Hey everyone, it is Monday, February 9th. This is another Tech Check live stream. I'm Deirdre Bosa. Thank you so much for those patiently waiting and sending in your questions and comments as we get started. Now, we continue to cover this major shift in AI over the last few weeks from chatbots to agents that actually do things like build apps and create software. You know, Jasmine and I have been doing it as well. And every week, we get more evidence that this is real and accelerating. Databricks just gave us one of the clearest data points yet. 80% of databases on their platform are now being built by AI agents, not people. And it's not just tech companies, it's non-tech like Mercedes, AT&T, Mastercard. More enterprise software is being created in-house than ever before. And AI is doing the building. So Ali Ghodsi, he's the CEO of Databricks and he has a front row seat to all of this coding wars, model race. What 20,000 companies, its customers are actually doing with AI right now. Let's get into it. Ali, you just jumped off TV. You can do it all. Thank you for jumping right on the live stream. Um, you left me in suspense on TV. You basically said that you're preparing, you raised all this money because you're worried about another 2022 situation in tech. And by the way, that is the year that the public tech trade just got wiped out. What are you seeing?
Well, I mean, it's just, you know, things seem to have just been accelerating on the AI side. It was all these deals, you know, that were like kind of circular in nature. People were very worried about it. So when we started our fundraise, that was the number one question we were getting: what's going on with AI? It's like, do you think these deals are okay? So we were getting a lot of questions about that and it had nothing to do with us. And then towards the end, people started waking up to like, oh wait a minute, if agents is the new interface, then the system of record maybe is no longer really needed, or is it needed? Are they going to switch to something else? Will there be a new system of record? What's going to happen to those companies? And then started focusing on, you know, what's the efficiency of those SaaS companies. So it's just, you know, when you raise so much capital like $7 billion that we just did now, you get to hear what these top investors in the world sort of really behind closed doors are worried about. And it's just there was too many questions about these kind of things that got us worried. So we said, hey, we better get really well capitalized now just in case. And you know, kudos to my board also which also pushed us like, hey, we got to capitalize now because who knows what's going to happen. So that was it. I'm not saying definitely it's going to happen but there were kind of a little bit 2000 vibes around it. So, we just wanted to, you know, be safe.
Right. And it felt like that started sort of late last year, right? All of these bubble questions were coming to the surface. And you, Ali, the reason I like talking to you so much is you've always been sort of like clear-eyed. It's never all good, never all bad either. But what are you seeing? Because you have like a front row seat. You see everything that's happening with your customers. You have such a good view of AI as a whole. Are you becoming more concerned? You mentioned the circular funding, valuations. How do you feel in this moment?
Yeah, I mean I think that what happened is that we had these amazing consumer breakthroughs with ChatGPT and then we saw coding happened but then we didn't see much else happening for a while. But then everything kept getting bigger and more and more headlines. So that was the worry. What I'm seeing right now is that we're finally starting to see that the knowledge worker is starting to use these things. As I said, we launched this Genie product that wasn't even our main focus, which is a way in which you just speak to your data, asking questions in English and it now has data intelligence. It knows the numbers. I actually myself had this quarter ended, our fiscal year ended last week and we saw a big spike in usage on Thursday and I couldn't figure out why. Is it normal? Is it like something wrong with the data? Normally I would have asked that question, it would have taken days for the finance team to get back to me. And I basically got back, Genie gave me back a report that had histograms on the data. It had like perfect visualizations on everything and I could just continue asking further and further questions. And I was just comparing, just a year ago this would have been a huge amount of work that would take a long time and I was just getting it at my fingertips. So I do think that we're going to start seeing this year more and more of these use cases where actually the knowledge worker is becoming more effective with agents.
Right. That seems to coincide with this like explosion in token usage as well. Even for like non-technical people like myself and producer Jasmine, we've been spinning up apps and building software and doing really cool things. Then what is Wall Street missing, right? Because it feels like yes these tools are becoming available to non-technical people, to the knowledge worker, and the hyperscalers in big tech are building for even more capable, more powerful models. Yet you're preparing and your investors are preparing for a world in which the tech trade gets wiped out again. How do I reconcile those two things?
Yeah, I mean I think the question is, you know, what has happened is that so we had these systems of records. The systems of records had a couple of really big moats protecting them. One was the user interface. So people, millions of people around the planet were trained on that user interface. So therefore you couldn't change system of record because you would have to train hundreds of millions of people. That's now being completely challenged by this English or natural language interface. And then the other thing you had is that the database, it's very hard to change databases. The system of record database is very hard to change. Both of those are being challenged on the database layer. These kind of lakebases that we've announced, which really you mentioned in the beginning, databases that cater to agents. So they're really awesome not for humans necessarily but they're really awesome for agents. So you're going to see that, that's a threat coming. And then the user interface, people will just use these agents. So I think all the market is just reacting to these two major moats for systems of records are sort of being lessened and they're going to evaporate. But that doesn't mean all systems of records are gone. First of all, you need a different system of record or some replacement to move it to. And secondly, the existing systems of records are not just going to sit there and watch that happen. So I think it's going to be somewhat volatile. Some will adapt and some will innovate and do it really, really well. In fact, if you think about it, if you have a lot of revenue, if you're a system of record and you have all these customers and the AI models are super cheap, you can just use them. There's nothing that's really stopping you from doing any innovation that anyone else would do to disrupt you. The question is just will they do that or will innovator's dilemma get in the way, which is, oh, I have to protect my existing revenue base and my seat-based pricing where I have humans using this and I'm charging based on that, not on one agent instead of let's say 10,000 humans. But if they don't do it, somewhere someone now has the tools to be able to do it for them, which is I guess what the market's reacting to, is just a lot of this uncertainty.
Okay, so there's like replacing software as a whole and it's mixed, right? You're saying that there's sort of companies, systems of record, cybersecurity for example, that can't be that easily replaced. But I thought this was really interesting from KPMG over the last week. They pressured its own auditor to lower fees because AI is making that work cheaper. So are you seeing that across your customers? Companies using AI not just to build, we know that's happening, but using AI to squeeze costs out of their vendors. Should they be doing that?
Absolutely. I mean that's one of the most common use cases. You know, a lot of the work that was done before you can now lower the cost of it. Royal Bank of Canada, basically similar, like you had these analysts, equity research analysts that would write these reports, earnings calls came out. How's this company doing? Now, you can with AI do all that research in a heartbeat. In fact, to be specific, in less than 15 minutes, they can get the earnings call from all the competitors, compare it, contrast with the rest of the market, get all of that done with an agent, and then the equity analysts can put the finalizing touches and really what is the actual recommendation and the really deep insights that they might have. But of course, that product, those kind of products, the TCO and the cost of goods sold is going to change for a lot of these businesses. So people are putting pressure, why should I pay this much when I can actually lower the cost of this portion of it. So that's happening as well. So the question is what exactly will come out of it. I think just last 10, 15, 20 years people were just thinking of SaaS as a safe haven and you could just place the money there and it's an annuity, just keeps paying every year. The metrics are very clear, churn and so on. You can just calculate everything. So it's like you could be pretty lazy. Now, it's like, no, you have to think hard. Which one is actually going to be able to innovate and which one is not? Because some of these SaaS companies are going to be wiped out because they're going to be lazy. They're going to focus on short-term revenue, and they're going to protect their existing revenue, maybe raise the price, not let people access their data, not innovate just to get short-term benefits. And then in a year or two, those businesses will go out of business. But some of them will also do the right thing, and they'll survive, and they'll do great, and they leverage AI the right way.
Right. Right. But what you're talking about is a whole rerating of an entire sector, which is I guess what we're watching. Um, okay, Ali, I want to get to some of like the fun stuff as well. Did you watch the Super Bowl? Did you watch the commercials?
Yes, I was there actually. Yeah.
Okay. You were there. Who were you cheering for?
Yes. Um, you know, Seahawks, but uh, you know, so good. Um, had to leave early. Yeah.
Okay. Okay, we're a Seahawks household and we can forgive you for leaving early because it was, you know, not the most interesting game in the world. Okay, so I'm sure you saw the commercials either way. Who won in your view in terms of AI commercials?
Well, I mean, look, there were different varying of how interesting they are. There's many different ways to analyze those commercials.
