I used to say studying semiconductor roadmaps was the easiest way to predict the future. That is still true, but in the angstrom era studying the WFE/manufacturing product roadmaps may actually be far more insightful about our compute future.