Every is pretty broad.
The pragmatism of self-hosting is key parts of the software will be missing and you may not get a comparable experience.
The issue is not charging. The issue is where key, and actually core, functionality is sitting on the paid side. The idea that "at some point" you hit a paywall is generally not true. It can border on crippleware, and might as well be shareware (still great to support small development teams).
Evolving licensing and revenue generation is key.
The reality is we wouldn't have the world we have today without open-source software, as in the totally free kind, and at some point it does need to be maintained, unless some of the few packages reach some sort of maturity.
Your point about sysadmins not being cheap is fair. I would offer a counter-balance and ask how many more things than need to be are made to be far more complex than they need to be. After all, open-source is also a place of learning, experimentation, and breakthrough, and technical debt.
Still, other projects get so much done with so little support. Restricted core features pretty much tunes out a lot of users to get over the value inertia. It's hard to call it open-source, when it's not quite open source anymore.
One license I've seen is requiring companies over a certain revenue or headcount to have to license. Even JIRA had some of it figured out with their lowest license offering, but for a ton of functionality)
AGPL is helping with some of this - the more projects evolve to partner (for example if someone wants to license the tech to be part of an unrelated project)
Maybe the product managers should be the development team.
It's far easier to financially support fully open source software even though it might not happen as much as it should. I've recently come across an excellent loom alternative in Screenity and the developer only has a few sponsors.
Packaging open source for supporting the development of the software can come in many forms.