Real time video and audio are a different beast to text. The only way forward for these is P2P, not federation.

Maybe some Matrix like protocol for chat rooms (call them Channels, Communities, Club Houses or something trendy if you want) that also setups a P2P audio room, and a P2P video sharing system.

No one can independently host and transfer the insane amounts of data discord moves, but P2P networks resources for serving data grow linearly with the amount of devices consuming data. And audio/video streaming can be ephemeral, no need to save 80tb of a streamer sleeping.

The thing is how are they going to use the piracy bogie man to make all this ilegal hahaha.