Chain of thought is just a way of trying to squeeze more juice out of the lemon of LLM's; I suspect we're at the stage of running up against diminishing returns and we'll have to move to different foundational models to see any serious improvement.