The big model companies will continue to hoard GPU and energy capacity for the next few years. This seems even inevitable post IPO as they will get access to retail investor cash.

With all this compute will open weight models, that currently boast of a 10-20x cost advantage over OpenAI, Anthropic, Gemini models be able to provide inference at competitive rates, even if they have 0 margins on the tokens? Wouldn't big labs over subsidize the large scale inference?

I think Chinese models would be the only ones that will survive the price war given state sponsored GPU clusters and energy access.

WDYT?


Comments URL: https://news.ycombinator.com/item?id=50531822

Points: 1

# Comments: 0