Post

Right now many services are subsidized and the companies behind them are buying up compute + memory + storage en masse, resulting in the hardware being more expensive. As subsidies decrease and prices go up, i expect demand for services and HW costs will come down. Could change the economics.

I may be uninformed, but my impression is that nobody is doing the combination of [training, inference as a service] profitably yet. AFAIK profits come from either: - running a model someone else trained - some other business that funds the AI work Am I mistaken? OAI/Ant profitable now?

my guess is the requirement of profitability on training+inference will start to make less sense over time, and that its largely all gonna be your second point, model training simply becoming a cost for another profitable business line. at which point opensourcing is usually pretty sensible as well