If I were an investor, I'd be pretty concerned with such a high valuation after the o1 release. It's great, no question, but in my usage so far it's a modest step up from 4o, much smaller than the 3->4 jump. Real world exponential growth is exponential until it's logistic, and this sort of feels like entering that phase of the LLM paradigm.

Talking to friends who are very successful, knowledgeable AI researchers in industry and academia, their big takeaway from o1 is that the scaling hypothesis appears to be nearing its end, and that this is probably the motivation for trading additional training-time compute for additional inference-time compute.

So where does that leave an investor's calculus? Is there evidence OpenAI can pull another rabbit or two out of its hat and also translate that into a profitable business? Seems like a shaky bet right now.