We could do 1000 tps if we want to require every operator to run 32 core super server with T3 internet, for example. But not many people can afford to run such a node, so it becomes more centralized the higher we demand. But for a private deployment of the tech, it's up to them. The higher the minimum specs you require of each node, the higher you can set the maximum TPS. So the question isn't really how many we can handle, it's how much power we want to require each node operator to have.
