Do you have a source for the streaming v AI power consumption thing? I don’t want to be an ass, just genuinely curious. When I run jellyfin and stream something from my home server, it draws 35 watts. If I run an ai model on my GPU, it draws way more than that.

That would be huge and would make me and my friends switch from mumble/discord to matrix. However, i fear that high latency may be an issue.

For me, the benefit of discord is not the text chat but the voice aspect. With push to talk/voice activated transmission and low latency for gaming.

The voice features of matrix are more of a conference call thing.

Until this gets implemented in matrix (not sure if ever) I’ll continue hosting mumble (or TeamSpeak)

That’s not a real picture, the Tesla is photoshopped in (and a bad render at that, not even a picture of a real cybertruck).