Name: The $100 Billion Data Question
Uploaded: 2025-12-10
Duration: 15 min 59 s
Description: Can one platform unlock the true business value of AI? Databricks CEO Ali Ghodsi believes so. He discusses the company’s ascent to its mammoth $100 billion valuation, the urgent need for enterprises to turn information into insight, and what the future holds for one of the largest companies in the private markets.

Can one platform unlock the true business value of AI? Databricks CEO Ali Ghodsi believes so. He discusses the company’s ascent to its mammoth $100 billion valuation, the urgent need for enterprises to turn information into insight, and what the future holds for one of the largest companies in the private markets.

The $100 Billion Data Question

Conferences

Autonomous and chaotic, wasn't that great? Don't you love the juxtaposition? Ali, thank you for joining us. Of course, my pleasure. All right, we've got a lot to rip through, so let's do it, my friend. Um, Data bricks started as a way, if correct me if I'm wrong, uh, as a way to write code and run it on a distributed computer. Now it's a unified analytics platform. You've got data warehouses, data lakes, data zeppelins. I don't know, um, you let engineering teams build and deploy their machine learning and artificial intelligence apps, and you prep their data for it. Is that a good assessment of what data bricks is? Yeah, we got all the buzzwords in there. We got them all. Yeah, did I say big data yet? OK, good, um, so what do companies need more? What do your customers need more? Do they need the clean data side of, of that offering, or do they need the custom tools? They need security and governance. That's like the number one thing that's holding the big enterprises. Like if your company is more than 10 years old, the big thing holding you back is that you can't actually do anything because you're so worried about getting hacked. Who's gonna get access to the data? Is the AI gonna suck up the data and memorize it and leak it? You know, there's lawyers now taking over. It's like AI lawyers now in every organization. That's everywhere. So when you say AI lawyers, lawyers who specialize in AI or actual artificial intelligence, yeah, yeah, no, actually they specialize in artificial intelligence, and they like tell you that, hey, here's the regulations. Here's which models we can use. Here's what's our policy. Here's what company contracts with our customers say, so we're not actually allowed to use this model. But if we do, we have to use it in this way. We have to have guard, and they enter all the and everything slows down in all the organizations, right? They enter the rooms and that's, so how do you do governance? How do you, and there's regulation popping up around AI all over the planet, right. And you know, so that's, so for you it's, it's corporate inertia, yeah, the biggest blocker, yeah, absolutely. And for those companies that's what I mean, you know, they're like, hey, this disruption with AI is coming, but you know we're being slowed down. Can you help me speed up? What do I need to speed up? Yeah, yeah. Now a lot of companies are looking at AI agents as a way to do that. You guys have expanded into that area, right? It's Agent Bricks, Agent Bricks. And so what are you seeing in that arena? We've been talking about it quite a bit in the last. 36 hours or so, um, what are, what are, uh, your customers trying to do with agents, you know, where are we, where are we now and where are we going? Well, let me just segue into it from what I was just saying. Your agent is just gonna go access workday and have access to HR data where salary information and so on is. How do you like lock that down? How do you make sure like suddenly it's not leaking information or going into wrong? Systems and accessing stuff like just this authentication security who's allowed to do what is uh is a huge ordeal inside any organization so we do that that's unity catalog. But what are people doing with the agents um there are lots of cool, uh, use cases, you know. I think the marketing stack is being marketing inside the organization is completely being disrupted. Um, so take 7-Eleven is a customer of ours, and what they do is they just use these agents to prepare all the material that before was taking huge amounts of time to prepare to target different groups, you know, what should I show you? What are you interested in? What's your campaign? That's now being largely automated and the content is being auto generated. The groups are being segmented, so that marketing stack is, you know, it's, it's, it's a big and actually at theater as well, a big disruption coming to marketing itself, a little bit of dog fooding too. We're gonna actually talk about that in one of the lunch round round round tables today, um, so. Everything in this category is about fast growth. Your company, uh, the technologies that you're dealing with, the, the, the com the company is trying to change rapidly, right? Um, are you, for your business, concerned about the rising costs of AI eroding your margins? No, the way I think about it is that, uh, you know, it's just so valuable, these use cases, and you know we can so easily see the ROI for our customers that I think actually we've seen that we can actually charge quite a bit for this. I think it depends on which layer of the stack you're in. I think the foundation model layer seems very competitive and the margins are very low or if you're in the business of like renting GPUs and reselling them, you know, that's hyper competitive, uh, but I think this, the layer on top where you're actually having agents. Do useful work, um, I don't think actually it's gonna be a big problem with revenue and monetization. I think the big problem is one, security we already talked about that. The second is quality, like, you know, are the agents good enough to do work, uh, inside of these organizations? And I think, you know, generally speaking in the industry, the answer is right now kind of no, even though like I mentioned one disruption with marketing. There are other disruptions I'm happy to talk about, but, uh, generally speaking, it's not the case that inside of our companies we have lots of co-worker agents running around that we work with all day long. You probably have like 0 coworkers that are agents. Anybody have any coworkers who are agents here? Raise your hand. 00, we got a hand. We got a couple, not like human agents, but we're talking about AI agents here. Yeah, alright, we got a couple of hands. This isn't, I mean, brainstorm AI. Hey, I'm saying the disruption is happening, yeah, yeah, in different verticals. Yeah, um, so let's talk more squarely about your company. So you, you have raised, as Ali said, uh, in introducing us, you've raised billions of dollars from Amazon, Microsoft, Nvidia. I've heard of these companies, um. Your evaluation right now stands at about $100 billion you know, for just to like ground that in reality, um, ExxonMobil, you know, that company is worth about half that in the public markets right now. So just like that's how big that number is, um, and they're number 8 on the Fortune 500, by the way, like big, um, be number one forever, right, right, that's right, um, you've read your Fortune magazines, cheers, um. You're flirting with an IPO. Why do you need an IPO and when? Yeah, I mean, um, why do we need, uh, uh, an IPO? It's a good question. I mean, it's like, you know, I would just say access to more capital, so I'm not sure we need that much capital. There's just much more access to capital in public markets. Uh, the second thing I would say is that. It is. I do think for American companies like in the United States, uh, we're global, so we sell in all the, all over the world, uh, there is a marketing boost to companies that are public. There's like free marketing every 90 days that they get, uh, with earnings call and everything, um, so that would be another thing, um, yeah, uh, and then, you know, maybe a little bit easier liquidity for employees. And investors, um, but it's close. It's not like, wow, there's like huge advantages and pros and cons. That's why we're not super religious about it. We will go public at some point, but to us it's like not a really big deal. I know everybody's super curious and interested about it, but for us it's like, uh, you know, it's close enough that, you know, we might do it, might not do it right now, so next year maybe. Depends actually, I'll be honest, the way I think about it is that, yeah, the way I think about it is that if, if there is a bubble bursting, which there might not be, uh, then it would be great to be private because we can continue investing because one of the biggest advantages Databs had was that in 2021, most of my peers, CEOs, they're like, we gotta IPO our company, we're gonna go public, and I told them like, hey, maybe we should hold off. It's like this whole situation is kind of out of whack, and those companies went public in 2021 and in 2022. You know, inflation, interest rates, are you suggesting the markets have been volatile? Yeah, exactly. And what happened is that they were like in cost cutting mode, layoffs, and that year we hired thousands of people and we continued investing, and now we have, you know, doubled the growth rate of some of those companies. Uh, so I think it's, you know, just having like a long term vision and just following that, I think AI is gonna be super useful to society. So I just wanna do that over long periods of time rather than having to follow like market hype. Yeah, I think, I think there's quite a few folks in the room who might agree with you on that one. so let's talk about some of the partnerships you've made in the interim. Um, you did a deal with Thomas at Google Cloud in June, who was here yesterday. Um, you did a deal with Brad at OpenAI who was here earlier this morning in September. Um, you know, I asked, uh, Corewave CEO Mike earlier today about the circular, uh, AI economy just briefly because we're running out of time. I will ask you the same thing. He believes it's a fundamentally flawed thing. It's not the right way to describe it. You're doing those deals yourself. How do you feel about that term? Well, let's separate something. There's not like, literally the, there's only like the arrow just goes in one direction. There's no, there's no arrow back. Like we're like, so there's a separate. It's like there's, there's deals a linear. There's, it's there's deals out there that are like, you know, this vendor invests, let's say 100 billion in this company and that vendor buys, say, products. From the same vendor back for 100 billion, there's nothing like that going on here. What Databix is doing is basically saying that, uh, you know, if you commit to Dabi as a customer and you, we already have 70% of Fortune 500 doing that, you get open AI, anthropic, and Gemini models in the platform out of the box because there's such a tight race between these, you know, now this one is better, or now this one is better at coding. No, this one is a better chatbot. No, this one has agents literally every week. Yeah, every week, yes, so we're just saying don't worry about it. Uh, commit to us and we make sure that you're covered and you have all of them. You don't need to even worry about getting contracts or anything, and we have all of those. We also have all the open source models, including even the Chinese ones. So thanks to you, you need to catalog our governance layer. Some organizations decide to lock that down and say we don't want to have a Chinese models, so then they can do that. So that's the, that's it's just one direction. There is no, like, you know, in return we'll buy. You know, tons of tons of something else from them or something. There's no, no, no such thing going on, understood. Um, I wanted, yeah, I can, yeah, go ahead. I can bite on the, on the, on the circular thing though if you want, please, please, by the way, just to make it interesting, um, I, I think that those circular things are just getting started now, so the inflation only like the, you know, the blowing air into the bubble is gonna start now. Is there a bubble in your opinion? I, I think there are, there is a huge bubble in many, many portions of the market. There's like, well, I mean the companies that are worth, you know, billions of dollars with zero revenue, that's clearly a bubble, right? And it's like insane. Like if you talk to VCs and investors, I mean they all know it. They're like this is like maybe I should just go on a break for like 6 months and come back and uh it'll be like really financially good for me. That's what the VCs will tell you. Uh, so there's definitely a bubble there, but I think this, the circular. Uh, aspect of it is just gonna this inflation of that is starting now, so I think like 12 months from now it'll be much, much, much, much worse. And I actually think it's sound that the markets are kind of reacting to that, right? They're wobbly and they're not growing. I think that's good because it's kind of telling the CEOs out there that, hey, maybe, you know, you should, uh, take a step back. Interesting. I just heard yesterday somebody saying in a year we're not even gonna be talking about AI bubble, so we'll see. We'll be back here in a year. Don't worry, um, OK, so, uh, uh, one more question and then I would like to come to the audience for them, um. You've said data bricks in the past data bricks could be a $1 trillion company. Um, we only got the first one of those, uh, a few, a few years ago and it was Apple in 2018, something like that, um, and that's the public markets, right? I mean what do you need to do to build, buy, partner. To get there, yeah, so you described very well in the beginning with all those words that you threw out of what databix does. I think the word's good. Yeah, you're very good with that, but you know there's a couple of new vectors that we're getting into that are completely new for us. One is transactional database market. This is the The traditional Oracle market, uh, that market has been the same for 40 years. So for us, the lake-based offering that we, we, we launched just this summer, that's a very, very big bet for us. Um, why is that interesting? Because for the first time we're seeing over 80% of the databases that are being launched on data bricks are not being launched by humans but by AI agents. This is actual stats. But we might ask them, well, what is this? Like we're already. Agents, it's people doing vibe coding, so lots of people are writing software. They're vibe coding stuff. Well, all that software needs a database behind the scenes, and the software prefers to use Lakebase. It's really built for that, so it's best at that. So that's just a huge growth vector for us, I think. If we just did that, we could maybe get all the way to 1 trillion. The second one that's really important for us is Agent Bricks. Yes, and Agent Bricks is important because Again, the real opportunity is, you know, it's a commodity now to have AI that has general knowledge. Ask it any question you want, as you said, the race is like going, it's a commodity. You can get free open source models, but something business specific is, but it's very elusive to get AI that really rocks and understands. That proprietary data that's inside the enterprise. So we can actually unlock the value of that and start automating more and more and more and augmenting the work that's happening inside the enterprise. So I give you like the 7-Eleven example with marketing. Here's another interesting one like in healthcare, Merck built a model model, AI model called Teddy. Uh, and it's a generative AI model, so it's, you know, stands for transformer enabled drug discovery, and this model is really have amazing understanding of cell behavior, and it can really differentiate between causal cells and reactive cells. So it has the opportunity. There's a lot of value there, yeah, it can really revolutionize drug discovery. uh, so that's another one. Another favorite of mine is RBC, uh, Royal Bank of Canada. Uh, this use case is equity research analysts. They do all this. Research on like oh there's an earnings call here there's an earnings call there, but what they've basically done is they now with agent bricks they're building these agents that get the earnings calls, get all the information that's out there about that company, everything that's going on, put together the research report and uh just augment significantly, get, you know, many days' worth of work down to like again yeah if we can just keep unlocking uh sort of uh agentic sort of use cases like that, I think it's going to be that that that. Can take us all back. That's your ticket to the 3rd, OK, to get the trifecta is also to have apps on top of this. All right, now you have apps that are vibe coded with the database lakebase and with agents. Um, OK, those are 3 new vectors for us. All right, we've got a plan. OK. There's a question right back here. Yes, please, your name and who you're with briefly. Steven W Preda, CEO of Alpha, it's a govern, AI governance intelligence company. I was at the summit in June's great job with, uh, Jamie Dimond, um, but you're hitting on a point and you kind of discussed it a bit with Jamie at the time. The governance issue, especially in the boardroom. And so you really see this kind of push pull between focus on all the risk, but they're really not talking about the growth opportunities. How are you actually seeing boards really kind of challenge the governance issues because you're obviously very important to that whole conversation. Yeah, that's a great question. I mean, I think that, you know, uh, mostly it's actually problematic because the boards are just pounding the table and saying, hey, like, where is the AI revolution? Like, you know, you're spending all this money, where are the results? And they're kind of frustrated and then the results are not coming fast enough. The reason the results are not coming fast enough is that there is a legal and a security aspect to this, uh, you know, and that, so that's what we talked about in the catalog where, where we work on that and there's new legislation coming and all of that. But then separate, even if you can figure those things out. The data architecture inside most organizations is an absolute mess because they've been buying software for the last 30, 40 years from different vendors and just piled it on. A lot of the data is siloed on premises in data centers still it's not in the cloud, and then in the cloud it's in different clouds. So just getting, you know, their ducks in a row there is also, you know, problematic. So most organizations are frustrated. There's a big gap between the CEO and the leadership and the board who's saying let's go, go, go, go. Why are we like AI revolution is already here. And then the reality on the ground, which is like, well, you know, give me another year and give me more head count and I need to spend way more to even get there, uh, so that's what we have to resolve and there are ways to actually solve this, um, you know, 11 of, one of the things I like to say is like, you know, I think the process change management with humans is actually the biggest problem, not the tech, of course, yeah, so I think the biggest issue right now is that now inside of the organizations there's like. Many people who say, no, I should lead data and AI strategy for this company. I'm like those guys over there, they don't understand anything. So there's like this tussle between different people who are fighting over, you know, becoming that AI person, and that's also making things worse. So like my advice to leaders there is, can you like pick one person for your company? Don't have a three-headed, you know, monkey. Doing that, it's always the humans. Thank you for your question, Ali. We're out of time. Thank you so much for joining us at Fortune Brainstorm AI. Thank you, sir. Thank you.