I’m no expert though, so I’m more curious than in disagreement.
Plus technologies like PeerTube can significantly reduce bandwidth expenses if that's really an issue. Google wouldn't use something like that because the federation threatens their monopoly, but a competitor with a different business model definitely could make it work for them.
They don't need to, they have caching servers on most ISP's internal networks. Popular stuff (that consumes most bandwidth) just needs to be sent once from the mothership.
A fresh new competitor to YouTube would face a combination of a lack of advertisers, and a lack of content seeking ad-funding.
On the other hand, when it comes to pay-to-watch content, there's space in the market for new competitors to arise, as the network effects are less significant there.
https://rakuten.tv/ and https://chili.com/ for instance, are far smaller than Amazon and Google.
I guess bandwidth costs are, in a sense, the problem. If if were easy to run a very-large-scale video-hosting site, that would help, sure. Ads wouldn't have to be worth much to make the business work.
1 tb bandwidth cost of aws is 236 usd a month.
[1]: https://calculator.s3.amazonaws.com/index.html#s=CloudFront&...
That will serve your 49MB video roughly 2 million times a day.
Let’s go with 50% average utilization, and say that means we need to monetize roughly 0.0065 cents per view to cover the bandwidth cost.
I think YouTube charges something like 10 cents per ad view, but won’t charge if the user clicks Skip in that 5 second window. I’m not sure what percent of views result in a billable ad, but anecdotally I’ve heard it’s surprisingly high.
I would assume storage is a much bigger cost factor, to say nothing of paying the engineers to develop and support the infrastructure.