I want to hear a short story before we get into this meat of AI and data analytics discussion. The time frame is 1985- 1989. I know he looks really, really young, but the time frame is 1985 - 1989. Two guys go to school on either side of the same street. One, and you will guess who it is that we're talking about, and you will guess who the other person, importantly, equally importantly. One person studies computer science at the top-ranked college in India, which is impossibly, freakishly hard to get in. The other guy settles for maybe a top 10 school, but mechanical engineering because he cannot figure out this computer science stuff. This other guy, it's becoming very evident who this other guy is, tries to program on an IBM 360 mainframe, punch card, gets his syntax wrong in a Fortran program. The first time he ever tried to program, he said, "I'm never going to do anything to do with computer science ever in my life." In the meantime, this other person not only gets a degree in computer science, but goes on to get a PhD and goes on to run a company, a tech company. I guess by now you know who the two are. Very proud to call you somebody that I did not even know, but I never knew that somebody from the other side of the street. It's a small world. It's a small world. We have common friends. I just found out that we have some really, really good common friends. On that personal note out of the way, a warm welcome to Goldman here. Thank you. I think it's the first time we're doing this conversation together. Let's talk about what is your vision for Snowflake in the next four to five years. You've got a rich background. You're off to a great start. You're firing on all cylinders. The company is rejuvenated. It feels like it's breathing another dimension of life in its relevance, in its core, and the opportunities. Not to put you on the spot, where are you going with this in the next four to five years? Yeah. For the past 10 years, data has been at the center of many companies, but mostly in the context of how do we do it more efficiently? CIOs cared about it. All of a sudden, with the advent of AI, people are increasingly realizing that high-quality data is going to be at the center of how they transform their enterprise. You know, that's our aspiration, to help enterprises realize their full potential with data and AI. All companies start with a certain history. We came off history as an analytic platform. What we are doing, and it's an ongoing process, is to become an all-encompassing data platform from inception when data was first born to insights, that sort of feedback into how systems should operate. AI is both a consumption layer. You can get the information faster. It's also a massive accelerant of the value creation cycle. That's what we aspire to be. It's an exciting time to be at the center of data and AI. I joke to people that actual mainstream journalists ask me questions about things like Iceberg. I'm like, really? Open formats. I think it is a reflection of the times that we live in, how much AI is changing our world, and the role that data is going to play in driving that change. Got it. I wanted to ask you, you've had a rich background at IBM. You ran the ads business, SVP of Ads. What about that experience has informed you better to be able to run a company like Snowflake? I'd say two things. One is sort of an intuitive understanding of the power of data when it comes to creating great systems. Google was exceptionally lucky that it landed on a business model that at its core, both in search and in ads, was a feedback loop. Search, as all of you know, came of age with the PageRank, which we can think of as the feedback loop of popularity. You had a great page if a bunch of other great pages pointed to you. Similarly, with search ads, I would drive our advertisers crazy when they asked me, what should I put in my ad to make sure that people click on it and convert on my site? My genuine answer would be, I don't know, but if you put the right things, we'll make sure you show your ads because the feedback loop would pick that up. To me, that's everything that we did in aid of that. We built some amazing streaming systems back in 2005 because you needed that to support that kind of scale. It's very much infrastructure as an enabler of massive business outcome. That's an early part of my career. The latter parts of my career were then about how do you wield an actual incredibly large business through tons of change? The mobile change was terrifying for Google because query growth on desktop, which was the driver of our revenue increases, had pretty much flattened out by 2009. Things like the mobile revolution were still a twinkling sort of in our eye. It had not really exploded. How we made the transition, what it took to steer companies through very large internal transformations of their business was also a particularly profound lesson. It's the combination of these two, the power of technology to change the course of businesses combined with what does it take to run a large business and navigate through change moments that feel incredibly daunting. Mobile was terrifying because that was the only place where we saw growth. Mobile queries made one tenth the revenue that desktop queries did. It's the confidence that comes out of being able to navigate through changes like that. At Snowflake, the thing that I tell our customers, CEOs that I talk to, is we want to bring world-class technology in data that can let them compete on an even playing field with the giants, with the Googles and the Metas of the world. That's how easy we want to make our technology relevant and applicable to our enterprise customers, especially in this era of just massive, massive change. The core of Snowflake data analysis, the old world investors would say that's data warehousing. This is data warehousing in the cloud. I'm sure you have a different view and a different frame with which you view your market opportunity. How different is that frame with which you view the opportunity, and why is it so? Why is the conventional wisdom that it's a data warehousing company with a limited TAM in the cloud so wrong? Yeah, because platforms evolve over time. Yeah, yeah. What used to be, quote unquote, just data warehousing became an incredibly scalable analytic platform in the cloud that could also do machine learning so that you could begin to feed the value of that data back into systems. Disney, for example, uses us to optimize guest experience when people are visiting in their park, also from your data warehouse. Part of what we did was turn this data warehouse into a collaboration platform. Companies like Fidelity went away from doing literally hundreds of IT integrations, bringing in files via FTP and SFTP, as error-prone a process as possible. Collaboration comes out of the box and can deliver business value with a couple of screens as opposed to needing to run an IT project. It is the accretion of this functionality. More recently, we have expanded pretty significantly thanks to Iceberg and Snowpark into data engineering. All of a sudden, the power of Snowflake's core IP, which is a data platform, can now be applied to data that is outside Snowflake. To me, the value comes from the addition of all of these pieces where we are now beginning to add both data ingestion platforms, but also transactional support with things like Unistore and Postgres. On the other side, with Snowflake Intelligence, which is our agentic platform, some of the pieces are all of a sudden a whole lot more than the individual pieces. Think about it. A hyperscaler is a Kubernetes platform plus cloud storage and a bit of networking. Yet these are trillion-dollar companies. To me, it's that power with data at the center that we are able to tap. Yes, our origins, which by the way, we are not ashamed of, we are proud of, is that infinitely scalable data warehouse on the cloud. Many things can come out of it if you add the right things into it. We had Summit, your conference back in June. It felt like it was not a technology company conference in a good way. It felt something bigger. There was a bit of sensationalism in the air. It felt like an AI conference. You had Sam Altman. You had all these. My friend Sarah, who interviewed Sam and me afterwards, texted me and said, Sarah Guo, who runs Conviction, said, thank you for inviting me to your rock party. She's amazing. We had her on a panel a couple of years ago. She's fantastic, going to be a superstar, is already a superstar. Her husband's going to be here the day after tomorrow. Oh, brilliant. Yeah, yeah. They're big shareholders. Yeah. OK, good, good, good. Summit, coming back to Summit. Yeah. It's been three months since Summit. Yeah. As you reflect upon the products that were announced, as you seep through the customer conversations, what is coming back as one or two products that are, you know, that we really hit it and that's got a big future? Does anything become apparent to you? Yeah. I mean, first of all, we announced a lot. We announced lots of things at Summit. In many, many ways, the mentality that I've put with our product team is it's a culmination moment. It's not a try and cram everything into one point in time moment. I just feel like we are living in an era where planning for one or two dates in a year is just like not that smart. We are very iterative. In terms of products that show incredible promise, I would put Snowflake Intelligence right up on top. It's an agentic platform. Our sales force is internally at Snowflake. A good number of them are using it. We are rolling it out to everyone. In brief, what it does is on my phone, it gives me access to all of the sales information that we have, our customers, our prospects, how much they've been consuming, what kind of use cases they have active, and the account hierarchy, all of that information, even attainment information, is there in one place. It's all permissioned so that I see a view that's very different from what an account exec or an SE can see. To me, it's an indication of what the future world of data access and data manipulation is going to look like. Honestly, I can ask questions off of it that I would not have dreamed of doing even six months ago, or I'd have had to go to an analyst who would then have to work for a day or two to answer these questions. I think it has remarkable promise. We are in the process of scaling it, so I don't have great revenue numbers to report. That very much feels like a before and after moment in terms of what can you do with data that's in Snowflake. You had Cortex AI SQL that's for the technical user, Snowflake Intelligence for the business user. Yeah. Can you give us the most resonant use cases for each of these products? Not within Snowflake. You already talked about that. Yeah. When you talk to your customers, what are the best examples that you're hearing about how these two products are lighting up the account base? Yeah. AI SQL, for those of you that don't know, essentially introduced some AI primitives into SQL itself. When you think about summing up, let's say, revenue numbers by region to come up with an aggregate, you can also think about, let's say, taking customer feedback and organizing it by product category, but summarizing the top feedback using an AI aggregate function. It's super technical. On the other hand, what this lets people do is use the power of AI on huge volumes of data without needing to figure out things like, well, how much capacity do I need? How is that going to be configured? How do I handle failures? Instead of doing all of that stuff, we take care of all of that data processing for you. Customers 100% use that to do a lot of sentiment feedback on feedback that's coming from customers. We just make a whole lot of these kinds of use cases trivial. It's no longer some complex pipeline or process that you have to set up and run. Snowflake Intelligence, the kind of solutions that, again, resonate. BlackRock, for example, is creating a customer 360 with it. A lot of customers, BlackRock is one of them, have a substantial amount of data sets within Snowflake. Some are structured. Some are also unstructured. Something that will surprise you folks that are used to thinking about Snowflake as a structured data company is people routinely store customer feedback, customer conversations. AI companies actually store things like model responses, text. As Snowflake fields, we support these columns called variant types that can hold a huge amount of data. All of a sudden, you can get a single view with a thinking model deciding, should I be looking at feedback? Should I be looking at the current account balance? What am I, as a customer service person, what am I allowed to see? What am I not allowed to see? All of that stuff is taken care of. It's use cases like that that are resonating. Cambia Health Solutions, similar kind of product now built with clinician notes on top of health data. It is that one-stop shop access to a ton of information, context around it, and an agent loop that can decide which tool to call when. That's the resonance. What I tell people is I'm sure everyone in the room at this point has used things like ChatGPT or Gemini deep research. What I tell people, Snowflake Intelligence is, it is ChatGPT deep research with access to all of the data sets that matter to you. It's the same kind of agent loop-derived answers to a question. That drives up consumption. You find more use cases, more applications to use the platform, drives more consumption. That's right. That's a consequence. Yeah. I'm actually very proud of the consumption model in here because it removes a lot of angst from our customers. The first worry, especially with all of the articles flying left and right about 95% of projects not doing well, this, that, is what does this mean for how much I'm going to be spending? What I can confidently tell our customers is you don't spend unless people use the product and get value from it. If something that you build gets no consumption, then there's no money to pay. I think that's what is helpful. As a company, we also stress starting from customer value with consumption as a consequence rather than the other way. Let me see if I can try to ramp up consumption by introducing this particular product. I get that. Let's talk about data integration. You said it opens up a massive TAM on the most recent earnings conference call. You also made an acquisition of a company called Datavolo. When I went around the booths at Snowflake Summit, people are buzzing about your newly branded product. Yep. Tell us more about why this could be a big opportunity and how you go to market, because this is a separate product than the core engine, or maybe there is some adjacency. First of all, part of what Snowflake does is it creates an integrated product. This often ends up taking time for us, which irritates our sales teams and our product managers. We've gotten better at it. There's only one Snowflake SKU that comes with everything. It comes with AI, and it comes with OpenFlow. It's an important structural advantage. What OpenFlow enables is just this ability to be able to connect to different kinds of systems and bring data over to Snowflake or to cloud storage. It's a lot of connectors. We also have partners do this. We don't see this as an either-or, but many of our customers end up liking the fact that, again, it's a one-stop shop. There is not a new contract to sign or a new tool to figure out. 100% I think this makes it much easier to bring data on a periodic basis into Snowflake. From there starts data engineering. There can be analytic workloads that are built on top of it and obviously access via AI. We think of this as a very good addition. We also acquired this company called Crunchy. It's a Postgres database. The idea, again, is very simple. Lots of our customers want to build applications that host transactional data within Snowflake. We want to make it super painless for them to create these Postgres instances. Postgres has become the de facto standard for OLTP databases. We feel these just significantly expand our TAM while keeping the product pretty cohesive. Got it. I wanted to ask you one more product question and then go tspark o market. The Snowpark connector for Spark workloads, you kind of brushed over it on the earnings conference call and tried to get at it in the follow-up. Can you tell us, what can you tell us more about? It is something, this opportunity to run Spark workloads on Snowflake has been there for some time. Yep. Did you just formalize it through a hardened connector? There is a real opportunity ahead of it. Tell us more about what's ahead on that side. Yeah. I mean, we have always aspired to do data engineering workloads. In fact, it is a significant part of Snowflake's business, but it has also been very Snowflake-centric, meaning it was always step one, bring data into Snowflake and then do data engineering on top of it. What Iceberg, which is the interoperable format, unlocked for us is all of the data that is sitting on cloud storage that can now be acted upon by Snowflake. The other learning that we have had is that over time, de facto standards form, and Spark is one such standard for data processing. What Spark Connect does is it makes it super easy to run Spark jobs without needing to translate anything right inside Snowflake. Snowflake's performance as a data processing engine is the best that there is out there. This just makes it easier. It is also a little bit of us meeting our customers where they want to be. Let's face it, people do not want to run, like, do custom stuff to be able to run Spark code. This just makes it a whole lot easier. It unlocks more for us. It's very much getting started in this area. I think OpenFlow fully rolling out, Spark Connect fully rolling out is what is going to unlock data engineering in a very big way for us. Is this a different opportunity that has opened up so that you might need a different sales motion to go after these unmanaged Spark workloads, et c.? I mean, we have a. Because the proposition is slightly different than structured data. Yeah, 100%. We have a good formula now for how we take new products to market, which is we hire a small specialist team. They go create a set of the early win market use cases that show that we can get great things done. Then we figure out what is the scale motion. For AI, for example, we decided to actually have a bigger specialist team for special AI use cases. On the other hand, we also did enough enablement so that the broad field sales team can do many simple AI use cases. At the end of the day, it's not rocket science to be able to build a chatbot either on structured or on unstructured information, the simple ones. The more complicated use cases, yes, it's going to require the specialist folks. I think we are increasingly getting better at being flexible about what is needed to take a new product to market. As a little bit of applying this recipe, all of you folks that have dealt with enterprise companies know that specialist motions can take a life of their own. We want to be careful about how we do it. Mike Gannon, our new CRO, has a ton of experience for when do you spin something up and when do you drive it broadly across the field so that you don't end up with five overlays that are as large as your actual sales teams. We feel good about the motion. Got it. On that, it's a perfect transition to GTM. What have you unlocked on the go-to-market side with the hiring of Gannon? As you build your sites towards what at least we think is a $10+ billion revenue company, how do you see GTM changing? Product engineering is there. You've got all of a sudden in 18 months a flurry of new products. What needs to change or be enhanced on the go-to-market side that you can get to the going from $1 billion to $5 billion is hard. Very few do it. You're there. $5 billion to $10 billion, it's a different league. It's hard. $10 billion to $20 billion, even so. How do you go to $10 billion? What are your sites beyond $10 billion if you do have sites beyond $10 billion? Yeah. I mean, first of all, I think go-to-market has evolved a lot in the last 18 months. Mike's arrival is a welcome addition to the team. In terms of stuff that we've been working on, I've talked about how we are now a lot more quantitative about the consumption lifecycle. We track use cases pretty carefully. There's even more work to standardize what a use case is and how you measure incremental consumption from it. The core proposition is you can only optimize what you measure. This is something that all of us can relate to. We've gotten much better at that. To the level of sophistication of what's the difference between a 90th percentile account or sales rep and the 50th from the median. Another big important change that I fully, that Mike is also pushing, is the role of our Solution Engineers. We got this amazing person from Microsoft who's run large portions of their Solution Engineering team in Azure to run our team. Now our Solution Engineers. Didn't get an angry call from Satya after that. Sorry? You didn't get an angry call from Satya for speaking to me. No, thankfully not. Our solution engineering folks now have much more of—they're the leaders of consumption. In fact, part of what we have done is make the role of the Account Execs versus the Solution Engineers co-equal ones, where the Account Execs talk about things like deals and the earlier stages of the use case lifecycle, while the SE leaders step up and talk about how we are driving consumption, how they are driving go-lives. It's a big change forward with an intimate understanding of what does it mean for somebody to be productive week -on -week, month -on -month, quarter -on -quarter. I think it just gives us a lot more flexibility about where we invest. Similar to ads, my attitude is I'm just a portfolio manager. I'm just looking for the efficient frontier when it comes to figuring out where do I want to put sales headcount. We're putting a lot of it. We hired 800 people in the first half of this year just into that function. The second big change that Mike is busy pushing is a rebooted partnership approach. Most of Snowflake gets delivered via solution, like system integrator partners, GSIs definitely. They are undergoing a world of change with AI, and we think we have products that can let them demonstrate value a whole lot faster. We hired a new Head of Partners as well from AWS. That's a huge focus for Mike. I think these are the things combined with a specialist motion for taking new products to market. I think these are the key ingredients that will let us go from the $5 billion over to the $10 billion. We are very, very—first of all, we are early in the on-prem to cloud migration cycle, and AI has now given a powerful reason for every CIO to now tell their CEO that having great data, having data in Snowflake is what is going to drive transformation for your business. We feel like AI can be a big pull for how data is brought into Snowflake, and that's the thing that's going to drive us, first of all, faster to the $10 billion that we want to get to. We'll end up creating a much larger TAM as well that we will continue to aspire to. Got it. One other thing I wanted to ask you was I know that you spent quite a bit of time at a big technology company. You've had a fascinating chance to watch the foundation model battle, which seemed like it was a two-horse race, then became a three-horse race, then a four-horse race, and five, and six. Some people think it's a race to the top. It looks like a race to the bottom. More competition coming in, equal amounts of, not equal, but surprisingly how quickly it takes for somebody coming from behind to catch up. Why are these models all doing the same thing? How does it all end if you have a perspective? Where is the next value realization from this model? Wow. Where are we going as an industry? We've not seen much business return. What do you think makes for a good AI prognosticator? AI? Yeah. OK. It is to predict early and predict often. This stuff is just really, really hard. While it is the case that some folks like Grok have come from behind and magically caught up, there are plenty of other trillion-dollar companies that are trying and not quite making it. I think there is absolutely a little bit of a je ne sais quoi to who are the great AI companies. It's not all that easy to compete. I think the word is very much still to be written in terms of how this world is going to transpire. The other thing that I'll tell people is that honestly, yes, there's a lot of success with AI. If you think about what are the two superhits with AI, it is coding agents, and it's ChatGPT. It's consumer chat. Everything else is pretty small in the big scheme of things. My rough take is it's still pretty early. I think we will see the impact on our personal lives, on enterprises. It's going to take a few years. It's going to be gradual. My take is that so much technology has already been invented that if it truly permeates the world, say the way that mobile phones did in terms of the reach that they finally have, what, 6 billion-7 billion people in the world, I think it's actually going to be transformational for society. This is without taking into account things like AGI. In that sense, I'm very optimistic about how much value can be created with AI. I think it's still pretty early. If I were to bet, I would bet that it is not a unipolar or a bipolar world, that there are several people with great capabilities. Is that going to get commoditized down to zero? I don't think so either because it is truly, truly difficult to be at the cutting edge. It's more than money. I think that's what is going to keep some of these companies unique. You folks, again, know this already. OpenAI has run off with consumer attention. People are not going to change all that quickly over to a new app unless it is significantly better. I think there are some things that have absolutely been established that are going to be much harder to break down compared to others. Right on the heels of this presentation is going to be a venture capital panel. These folks, I'm going to be asking the same question. I've been doing that panel with these guys for about 10+ years. We're going to call it even better than the All In podcast because that's how high the quality of this venture capital panel is. It really is. If you have a couple of minutes, you should watch it. The other stat that was very interesting was 50% of your new customer wins in the quarter were attributed to the AI. They had an AI influence. Absolutely. Tell us more about that factor. I mean, look, every customer that's betting on Snowflake is betting on the next 10 years. It's already very clear that AI is a big part of whatever it is that's going to show up in the next five years. This is where our ability to make AI simple for our reps to be able to say, let me show you what is possible with data on Snowflake becomes such a big deal. It also points to the importance that AI is going to have in the future. Another stat that we released as part of earnings was that something like a quarter of deployed use cases have some element of AI in them. I think this points to both the ease of use that Snowflake AI has, but its increasing importance for the entirety of the data lifecycle. Got it. Two minutes. Anybody has any question? Standing room only. This is so cool. Yeah, in the biggest ball zone. Anybody? OK. Maybe. Stun the room with your brilliance. No, you have a question for me. Let's turn it around. What is your prediction for software? Software is not dead, first of all. I think there is a view that maybe I'm a tired old. Despite the fact that I could not execute my Fortran program on an IBM 360 mainframe back in college when you were blazing new trails just a mile away from me, I do believe that we can confuse the user interface and how attractive AI makes it to be to visualize a complete disruption of the software stack. I think what's going on is when Netscape went public and four years after that, the web browser became the new fascination, the front end. The enterprise software industry used the web browser as the front end to revisualize the way in which end users interacted with the software. The backend logic did not necessarily change. The backend logic, the logic of doing business is the logic of doing business that's expressed in code. What it did do was to help you rip out the user interaction model. In the same way, I think AI is the new UI, it does not change the logic. Certain things don't need you don't fix things that are not broken. We know what's broken, the engagement model, the front end. I think many of us confuse, not me, not you, but many confuse the lack of usability or the complexity of the user interface to be an inherent flaw with the software. I would beg to disagree. I'm extremely optimistic about how so now, there are a lot of cross-currents. Yeah. You guys have emerged from this period of declining NER. Now you finally hit stability and started to see improvement. The same thing needs to happen to the rest of software cohorts. If I have to say, the software prints actually all, most of them looked better than expected and showed some sequential acceleration. I am very, very bullish. On that note, let's give a round of applause for Sridhar Ramaswamy. Thank you. Thank you so much. Yeah. Thank you.