Ramez Naam quoted the post and Ben Golub reposted it.
Ben Golub#1451
It's bizarre that some claim the LLM industry might be a monopoly. Based on what I've seen, this is one of the most competitive industries ever. Yes, the number of players is small, but the product improvement and leapfrogging is unprecedented. Of course, things may change in the future.
Users support claims of high competitiveness in the LLM industry because of abundant billion-dollar funded research alternatives and the ease of switching models via API keys.
@AndreyFradkin monopoly is not the right word, but there is a certain lack of diversity in research/investment approaches that a broader set of (well funded) competitors would bring. having said that, I have seen the list of 100+ neolabs, so maybe thats changing. time will tell.
@abhishekn @AndreyFradkin But surely this is the only industry in human history with a dozen $1B+-funded alternative mechanisms that are pure blue-sky research, no? That's totally unheard of in most fields.
@AndreyFradkin Definitely not a monopoly, but the barrier to entry at the frontier is enormous.
It would also be healthier for the industry if consumers saw open-weight models as a viable alternative. Not every task needs a trillion parameter model.
@AndreyFradkin monopolies usually mean you cant leave, with llms you swap one api key and youre on a different model by lunch
Ramez Naam quoted the post and Ben Golub reposted it.
Ben Golub#1451It's bizarre that some claim the LLM industry might be a monopoly. Based on what I've seen, this is one of the most competitive industries ever. Yes, the number of players is small, but the product improvement and leapfrogging is unprecedented. Of course, things may change in the future.