The hypothesis is about theoretical future models, not existing public ones. I'm making a prediction based on a pace of improvement that I believe will probably continue unabated and which likely is not going to hit significant bottlenecks for very long.