I also have late 90s Red Hat, Suse, and FreeBSD official media stashed as well lol.

You're not going to get any of that happening while the AI bullshit bubble is ongoing. Bandwidth hasn't skyrocked of course. But storage space has. I don't see any large central entities stepping forward to do this without some well-defined means of support. The kind that have never existed for such fediverse endeavors so far. And not just for storage space and bandwidth. But for the lawyers as well. You better believe the IP police will be all over them.

If it is to be Well decentralized. Well inoculated from legal challenges and financial constraints. Blind caching proxies Would be the most sensible option. Caching proxies are vital to every streaming service including YouTube. Which peertube is imitating and not just a name.

If you want to know why you have to buffer longer and more often than services like YouTube. Even for things just like rewinding or skipping ahead. Proxies. Your peer tube request may have to traverse mountains and oceans. Your YouTube Netflix or Hulu stream usually is from a town or two over.

Technically we have webRTC and the other Technologies behind peer tube to do similar things when there's a large group all watching a single stream. But the Swarm has to get set up again over and over every time the group drops. Which wouldn't be a big issue on a well-established service with people constantly using it. But right now. It means that you are generally the person caching and streaming to others if any. But no one streaming to you. (But the source) Hence lots of buffering and slow streams.

If I could as a user run a small program to cache for my favorite content creators. I'd be there for it.