To a first approximation, I expect companies to spend nothing on AI and get put out of business if they are in a sector where AI does well. Over the medium-long term the disruption looks so intense that it'll be cheaper to rebuild processes from the ground up than graft AI onto existing businesses.