00:08:57 · Aditya Agarwal
[00:08:57] S1: Yeah. It's interesting. Some of the kind of like at least at SPC when we are talking to potential founders, probably some of the most determinative kind of characteristics are just what we call internally clarity, which is that for a given problem, are they able to essentially... Like, they can spend thirty minutes walking you through all of the logic and the detail. But if you just ask them ultimately what are the one or two things that matter for your business in the next six months and the next eighteen months, they often have very short succinct answers which is like, this is just kind of the one or two things that matter. If this happens, then I think we have a shot to kind of take it to the next level. If it doesn't, we're kind of like, we'll have to zig and zag but that's a pretty bad sign, right? They have a lot of clarity. And in some ways, it's almost not emotionless, but they're very passionate about it. But they also just have... These are the two things. I wake up and I think about it and this is what I'm trying to do, especially in the early days. To use this framing, and you already talked about Neeva a little bit. Left Google after fifteen years to go. What are the one or two things that were going through your head that you wanted to tackle with Neeva? Because part of where I'm going with this, Sridhar, is it that you had the right idea but perhaps one year too early, two years too early? That's actually a hard way of thinking about the world, kind of like to kind of play the counterfactual back.
00:10:17 · Sridhar Ramaswamy
[00:10:17] S0: Yeah. I mean, to be brutally honest, I think that first of all, the idea was born abstractly that a search ads, I mean, an ad supported search model had reached sort of its useful limits. And there's nothing wrong with that conclusion. If you're not bound by ads, like obviously you can do a better job. I think the real sin of Neeva and my own arrogance in it was a lack of a clear roadmap on, okay, how? Like what must happen for there to be a much better search engine? It's not something that I should have figured out by building. It's too big a thing to build if you can conceptualize in the first place. It took the arrival of GPT-three nearly three years later for us to realize, ah, I want to click on a link, said no one. And so that's like the brilliant insight for what can AI powered by a search engine do because you've removed the click. You give the person what they are looking for in one shot. I think the lack of that clarity, there's a lot of detail about how... And sometimes you can draft onto a wave that comes that you are not quite expecting. That's all the, you know, you know, it's it's better to be lucky kind of story. But the fundamental sin of
00:11:52 · Aditya Agarwal, Sridhar Ramaswamy
[00:11:52] S1: Neeva was that lack of clarity for, okay, new search engine, but how? You know, arguably, the people who built GPT-3.5 and ChatGPT didn't really know what they had, right, until it kind of came out and we started using it and everyone was like, oh, this is actually the right way to kind of do everything from like Internet search to commerce and so on. So sometimes you also luck but also just serendipity. Just kind of get the stuff out there 100%. And then you see it takes on a life of its own. [00:12:24] S0: I'm not first of all, serendipity, absolutely. But execution still matters. Sure. If you go back and look at early twenty twenty three, how many people here remember the we will make Google Dance speech? [00:12:38] S1: Oh, yeah. Yeah. Yeah. Yeah. This is from Right? [00:12:42] S0: This is from Satya? [00:12:43] S1: Yeah. Yeah.
00:12:43 · Aditya Agarwal, Sridhar Ramaswamy
[00:12:43] S1: Yeah. Yeah. [00:12:44] S0: Yeah. The opportunity was there for the taking. I'm sure most of you don't remember Sydney, but Sydney was way better than ChatGPT. ChatGPT would like spew out trash through much of 2023 anytime you asked a factual question. If you ask for a high query to do a great job, if you ask for a fact, it'll just like happily make it up. Sydney Bing chatbot had solved all of this. It had this fatal flaw that every interaction with it would take two minutes. They never fixed it. These are the people that are supplying capacity to open AI. They didn't see the opportunity. Bing itself makes 10 plus billion dollars. So they had a lot to lose by pushing on Sydney as the definitive search experience. But these are the people that actually created a much better search, not OpenAI, and they still blew it. So [00:13:34] S1: It's hard building new products. It's really hard. Before we kind of get into Snowflake a little bit, let's talk. I mean, I'm curious to get your take right now. We are... Is this kind of an interesting time to be an open source kind of like model right now? We obviously had GLM five two come out, which is getting great reviews. And in general, it feels as though that, yes, GPT 5.5 is exciting. Fable, whatever the... Whatever it is, you know, if I could use it, is sounds pretty cool. [00:14:02] S0: There's a line of thinking, by the way, that says that all model progress is just dependent on underlying hardware. And Fable and 5.6 are merely B200 models.
00:14:02 · Sridhar Ramaswamy, Aditya Agarwal
[00:14:02] S0: There's a line of thinking, by the way, that says that all model progress is just dependent on underlying hardware. And Fable and 5.6 are merely B200 models. [00:14:13] S1: Yeah, yeah, yeah. Know. So I guess my question coming to you is, what is your current take on the model ecosystem? Kind of like the leading lab providers like obviously Gemini, OpenAI sorry Gemini, ChatGPT, Claude, and then also obviously the six or seven kind of big open source models. Like how do you think about navigating in that world from your current position as the CEO of Snowflake? [00:14:38] S0: The Snowflake CEO, I mean, first of all, these have been recurring cycles of like frontier model labs, creating models that are significantly better along some variety of dimensions on the open source models- Catching up, yep. Catching up. I think, by the way, what is remarkable is that there's just a small handful of companies that can truly create frontier model. Still blows my mind. [00:15:09] S1: Still kind of crazy.
00:15:09 · Aditya Agarwal, Sridhar Ramaswamy
[00:15:09] S1: Still kind of crazy. [00:15:09] S0: Yeah. Yeah, I'm kind of like, what money can't buy stuff? No, it can't. I still find that just wild. I think definitely what is happening with the power of the Frontier Labs is that at least with lots of enterprise customers, there's a little bit of a worry of like, wait, all of us often you see like Satya had both a tweet and I think an article, Wall Street Journal, on this, where if you look at a world in which kind of all software disappears and these models are the interface between pretty much all of humanity and data underneath, let's say, it's a very different kind of world. [00:15:50] S1: Correct. [00:15:51] S0: So there's an amount of trepidation about being too reliant on a single model source. [00:15:57] S1: Right. Which is really kind of an intrinsically a trepidation about centralization. 100%. Yeah.