Welcome to the forum!
And sorry if this reply sounds rude (not my intention at all). Wendell did a vid on a minisforum, framework or beelink, idk which one (can’t be bothered to look tbh). One of the mini-PCs reviewed lately had a lot of performance, close to threadripper performance.
That comes at the cost of power consumption. So now the question is “what are your demands?” 'Cuz running just plain docker in a VM is not that resource intensive. You can do that on something like an odroid h4+ or h4 ultra (not shilling or anything, I’m giving an example of a really low-power system that I personally use, the h4 base, so I can vouch for it).
But if your workflow demands a lot of power, then you might need something beefier, like an 8-core mobile chip found in most minisforums. If you’re going to run something like gitea, vaultwarden and 3-5 other lightweight containers, you can do all of them on the lowest-end intel n100 / n97 / n150 / i3 n350 (and obviously run more than just these). But if you’re doing something that takes a lot of resources, like running gitlab with many gitlab-runners and you have an advanced pipeline for compiling software, then you have to add a bunch of other non-light things, then you need moar powaaah!
And then there’s the question of power consumption. Electricity prices have been getting crazy for a while (at least where I’m at) and running a constant 90W / hour workload 24/7 could be something that chews a decent portion of your wallet. Mini-PCs aren’t that big power hogs generally speaking, but it depends on what software you’re running on them.
When it comes to proxmox, you should enable power savings in your PC’s firmware settings and run something like powertop in proxmox. But it’s still going to be dependent on your workload (again, if you’re compiling software for hours at a time, no power-saving mode will save you).
As for its capability as a hypervisor, it’s fine. It has some idiosyncrasies specific to it, but it’s still something I’d recommend beginners to use. The alternative would be running a desktop linux on it and using virt-manager or something like gnome boxes or karton to run VMs. It takes a bit more to figure the latter out, but depending on your needs, it can be done. But if you’re planning for a headless build, I’d say proxmox (although you can run virt-manager via SSH to your server, to keep it headless - but if you’re using windows, you’d have to jump through hoops, like using WSL to run virt-manager to then manage your server, just don’t bother if that’s the case and go with proxmox).
The advantage of running a hypervisor is the ease of deployment of VMs and the ease of creating a lab for your yourself. You can have multiple labs, segregated by a virtual router, so you can test to your heart’s content (you can have 1 playground for fast debugging and troubleshooting and a final UAT-like lab, where you strip all the debugging and testing options and you deploy a container close to what you’d ship to prod).