Maybe I'm just oldskool but I immediately want to get the whole dev && ops team on board to lower this number! Then the ultimate bragging rights can begin. What doesn't sound better than we run a multi-billion company on just these 5 machines. Perhaps more realistically 200k/4k ... 50 :)

Does it make sense to have so many nodes in a single cluster? Isn't multi cluster a better approach at that scale?