HF wise, the 360 can be seen as a provable, public witnessed, metric for amount of users being able to be on a single instance but not fully objective one as there are too many unknowns and was most likely unable be objectively proven by talking to absolutely everyone in the crowd to prove them as individuals because there are just too many.
It shouldn’t be total concurrency you compare, instead it should be how many you can have in a single instance or local concurrency.
Total Concurrency wise, High Fidelity is laughably tiny but skewed as only some use steam clients, and of those a fraction have steam friends enabled: So hard to get proper concurrencies for everything without someone redoing that concurrency app made a while back, which unfortunately does not differenciate between users and assignment clients.
We barely keep around 5 people online at any hour outside of events, sometimes none. Some would consider it Dead; but thats how it has been for years. These blips in usage when we peak at hundreds is when we cram them all into a single instance.
Regardless the details are fuzzy and its hard to compare to anything else as there is nothing else to compare to.
Closest Analogue would be SL, but they dont stream animation data, that is just voice based, and they can hit 60s stable according to their metrics (100 max), with events located in-between four regions, setting it to 60 x 4: But that wouldn’t be on a single audio mega server, and you’d have scaling issues and voice communication latency issues between the ‘regions’ as everything scaled.
MMOs tend to heavily use to instancing; so even a single server, a single area can have many instances. You don’t need to look any further than WoW for example or even OrbusVR, which would be closest to Hifi, interms of context and population (VR MMO). So its hard to say, but generally those hundred thousand or millions players, tend to be separated into many, many instances.
MMOs also don’t tend to include audio servers, because they know Gamers will use external software, so they save dev time by just focusing on more important things to them. VR might make it a must, but you still will end up people using mumble or discord for Games. This allows the Provider to save money on bandwidth, as game data tends to be a lot more lighter to transmit back and forth than audio packets. It would also requires a separate networking team that specializes in audio streaming.
Collectively every time you mention In-game Voice chat to dedicated gamers, they will groan and most likely just mute it as soon as possible, because of no quality control on mics, toxic players, and overall feeling that a whitelist approach to voice chat is better. Even I agree with this, whenever I play any Game.
SocialVR is different from the MMO Games genre, but it is relatively unexplored territorry. Second life might be the prelude to it: VRChat is the most popular one.