Alessio Fanelli: evaluation

Transcript context

…Yeah. So, um, my comment on, on the business model shift, this is super interesting. Apparently it is wide knowledge that OpenAI wanted more than 6. 6 billion dollars for their fundraise. They wanted to raise, you know, higher, and they did not. And what that means is basically like, it's very convenient that we're not getting GPT 5, which would have been a larger pre train. We should have a lot of upfront money. And [00:17:00] instead we're, we're converting fixed costs into variable costs, right. And passing it on effectively to the customer. And it's so much easier to take margin there because you can directly attribute it to like, Oh, you're using this more. Therefore you, you pay more of the cost and I'll just slap a margin in there. So like that lets you control your growth margin and like tie your. Your spend, or your sort of inference spend, accordingly. And it's just really interesting to, that this change in the sort of inference paradigm has arrived exactly at the same time that the funding environment for pre training is effectively drying up, kind of. I feel like maybe the VCs are very in tune with research anyway, so like, they would have noticed this, but, um, it's just interesting. Yeah, and I was looking back at our yearly recap of last year. Yeah. And the big thing was like the mixed trial price fights, you know, and I think now it's almost like there's nowhere to go, like, you know, Gemini Flash is like basically giving it away for free. So I think this is a good way for the labs to generate more revenue and pass down [00:18:00] some of the compute to the customer. I think they're going to keep going. I think that 2, will come.…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.