So yes, I'm basing my calculations on YouTubes service. What you're imagining is obviously an option for individuals that only want to serve to few individuals at a time, but is entirely offtopic to what was proposed here
He suggested $100/month for 300 videos and noted that he was experienced in content creating.
Literally the the first paragraph
He didn't actually say or imply what you think he did. In fact a later part of the comment ("10/month? That's 30 videos in HD max. $100? 300 videos in 4K... etc. ... or whatever.") clarifies that he meant what he wrote and it is something quite different to what you're picking out.
My point from the beginning was that $100/month is not enough to host a successful channel on YouTube. You absolutely can do it on vimeo, self hosted jellyfin or similar. Plenty of options around. It just get extremely expensive if you're serving content to hundreds of thousands or even millions of viewers.
You can probably start at $10/month with a hetzner shared instance and 1tb storage and unmetered 1 Gigabit Ethernet. Gonna be fine for ~2 simultaneous viewers
I only addressed the storage first because his idea was so far outside of a realistic price. Storage cost is a rounding error on the invoice if you're serving videos to consumers with views as a semi successful YouTube channel gets
I think that is still overcooking it. Peertube has a FAQ on the subject [0], going off that it's probably achievable to have 1,000 concurrent users at ~$50/month+storage. Not commercially advisable, it is pretty obvious that everyone in the game ends up going ad-supported and that makes sense to me. But YouTube could in support something like that for small channels without too much trouble. The issue here for the median channel case doesn't look like costs in the >$1,000/month range as much as foregone revenue and operational complexity. The vast majority of content creators don't get hundreds of thousands of views.
[0] https://joinpeertube.org/faq#should-i-have-a-big-server-to-r...
But your calculations are for a different problem. You're not solving for "what is the best I can do for $100". You're solving for "what do I need to clone Google's infrastructure for thousands of hours of video footage".
Eg, you doubled the amount of storage space to avoid rebuilding RAIDs when a disk goes down. There is an obvious alternative - don't do that, let the RAID go down and rebuild it while eating the downtime. That sort of thing isn't a big deal for discount video hosting. Maybe even force people to keep their own backups.