You can't stream hundreds of videos for tens of thousands of viewers, which is what Dave uses YouTube for.
Bittorrent has proven its scalability for well over 17 years.
Well, just because you can download a copy of a recent Blu Ray release of some blockbuster in a couple of minutes doesn't mean you would be able to stream hundreds of videos by using the same protocol.
Bittorrent transfers files by little blocks and it usually does that without any specific order. So if there are a lot of seeds, each of them can just transfer slowly their share of blocks - so the overall speed will be incredibly fast.
Streaming works differently. Each seed would have to transfer with the speed that is equal or faster than the video's bitrate. A FullHD video with decent quality can have a bitrate of 10 megabits per second. That's quite a lot for a single seed. Assuming full-duplex 50 Mbps connection it could probably serve 3-4 clients before totally saturating the connection.
Then you have the storage problem - how many people would actually have a lot of videos from, say, Dave's channel, which currently has more than a thousand? Then what about monetization? A lot of video creators are only able to create because they receive some income from doing it, how you would provide it with Bittorrent?
And we didn't even start talking about initial distribution, video encoding, search engines...
Bittorrent is sure great for file sharing. Not so sure about video streaming. It can work in some cases. Probably not in general case, until 1 Gbps speeds or similar will become commonplace.
You can't stream hundreds of videos for tens of thousands of viewers, which is what Dave uses YouTube for.
Why not?
I'll try to be as short as possible... Basically, it just requires way too many resources. In order to make your own little YouTube, you have to provide the following:
1. Video encoding. It's a very costly task in terms of computation resources even right now.
You can try that at home. Just attempt to re-encode a FullHD movie on your home PC. You will quickly discover that in order to achieve a high quality/bitrate ratio you will have to use aggressive settings. Those will make the encoding incredibly slow even on modern CPUs. Soon you'll want to use a dedicated server for that task, with a specific CPU that has a lot of cores or some sort of dedicated hardware for video encoding.
Also remember that you won't have to do just one encoding. You'll have to make versions for 720p, 360p, etc., since there are people on slow connections and old devices.
Those kind of processing power is expensive. It's not a cheap VPS you can buy for $20 per month.
2. Video streaming requires lots and lots of bandwidth. Even if you find exaggerating that 10 megabits per second figure I provided earlier, even if you assume just 2,5 megabits, for example, it will still require insane amounts of bandwidth for any kind of significant audience.
The latest video from Dave already got >30K views in just 21 hours. I can safely assume that there could be a thousand of viewers watching this video simultaneously. Assuming it has a bitrate of 2,5 megabits and everyone watched it in that quality, it gives you 2,5 Gbps bitrate. This is not a usual speed. Most hosting services offer just 100 Mbps data links. Everything above that will surely cost additional money. Pretty sure it won't be cheap, and not every hosting provider is technically prepared for that kind of load on its infrastructure.
Also data transfer itself usually costs money. Those 30K views assuming the same bitrate for a 10 minute video would be ~5200 gigabytes of data. Not an insignificant amount. Definitely not the usual amount of data transfer that this forum, for example, has.
3. But assume you can deal somehow with tens of thousands of views per day... What will happen if one of your videos will become viral or simply more popular than the rest? Will your infrastructure handle hundreds of thousands of views? Millions?
When a video becomes viral or significantly more popular than the rest, it is a huge opportunity for a channel. It can bring a lot of new viewers and help to achieve additional notoriety. That is, if your infrastructure can handle it. Otherwise it will become just a missed opportunity and a service outage for the regular viewers.
4. Now think about how people find videos in the first place... Dave mentioned in one of his videos that around 50% of viewers come from searching. It's possible that I myself discovered his channel when I was searching for some specific topic on electronics.
This happens because YouTube is integrated with Google search. Will your own hosting service offer the same kind of visibility for your videos?
5. And that's not all. But let's forget about little details, I think what I described is already enough.
If you don't agree with my points, really you don't even have to argue. Just go and implement your own "YouTube" that's capable of reliably serving tens of thousands of views per day and offer it as an alternative to YouTube for Dave. I'd be really happy to be proven wrong and I would use that alternative instead of YouTube.
Unfortunately, it's unlikely that it's possible to do that without spending an amount of effort and resources that is completely out of reach for most people.
The argument people often make is based on costs which were much higher in the past combined with file sizes that were also much larger. If you have a modern computer it seems that the file size now for many videos is not so super large. So, maybe if you do it that way you cant get the huge volume and you do have to pay somewhat, but I am also speaking more generally about other kinds of content, even static HTML. I find it totally depressing that people dont make their own web pages+ serve them.
Right now you can buy an inexpensive VPS for $5 per month that can effectively and reliably replace your mail from Google, host your own blog and website (not static HTML, but something more interesting, like Wordpress), which you can use as replacement for your social networks and do some other neat stuff, such as file hosting, code repository, etc. It requires some effort, but it can be done.
Video streaming, however, is still out of reach for the reasons I described above. Yes, technically you can do it... For 10 simultaneous viewers or such (discarding the video encoding and all other stuff). But for anything serious it just doesn't work.