Do we have to find out where we start sharding the instances, or breaking them off into others, like how many people can fit into one instance?

Part of it's from game design and how the game feels. Part of it's performance-wise, since we are still fairly early in the project, so we haven't done all the optimization we're going to to maximize how many we can fit on one game server. Then there's going to also be how many we can fit on one machine, how many game matches we can fit on one machine at a time.

Now the machine, you mean the ones in the cloud?

Yes.