It appears as if improving data corpus quality and size and improving processing capacity are still driving performance gains. I have no idea of the functional relationship, and its likely not a Moore's law kind of thing, although that would be an underlying driver of available capacity to saturation.