00:35:07 · Guillaume Moubeche, Ron Gabrisko
[00:35:07] S1: Okay. So you're keeping like the the very like open source mentality and you know what you pay for and... Okay. [00:35:15] S0: Totally. Okay. Totally. And I think... And that's that's, you know, been really good for us. Right? Because people wanna know, like, you know, transparency. And that... That's helpful too when you go in for renewals, because, like, when you go in for renewal, if we did a three year deal, they wanna know what kind of value did I get out of the last three years. You know, you'd... You need to be able to show them the actual value, And so I think the value based pricing starts to fall apart once you try to go for those renewals because you've already built it. Right? You have to go after new projects. So... [00:35:51] S1: Yeah. I agree. And I think in any case, like, usage based, to be honest, is is so obvious. Because if you look at... And I think that's also what happened recently, you know, like a lot of companies... Or even when when I think Elon Musk, you know, went to the the government and started looking at what people were paying, and they realized that, you know, they had... I don't know I don't know how many licenses of Adobe it was that were basically just unused. It was a seat based pricing, and Yeah. No one was using it. So, yeah, it makes a it makes a lot of sense.
00:35:51 · Guillaume Moubeche, Ron Gabrisko
[00:35:51] S1: Yeah. I agree. And I think in any case, like, usage based, to be honest, is is so obvious. Because if you look at... And I think that's also what happened recently, you know, like a lot of companies... Or even when when I think Elon Musk, you know, went to the the government and started looking at what people were paying, and they realized that, you know, they had... I don't know I don't know how many licenses of Adobe it was that were basically just unused. It was a seat based pricing, and Yeah. No one was using it. So, yeah, it makes a it makes a lot of sense. [00:36:27] S0: Yeah. Totally. Totally. I... Yeah. I think the... I mean, even the companies that were traditionally seat based pricing like Salesforce, they're trying to move to consumption based models. I mean, you have to just, you know, you run out of seats, especially in the, you know, in the new world. So... [00:36:46] S1: In what way do you think, like, or like, I'm gonna rephrase my question. In five years from now, because you've raised now at like a $134,000,000,000 valuation, like, five years from now, like, what do you think Databricks look like?
00:36:46 · Guillaume Moubeche, Ron Gabrisko
[00:36:46] S1: In what way do you think, like, or like, I'm gonna rephrase my question. In five years from now, because you've raised now at like a $134,000,000,000 valuation, like, five years from now, like, what do you think Databricks look like? [00:37:00] S0: I'm here to to help build. I think it'll be one of those generational kind of transformational companies. I don't think enterprise AI is valuable without your proprietary data, and I think Databricks is the most trusted, you know, product and company to, you know, manage your data and connect it to AI. So, I really like the space we're in, I think it's... I think it will, you know, touch every human and change every company, and that's our goal, you know. You know, with Genie, we can expand every single user in a company. We've released Databricks One and Databricks Mobile, so now you can get it on your phone and, you know, get all your... Ask all your questions to your data while you're, you know, traveling or whatever. So it's it's pretty cool. I think, again, I think Databricks will be an absolutely iconic company, so, you know, we're on it.
00:38:02 · Guillaume Moubeche
[00:38:02] S1: What... What's your feeling with with the the kind of, like, chat usage? Because I feel like this is actually quite new, you know, for a long time, if you look at how software were designed, it's mostly like you have like a mouse, you click on buttons, you do actions, and this is how like things work. But the the chat interface, I think it's it's not something like that people have been... I feel using that much to get answers from software. It kind of like started with chat GPT. And what's your view about it? Because I'm assuming that you also have tons of data on how much query people are making. Do you feel like for for data, the the chat query is usually the best, or do you try to also give like IDs to people and be a little bit more proactive in a sense, you know, like sometimes Anthropic does it, where they want to refine a bit what you are saying, and they would propose like a few set of options so you can dig in a little bit more.
00:39:13 · Ron Gabrisko
[00:39:13] S0: Yeah. I mean, g... So a couple things, like one, like Genie code will actually help you write data engineering and data science, like models and workflows. And... Right? So we're starting to automate, some of that to help out data scientists and data engineers. It just makes them way more productive, you know, for our business. I think the other thing that's happening with a lot of these, vibe coding tools is I think every single app in the world will get recreated. It'll be much more custom to what they want. I mean, we did that with our own... You know, I run our business on Databricks. We call it Databricks on Databricks. Like, I can predict our revenue within a couple percentage. Right? I can see which customers might churn, what products they're using, like, so I think every app will get recreated, and it'll get recreated with vibe coding. You wanna have AI, you wanna have predictions, you wanna have agents in there, and so that's a big part of why we release Lakebase. Like, Lakebase, if you look at the old databases, like, every single app needs a database. Every single one of these vibe coding needs a database. You know, we bought Neon as a part of Lakebase, which is really an AI database. Database. Like, the next generation of databases need to be able to handle massive volumes from AI and agents, and be able to, you know, set them up and tear them down and branch them immediately. And so, you know, we think Lakebase will be the foundation of all the new apps that people, you know, use vibe coding to build. And so that's again another huge transformation for every one of these companies instead of using, say, out of the box software.
00:41:01 · Guillaume Moubeche, Ron Gabrisko
[00:41:01] S1: Can you can you explain like... Yeah. Sorry. Can can you explain like why you think like you're gonna need the new type of databases with with AI? Like, what does that change from a conceptual perspective? [00:41:16] S0: Yeah. The biggest part is we separate storage and compute. So similar to what we did in the analytics space with the Lakehouse. Right? Like, if you have to expand storage and compute at the same time, it makes your Databricks super rigid. Right? And so a lot of these AI and agent based applications, they wanna be able to spin up a database in, you know, milliseconds and then shut it down. So you don't wanna have to pay for it when it's not running. Right? Because, you know, think about it, like in the future, you have millions of agents hitting these, these applications. Right? It's gonna be, you know, today applications, you know, are for humans. There's so many humans that are using it. Now it's gonna be agents. It's gonna be 10 x, a 100 x. And so the difference of separating compute and storage is super, super important. And so, you know, if you think about our Lakehouse concept, where it's like take all your, you know, it was all your data for your data lake and your data warehouse, put in an open format so you can do all your analytics on one set of data. We're gonna do the same thing with Lakebase for databases. Right? And so, I think it's a huge transformational shift in the database market, and that's what's needed for AI and agents.