I'm imagining a 30 year old Pentium Pro server grinding away in a broom closet somewhere. It's next to the one still running the old Space Jam website.
Once upon a time I mirrored Project Gutenberg. As this was a pain having only 64kbit over ISDN, I was expecting great things from attending a LAN party where we actually had Ethernet speed connection to the internet.
When I started my mirroring script, though, it was not faster than at home. Turned out that the mirror host limited connections to clients in unknown networks to 64kbit.
Luckily I found another mirror host that had no such limits.
It's a bit of a problem that there's no serious contender to NixOS. Especially Guix is in a good position to become an alternative.
But it will never happen, because of GNU. And before I continue, I want to make clear that this is not to shit on them.
But realistically, only a fork could make it relevant. NixOS, despite its issues (documentation, flakes, whatever), has a massive mindshare: it's a huge repository with very up-to-date packages, a lot of modules, and devshells are just a very handy thing for developers. You often find flakes in random GitHub repositories for that reason. There are sponsored efforts around the distribution (like lanzaboote). There are (semi-)commercial entities set up around it (numtide, determinate systems, tweag...)
The difference between NixOS and Guix is probably so large that no commercial provider would want to put in the required work to bring Guix up to speed, and GNU is committed to other values. As such, I think only a very big volunteer effort could make a difference.