LLMs won't probably be the models for "super intelligence". But nowdays, how corpos can "justify" their R&D to spend gigantic amount of resources (time + hardw

LLMs won't probably be the models for "super intelligence". But nowdays, how corpos can "justify" their R&D to spend gigantic amount of resources (time + hardware + energy) in models which are not LLMs?