 Hello and welcome back to theCUBE's live coverage from the Lake House, Databricks event, Data Plus AI Summit. I'm John Furrier with RobStretch, a CUBE analyst, keynote analysis day two. Rob, we're at day two, wall-to-wall CUBE coverage. Databricks, quite the mojo going on here. North and south here, Muscovny, we're in San Francisco. Snowflake had their event in Vegas. You saw the forking going on between head-to-head events. Databricks, got to say, not too shabby. Not on the messaging, not too shabby on their tools, not too shabby on their positioning. Yeah, no, I thought it was very strong and I think going into day two, there was a lot about open source today, there was a lot about their contributions back to communities, there was a lot about AI and how it's going to develop some really good thought leadership that was really taking it to the next level and I think the vibe has been awesome here this week. So let's talk about, you know, we had Mark Andreessen, who's just repented the, you know, AI is going to stay society, he's most known for, besides knitting the browser and the internet, the web, software eating the world and data is eating the world of AI right now and I think data is moving to, from a cottage industry to a full-blown architectural, enterprise-wide thought process around how do you organize your company to re-architect the cloud scale with data in a secure way, with choice, with the most developer-friendly mission? That's what's coming out of this, not some analytical siloed industry vertical that Gartner tracks. Yeah, no, I think it was- Not a database. It was very clear that, you know, Mark was looking out and saying, hey, listen, you know, it's not that this is all brand new, it is, but at the same time, the papers were written in the 1930s that really have led to this. So I think, again, you start to look at it and what he was trying to get across is some of the challenges that we have here, you would have even with humans. And I think, you know, our favorite topic about hallucinations, he went into that and said, how many times have you asked somebody a question and got the wrong answer? And that is some sort of hallucination in that way. So I think, again, he was trying to bring it back to its early days and that, you know, these hallucinations, in fact, Jamelle Brown from First Orion and I were talking about that here yesterday about the hallucinations and how people have them as well. Yeah, and what's exciting about this market that I think it's important to share is, and I'd like to get your thoughts on how people should think about it going forward is, it's so early that it reminds me of the early tooling days and platform days where there's a lot of like inadequacies around some of the key concepts, but that's natural, that's how things get started. So there's a lot more work to do. So LLM engineering is a term we heard yesterday from some of these hot startups. LLM, large-level engineering discipline. What that means is data is going to be a completely different industry than the classic, we're a database, a 3% market share, no one really has dominant share. It's not about the database, it's about the data corpus and some of the new techniques to get this LLM and foundation model built into applications. Right, and I think you could even see that today where on stage they had the CEO of DuckDB. It's not about competing against the database, it's about where is the data, how do you bring the data together, how do you get to make the data usable? And I think, again, to your point, it's about the contributions, and I think talking about Delta Lake, they talked about over the last 10 years it's had the most contributors of any data centric open source platform, and they really are hitting that platform, though, and they were talking about how there's over 40,000 commits and how there was over 3,600 contributors. That's massive when you start to look at that community bringing it together. So I want to ask you something, because you and I were talking yesterday, we interviewed Jeff Denworth with VAS Data, storage company, turning into a data platform, but they're underneath the storage level, physical layer, they're going up building high-performance platform, but that's not competing with Databricks, you've got MongoDB partnering with Databricks, they're a developer platform with Atlas, so you have a lot of kind of new emerging vendors with really innovative strategies, Mongo, VAS Data, Databricks, Snowflake, this is all emerging, and it's not this or that, they all kind of work together, but storage, in particular, is an opportunity with VAS to go up. I agree, I think what's interesting about where VAS sits in there is that it's, they're kind of, I guess you could say, object-file or block-independent in the way that they go to market with their platform that can actually sit underneath all of this, and I think you're going to have different storage technologies at different price points for different workloads. I think also, to your point, you start to see that today was the unveiling or a re-announcement of that Oracle is doing Delta sharing as well, and so is Twilio with their CDP platform called Segment. That's huge, Twilio has kept that a closed CDP platform for years, so them opening it up and saying, hey Databricks customers, you can now bring the data out of Twilio through Delta sharing is huge, same with Oracle, I mean I think that's probably one of the least written about pieces so far this year. I want to get your thoughts, you're working with theCUBE Collective and our customers on trying to figure out strategy, and a lot of people are kind of like moving in real time right now to put a new plan in place, not to replace any kind of go to market or strategy, they got to add a generative AI story, okay, to the pre-existing cloud native story, and I was just commenting with Joel, VP of marketing here, Databricks, about Amazon and Microsoft, how they're changing quickly, so I won't say it's a pivot, but it's kind of fashion, you got to wear the clothes that people want to see you wearing, and you never fight fashion, and the fashion right now is generative AI, and a lot of companies have it, and they don't flaunt it, and ones that don't have it are AI washing, so you have a dynamic going on where companies don't know how to leverage their story, and add AI to it in a credible way, you've been helping companies do this. Yeah, I think that it was even, I mean you've been quoted last year in this when I did some analysis for the AI ops cohort, and saying there's a lot of little, there's the companies that are talking about AI ops the most have the least AI, and I think that you start to get into it, and AI is such a broad topic that to your point, that the AI washing is at an all time high, and even Eric Schmidt talked about that today, and about how we're really good at hype, in fact getting it hyperbole almost to that, and I think right now is you have to break down and be clear to who you're selling to, and how you bring AI to that party, because people are really, it can go one of two ways, it could be cause concern, because thinking about data leakage, or privacy issues, or it could go the other way where, hey they think you're too far out, and maybe I'm going to have hallucinations, or something like that, so I think again you have to do, and be careful about how you use it as part of your go to market, but it has to be there, I mean you have to have a message, otherwise you're nowhere. So I want to get your thoughts on standards, we were talking before we came on camera, about the observation yesterday with Databricks, with the format unification, Ali said goodbye to the format wars, they're essentially saying, we will lay down standard here, and bring people together on the formats, on metadata, that's one example of many coming, so in every major inflection point in technology industry, some sort of de facto standard kicks in to make everyone snap together to capture growth, this is one of those moments, how do you see that evolving, what's your analysis, because there's many examples, you got the format, the uniform, that's just one, there's potential other ones, what are you watching here, where does the standards need to come together, what has to click? I think it's around adoption, and I think it really comes as delta uniform going to really as a standard, going to be picked up and driven by others, where I mean their running joke was that, using delta to make iceberg tables, you could actually make them faster in delta through uniform than you could through iceberg itself, so I think again, you start to look at how these adoptions and it's not one or the other. Do you think it has legs? I think it has huge legs, because I think people don't need to know that they can protect things that they've already invested in today, and that they can take it into the future with them as well, because the data has so much gravity and they've spent time engineering it, they don't want to move it. Yeah, that's the great analysis, I agree by the way, good point. The other point I want to ask you about is open source. If you look at the moat that Databricks has, their moat is open source, so open source has become a huge. Delta sharing has got traction. That's an interesting protocol, I'm watching that. Also the downloads and numbers are off the charts. So one little hidden kind of like differentiation competitive advantage for Databricks over say Snowflake and others is they got penetration on open source, and they got adoption. That's a hidden gem right there. Not anymore, we're talking about it. Well I mean it happens, and it's happening early on. I think it comes from the pedigree of, again, the Berkeley kind of cohort that they had researchers up on stage today from MIT and from Berkeley talking about how they're doing research around this, and I'm sure that a lot of them are using Delta Lake as the underpinning. I always like to joke Stanford versus Berkeley, everyone knows Berkeley is more progressive, more revolutionary than Stanford. Well Stanford's got chops too, which might, probably moving over back to Berkeley, but I'm always Berkeley professor, and you look at Berkeley's history, there's a lot of vibe in Databricks coming out of Berkeley. The idea of disruption, positive, enabling disruption, a disruptive enabler is something that disrupts but enables. And I think if you look at everything going back to say BSD, out of Berkeley, which really helped beginning pale eunuchs, a proprietary operating system. So I think this idea of democratization with the notion of pushing open and choice, it's huge, that's a Berkeley kind of vibe. Yeah, I agree, and I think that the folks from MIT that we're on today as well talked about it, and I think that pushing open, as we've been saying for months now, open wins, and I think it becomes a how do you play this and the ecosystem that they built around it, around open, they had a number of different companies on stage today that have started out as open projects, like Fireworks who was PyTorch and now are starting to grow and follow that Databricks model. So I think you're getting a lot of that out of these different companies where they're starting out here and growing up. And there's a number of others on the floor here as well. Bottom line, what's your assessment of the overall show so far? How would you grade these guys? What's your takeaway? You know, I think the overall show has been an A. I thought the keynote today was on point and I thought that the demos were there, they tried to bring the ecosystem in. The show vibe has just been awesome this week and I think that people have been engaged. I mean, even having the hackathon where they showed the results of that today around building AI, using some of the Lake House IQ stuff is really cool. And I think that the online feedback that we've been seeing has been phenomenal. Great stuff. Well, I dropped a big post about Amazon's master plan with Matt Garmin. You're seeing what's going on Snowflake. You see what's going on Databricks. The horses are on the track, as Dave Vellante would say. And not yet clear who's going to bust up, but Databricks is looking good. I agree, I agree. I think they're neck and neck and I think it's going to be a race to the wire to say the least. All right, this is the Cube coverage. We're live in the Lake House. This is the studio on the floor. I'm John Furrier with Rob Stretch. They're breaking down all the action. More guests are coming in. We're talking all the thought leaders. Go to silkenangle.com, the cube.net. Those are the sites. And check out our new, thecubeai.com. We have our own LLM too. We got a little AI there. Check that out in data. Register, I'll prove you if you're watching this video. Thanks for watching. We'll see you next time for more Lake House after this short break.