upgrade unless you're using a potato from several generations ago.