The only thing the $1T model needs to do is find some algorithmic speedup which allows it to be trained at $100B. I'm not saying that's easy or that it will happen but I just don't see why not.

I don't understand the logic here - a methodology to increase efficiency by 90% in training doesn't exist until it does - could you explain what you mean by "I don't see why it won't exist"? Are you seeing consistent gains by some process?