>>106293486
Sam Altman just did an interview where he said OpenAI makes money on inference. Anthropic definitely has fat margins on inference. As models become more efficient (e.g GPT-5 thinking token use is much lower for comparable responses compared to o3), and algorithmic progress continues allowing model makers to train smaller models with comparable output (just look at the capabilities of models of the same size now compared to 2 years ago), it won't be long before value added inference resellers like Cursor, et al are break even and then profitable. Literally everybody will win. Denial of the inevitability is a sign of low intelligence or typical contrarianism