on the scale of YouTube
That’s precisely the trick. Don’t try to copy Youtube. You’re gonna lose. And it’s not the Peertube intended use case anyway.
Instead, Peertube and other such platforms should work as cross-indexing domain-specific, configuration-specific video galleries. A retro videogame video archive does not need 4K 120fps Dobly 14.3 audio; they can just encode most of everything in 480i 30fps and their storage costs will go down significantly. A news report / news reel archive can save some costs by encoding as SDR (or even lower) with ~80kbps mono MP3 audio or somesuch, since most of everything past the intro jingle is human voice.
Play to your advantages. Trying to break into a monopoly game where the rules are broken, the only other player is broken, and the entry fees are broken is self-defeating.
Same as usual: the mouse loses unless it assembles and unionizes with other mouses and they bring in a guillotine.