My understanding is it means half of what a subscriber pays is spent just on the compute required as you chat with the models. Which leaves the other have to be

My understanding is it means half of what a subscriber pays is spent just on the compute required as you chat with the models. Which leaves the other have to be divided up among salaries, marketing, R&D, etc.