 Hello, and welcome to theCUBE pod episode 52. I'm John Furrier with Dave Vellante, Extracting the Subliminal of the Cubes. Look at the angle, Cube Research. Dave, one year in, 52 weeks of straight podcasting. One year? This is our anniversary? I think we missed one week, I think. Yeah. All right, we did it. I think we're in the top 10 in the overall tech industry. I'll have to go check the rankings on Apple iTunes. But yeah, we're rising right up the charts. Oh, it's good to be with you, John. All right, so, I mean, plenty of news. Obviously, it's a couple of things going on. I'm wearing my March Madness for Carolina, my daughter's senior there, UNC, big upsets in the March Madness, Bracketology. SBF is going to jail for 25 years. And that's the huge news. Obviously Databricks announces a potent open source LLM called DBRX. And Big Money Amazon pulled the trigger on their next tranche of their $4 billion investment in Anthropoc. I think they're adding another $2.75 billion into Anthropoc. Code here gets a $500 million. Jeremy Burton, keep alumni raises $115 million in Series B. Intel, Samsung, Qualcomm, all form this UXL foundation to take on NVIDIA's CUDA software. Okay, and also the AI officer role is in flux. We talked to Cisco and a bunch of other people working on their AI strategies. It's clear that an AI officer is coming down the pike. And of course, we've been a bunch of different events. We've been covering EC Connect, Adobe, all the different events, and Google Cloud's coming up. So we're about to slide into heavy-duty season for theCUBE. Obviously all the hot enterprise tech events will start trickling in. Like I said, we coupled a couple of months this week. We got Google Next, cloud events coming. Just a ton of activity. I think April, May, and June are going to be like literally nonstop for us and our team. We're going to be packed. And there's so much more action. I mean, every day it's like, I feel like I'm getting behind every day. It's like, oh damn, I got to dig into the model of experts. MOE, the new thing everyone's talking about, the MOE. I didn't have that on my bingo card, as they say this week in the news, but Databricks brings up this MOE, this model of experts concept to the LLMs, but faster, smaller, cheaper. This is the kind of a movie we've seen before with AI right now. And Databricks, I love the open source, open angle on this because this is going to be a boom for developers. You're going, we're going to see the beginning of a more of accelerated speed game. How people are using LLMs, using AI. I was just reading a paper just today on how academics are using the LLMs for peer review. You know why? Because it's more polite than peer review. They write in a coherent paper. So, you know, research mall, I mean, everything's being impacted, but it should be a great event season as we go into year two of the Q-Pod and also kind of year two of the AI hype day where the consumer will lead the market. You're going to start to see enterprise starting to put stakes in the ground. And it's going to be the year of, you know, put the stake, show the stake on the grill. If you get the sizzle, you better show the stake as they say. So it's going to be the year of open up the cover and let's see how the stake looks. Who's got the beef? That's going to be the question this year. Who's got the beef in AI? I was 16 months in, right? 16 months into the AI wave. A lot of sizzle, Dave, a lot of sizzle. Where's the stake? Well, I thought that Databricks announcement was pretty sizzly. They made a lot of claims. I mean, you were saying, I mean, I looked at it and said, wow, so much for small language models. I know it's 36 billion parameters, but that they use out of 132 billion. But that still feels pretty big to me, John. I mean, and you know, you just, I think as we were coming on the Q pod, you said somebody else just lead, lead problem. Samba Nova systems out of Palo Alto, just announced their COE version 0.2 outperforms TBRX and Databricks and Miestrel and Grock, okay? At a breakneck speed of 330 tokens per second. Again, this is just the breakthrough speeds without sacrificing precision is going to be the key. And it's going to come down to this whole, what will AI run on? That's the big enterprise battle. So, you know, remember we used to use the NASCAR analogy. She used to say horses on the track. I use the NASCAR analogy. All the cars are kind of in a cluster. Someone gets ahead, someone slingshots out front. I think we're going to see a lot more of this where it's going to be an arms race in AI like we've never seen before. We've seen arm races before where people who ever can get out there out front will win. But you're starting to see a technical arms race and a financial arms race. You're seeing the investment community going all in even in spaces like observability. Jeremy Burton, we mentioned, Observe got $115 million in a Series B, which is a good Series B round, but that's a very tight market observability. So he's betting that they're going to be, they're going to have a war chest of funds to propel themselves into the position to be in the pack. Whatever analogy you want to use NASCAR, you know, Saturday moving day in the master is you want to be in the hunt. If you're not in the hunt, you're going to be left out. I think this is what we're seeing right now in AI. On Jeremy's raise, it looked like to me, like he didn't do the typical, you know, Jeremy's got wisdom, it's been around a while. He didn't do the typical try to maximize the valuation. I think he tried to maximize his balance sheet and his runway and create an attractive deal for investors. That's the way it looked to me. But I wanted to come back to DVRX, if we could for a second. Yeah, absolutely. I thought it was, I started to dig into it and I'm just sort of getting into, I was going to do it for my breaking analysis tomorrow, but it's like, tomorrow's good Friday. I kind of want to leave early and do the stations if I can. But so I got to pick an easier topic. But so the question is, so they've made a big deal out of being open source, but I was started to read the licenses. I don't usually read these open source licenses. There's some stuff in there, John, that is pretty interesting. Let me share some of it with you. Yeah. I found... On Twitter, by the way, they were commenting on this. So go ahead. Well, so that's how I saw it. And I was like, hmm, somebody said, it's not really open source. And I'm like, well, why not? So let's go read the license and see. So I found a few things that I think are worth discussing. And yeah, maybe it's okay, but I'd love your thoughts. So it says all distributions of DVRX or derivatives must be accompanied by a notice text file that contains the following notification, quote, DVRX is provided under, and subject to the Databricks open model license, copyright Databricks Inc, all rights reserved. So I'm like, that kind of caught my attention. That was one. And then it said, if on the release date of DVRX, if the monthly active users of the platform you're applying it to are more than 700 million, you can't use the model without Databricks permission. So that's by dance, I guess, and meta. It also says you can't use DVRX to improve other LLMs. So that's kind of restrictive. And in the sense, I found another clause that looks like they're maybe not hardcore about it. Maybe they won't enforce it, but forced model updates, you got to be the most current. And I can see why they do that. If an earlier version is deficient, they don't want it out there. But so I guess it's open in the sense that it's not a black box, the weights are open. But I don't think the training data is been revealed. And I think just, I don't know what other open source licenses look like. I got to dig into it, but these three or four things caught my attention. And I was like, huh, maybe it's like quasi open. What do you think about that? Well, first of all, it's a great call out. And I got that, so that same tweet might have been hitting us both on that one. But I think it's a really important point. This isn't a new era. If you looked at what Jensen Wong said at NVIDIA's GTC and all the covers we've been doing with AI, everything points to this new generative model is new. So if you combine, I call classic open source licenses, they've been under siege lately. You've seen what happened with MongoDB, you got Hashicorp, even Redis recently have changed their licenses all to get commercialized. And that's caused all kinds of forking. So just generally speaking, open source is upside down on the licensing anyway, right? So Matt has say he's been very vocal on this, on Info World, his column and on SiliconANGLE. So you have this pure open source model, commercial open source as it becomes more entrepreneurial and there's wealth to be created happening. And then enter this LLM stuff, which in and of itself is kind of like code. I mean, data, as I've always said, is like software. It's like code. So when you start looking at LLMs, you almost got to look at them like code bases. They're not code bases in the classic sense of software code, but they are code in the sense of they are data and it's got value. Like software used to have value in a proprietary sense. So how do you apply an open source concept to data that is constantly changing, right? So it's a very tough licensing call because one, it's never been done before. Then you got to sit down and figure out, one, what's the future going to look like for the data as it evolves? Number three, what's the best license to apply to it? So these are a lot of unknowns. So my first reaction was you play it down the middle of the fairway and see what happens. That's the way I see this going on. So I'm not too concerned with the license. I think it's more of a hedge from Databricks not to get into trouble and play it straight down the middle and say, let's see where it goes. I think it's an opportunity. I think what should happen is a team of people in the industry should form a little collective council or open source kind of vibe or, and just figure it out. I'm not deep enough on the licenses to understand the nuances between work on the data side. But you know, to me, you know, I can see someone saying, I don't want to do all this, let this work get pulled and stripped mind into someone else's model. So there's going to be a game of thrones going on and around this area. So again, the contributions are going to be like, we beat you and it's going to be the arms race. And then you got this license under the covers. And then by the way, who owns it? So what if they're mixing in data that's owned by somebody? So it's a can of worms to begin with in a good way. That's what innovation looks like. It's not pretty. It's not always tight. It's kind of ugly. And as they say, it's sausage making in the early days of these big waves. So I love what Databricks did. And I think the way they handle themselves over the past two couple of years, they've been good at managing some of these open models, especially how they handled all that stuff when opening up the data formats last year. So they have a track record. So I kind of trust Databricks on this one. Well, so yeah, and I couldn't find anything about the training dataset as to whether or not that's open, so I presume it's not been revealed. And this came out of the Mosaic ML acquisition, right? And they were, did you see the benchmarks? Did you see the, it looks like the graphs were kind of thrown together. Like, I was, you know, maybe I'm being a little bit too pedantic, but like the 71% graph for a mixed roll was under 70% or whatever it was. It was just, they weren't lined up. It was sort of maybe rushed a little bit, but I do a lot of crappy slides too. So I really can't. Well, Rob Hofen and Paul Gillin wrote that because Rob went to the dinner and had the briefing, but I think I read on Twitter, I was talking to Ali and Naveen on Twitter around some messaging back and forth when they announced it. The guy who did the graph spent all, holding all nighter, because he had to get him out. I've been there. I mean, I've been there. So I can't, that's, but I just- I got him some slack on that. But I don't totally, but it was just like, it just was interesting to see. But it's a, you know, just looking at it, some of the benchmarks, it's like two and a half times the size of mixed drill with about, you know, half the inference speed. So it's like, okay. And I don't know. I don't know how much to put into these, you know, the benchmarking. But then I was thinking, all right, well, what is, how does this fit with Databricks business? And I think it sits on top of Databricks. And they're basically saying, look, bring your data, bring your machine learning, bring your gen AI, we'll rag it, we'll bring vector search to it. And I think their play is to sell DBUs, right? That's what Databricks sells, Databricks units, which are some opaque, at least to me, Databricks processing unit, and they sell a crap load of them. And they'll probably do really well with this. So that's kind of how I looked at it. And it's, I guess it's a pretty good play. And it's, you know, it's, I think trained, I read it was trained on 12 trillion tokens, which is a lot more than like... Well, the large has got 130 billion, 32 billion parameters. And it's the design is around the MOE model of expert architecture. And it switches between which experts pieces to go with in the code. That's the key innovation. So it's designed to work in a very low compute footprint, which we've been talking about LLMs being optimized to say edge devices, whether that's cameras on an intersection or a human wearable device or a small machine in a retail store, it doesn't matter. It's a non data center device. So you started to see that. I think that's something that we said from the phone to the handheld, human wearable to sensor on a network to a light bulb. All this is going to be low footprint size. Well, it was trained on 12 trillion tokens, but I think you're right. It's the sort of mixture of experts is how they did it. But you compare that to like Falcon, I think was trained on three and a half trillion tokens, the Falcon 180. So it should be substantially better, right? But anyway, it's a good marketing. First of all, Databricks is great. They a lot of cash, a lot of smart people and they pull a lot of talents to the company. If you look at how they handle their projects, they're doing good on the talent side. But what this really proves to me is that continuing to validate the developer market right now is so hot right now. These companies want these developers using their tools. And we've seen in the cloud native wave that whichever developer adoption wave hits, those companies around those tool chains are extremely successful. So if you look at all the public companies that went on past decade, all were adopted either organically and or because they had a great product for developers. So as the infrastructure players like Databricks, Amazon, Google all try to win over that infrastructure. We saw Broadcom having chips and we have Nvidia wants to be this big data center. The infrastructure is building out and fast as it possibly can for the picks and shovels and the platforms and tools because the developers are hungry for the best. And Databricks is doing this to get the developers hooked on their products and services. Yeah, they're not dumbasses. I mean, I think they recognize that slow outputs are a problem. It's funny, John, right? Remember when we were talking to, we were in the private meeting with Jensen for a couple of hours last week. He said, well, imagine if you could let the AI go off for a week, which is kind of interesting. But at the same time, we know users want instant ratification. So I think Databricks recognizes that slow getting to an outcome slowly is a problem and they're addressing that and they're saying, run this on our stack. So it's interesting, the data business, like databases, it goes through all these cycles, it's got a query response. But what I think Jensen said that in our private meeting, as you pointed out, what's interesting is that the world that we lived in prior to Gen AI was programmed for us, even databases. You type a query, you get a response. And so that was kind of a static world. Now it's completely generating at runtime, generated response. But what I liked about his talk was, he said AI will fill two use cases. The prompt response model type of query and like chat GPT and get a generative response, that goes from basic text prompt to multimodal, some sort of runtime assembly response. And then he said, send it away to go reason. So there's two use cases, prompting and responding and generating at runtime and then go away and think. That's reasoning. So all the big conversations now in AI that's happening inside the ropes in the industry is this whole reasoning aspect is huge. And it's the number one thing that nobody's talking about that's the hottest thing. Ever saw about chat GPT and AI and inference is better than training, blah, blah, blah, blah, blah. The real action is reasoning. That's complicated. And that's where these GPU classes are gonna come in. That's why Zuckerberg is buying up all the GPUs at Metta because he wants to squirrel away all those resources. So what's gonna happen is we're gonna see another AWS emerge possibly. It could be Metta or AWS with Anthropi. They're already making that bet hard. You've seen them deploy the capital. So the infrastructure for AI is really the battle Royale right now. And it's super exciting because the developers are hungry. I mean, everybody's moving to AI. Everybody, consumers, developers, every single industry, every single company wants AI and they can't build out the infrastructure fast enough right now. AWS would be the next AWS you're saying. But so, yes, and I- They're trying, they're at risk. I mean, they could. Before we go there, I agree with you what you're saying about Jensen, but Databricks seems to be doing the opposite. It's like, we want fast outputs. We know that's a problem. And so we're gonna make it easy for people to get fast outputs, bring your data, bring your ML, bring your gen AI, we'll rag it, we'll vectorize it, we'll API it, we'll give you a model optionality. Do it all. Because that's where the action is right now. Right now, the action for developers is get your data going. But it's a smart play. And if you can squint through the benchmarking, which they love the benchmarks, you remember they in Snowflake used to go, the urinary Olympics. But that works for them. But I would caution users to just be careful about the benchmarks. I don't think they mean that much because as you pointed out, everybody's leapfrogging each other. Well, I mean, we said last year, Google necks that whoever wins the developers in the AI race will have a big cloud position because it's going to come down to trust. And developers don't want to work with someone who's not going to be innovating. So the leapfrogging will happen. I think that's table stakes and people should expect if you want to be in this business like data breaks and others, the arms race is on. And that's the market. So again, the prize is the developer uptake. So, and that'll translate directly into software and infrastructure customers. I do want to point out too, and we've talked about this, we've written about it in breaking analysis quite a bit. There's still a huge gap. When you look at the LLMs that are being adopted, open AI has a huge lead over virtually everybody, including Lama. And so, but there's a lot of money out there and everybody's going after them just like everybody's going after Nvidia. You mentioned this Intel-Samsung Qualcomm thing. Intel's now in both, right? Intel, isn't Intel in that AI alliance that Meta and IBM did. So Intel's playing its cards in both decks spreading its, which to me, look, if you really want to take it on Nvidia, don't split, don't fork these alliances, get together. And let me ask you a question. Let me ask you a question because this has come up this week in some of our meetings with people, Dave. This momentum around ecosystem land grabbing is happening. There seems to be a trend where everyone wants to have their quote AI Karitsu, their crew, their tribe. They want to put together their ecosystem. So, as we know in the enterprise, having an ecosystem means you have partnered in adoption because it's in everything's integrated these days. So not having an ecosystem is basically saying, we're not relevant. So is there ecosystem washing going on right now? Are people standing up, kind of ecosystem stories that aren't on solid ground? And what's your take on this? And if so, is that a feature or a bug? Is that just the way it is right now? And what will we need to look at to see for a successful ecosystem? Well, I think you're seeing history repeat itself. We saw this in the Unix days. And remember the original Unix was the original open systems, right? And you'd have different factions. Remember digital equipment led one and the sun led another one. I can't remember exactly what they were. And they ended up all going to Linux, right? I mean, which is the true open source. And so I think you see this all the time. These cobbles do form and these Karitsu's form and some of them make it, some of them don't. I guess my point is right now, everybody's going after Nvidia. Nvidia's got the mantle, they got the margins. Your margin is my opportunity, that's saying. But so it's kind of surprised me that these guys would fork it. And I guess they do it because you've got founding members and you've got laggards who come in later and they don't maybe have as much influence. But I don't know, you look at what the CNCF has done. I mean, that's worked. I mean, your margin is my opportunity is the basis line. But I think it's more of in any, you learn this in business school. I mean, anytime there's a market, the barriers to entry are as a key variable in assessing competitive advantage. And certainly Nvidia's got a huge bar, a barrier to enter. So they got the product, they got the software, we call it the moat. So anything that could increase the ladder to get into the market. And I think the plot to break up Nvidia's grip on the AI software business and hardware is to try to put an industry group together to have some sort of unified acceleration layer. It would be the equivalent of, to put our historian hat on to saying, IBM and owns the systems, network architecture operating system, the network operating system for proprietary mainframes. And let's create the OSI model, TCPIP, Ethernet. And I think you're seeing another kind of situation lining up where Nvidia is kind of like the monolithic mainframe. They even use that word in the event. But it's not proprietary as a vendor, but they do have a lock on the operating model with just the inherent barriers to entry. So in a way, there is a lock spec. It's called scale and software. So how do you break that? Open standards. Now the only way to connect that and try to get as much momentum. And of course, Nvidia will try to extend their lead during their time of prosperity by cutting deals with the metas of the world and Amazon's to have any product constraints control the supply. So the unified acceleration foundation, the UXL is a great initiative on paper to do that. But you know, Dave, we've seen the movie before where you can prop up these standards bodies. Sometimes that dog doesn't hunt. And so we'll see. I mean, it's going to be great to see. And by the way, do companies even have the time to participate in this? So my one API view on this is that I'm not sure I buy that. I think right now it's who's got the tools and de facto is more important than standing up and holding hands in public. But I think open source drivers. So what are you saying? I mean, you think that what Intel, Samsung, Qualcomm are doing in addition to what IBM and Meta did with the AI Alliance is that you're saying that's good because let a thousand flowers bloom or do you think, I mean, I'm not hardcore about this, but it just seems to me, if Nvidia is the target, which we know it is, why not just bring all those resources together and say, okay, we're going open because collectively that's the only way we're ever going to catch up to Nvidia because they have such a huge moat and we'll. Yeah, I have no problem with that. I mean, I'm just saying it's like, again, if alliances have to be functional to work, like I've seen many alliances that look good on paper, but the market is so focused on other things that there's no time to make the alliance work. If the market changes. Well, one, there's a couple of factors. One, do they truly committed to the alliance? That's one, two, is the market moving too fast for the alliance to even get their eyes on the beachhead to be secure. So that's going to be another one. If they're motivated by the fact that together they're winning together, then that's something that could work. That's how OSI happened. The open systems interconnect, the OSI seven layer stack worked because the people were highly motivated, open standards just distributed computing, but the full stack wasn't standardized. It was only up to layer, I think four or five was really got the most traction. Layers one through three were really standard. And that's why, that's how we got inter networking and spine and leaf is a critical rack and stack topology. What was the target back then? Was it SNA, IBM SNA? And DECnet SNA and the whole mini computer and proprietary operating systems of Unix. Unix systems and proprietary network operating systems. They control the physical layer up to switching and transport. So, you know, layer two or three. Which was essentially the mainframe. Many computers are basically small mainframe. Cisco became a company because of open standards. And they once they locked in, you had routing, you had layers two and three, you had switching and routing was your goal, you'd done off to the races. Now you can connect buildings together and not get locked into the proprietary mainframe IBM or DEC, which by the way, if you had IBM you couldn't run DEC. So it's like a proprietary didn't interoperate. So interoperability was the big thing. So we'll, I mean, we'll see. I mean, I don't want to get into a rattle, but to me. No, I just, I'm trying to do a historical comparison. It comes down to money at the end of the day. If there's a shortage of GPUs and the main supplier NVIDIA is controlling them. And a lot of the big hyperscalers control the inventory because they have the customers. That's going to create market opportunity for capitalism. It's like, hey, there's a market. And if you got developers, which is really strong right now get developers building apps and you got entrepreneurs and motivated suppliers to create an alternative, then you got competition. That's a good thing. And I think that's natural. And that's the benefit of capitalism. And I think that for that reason, it's not bad to deform it. Let's just see if that dog will hunt. Like I said, if the market's moving too fast that's a challenge. And if people aren't playing together and sandbox properly, then that could be dysfunctional. So those are the two areas that we're going to watch closely. So what do you think of the Reddit IPO? It was hot, rocketed up. Kind of pulling back a little bit. I loved it back at birth. First of all, I love the fact that Reddit was a great company and form. Love the founders, love the whole story. Love why it exists. I think it was a great IPO. Even though they have no revenue, it's a power dynamic that has all the elements of going big. It has a huge audience. Let's not forget the GameStop generation. So there's plenty of people to activate that stock. So it's got built-in power dynamics in the customer base. Love that piece of it because that brings that whole new vibe in there. And it has the opportunity potentially to monetize given the scale of the audience. So a huge fan of what Reddit's doing. Love the fact that it went public. It also shows that the IPO market might be opening up which is great sign because I think the stock market S&P hit an all-time high yesterday, which is interesting. And just again, it's weird how this market is. You have startups falling out of the sky, but you know, you've got massive economic engine. Again, the productivity piece is coming in. So I just think it's going to be a good IPO years coming but I don't think we're going to have a dark time. I think it might be dark for some companies on the wrong side of history. But I think the Reddit IPO, if the election doesn't screw everything up, then maybe we have a window here. So this could change a lot, you know, a lot of activities. I'd like to see Databricks go, you know, dying to see those guys go public. Arctic Wolf is another one I had two years ago. I predicted they were, I think in 2021, I said they'd go public, but kind of missed that one. But I think that's another company that could go and they're going to line up. I mean, there's got to be a big pipeline of companies that want to go public. Cohesity is another one, you know, that with the Veritas acquisition or merger that's called, I mean, Sanjay definitely wants to take that company public. So, you know, do you have to be, you know, a billion? You know, can you go with a half a billion, ARR or even revenue, right? Do you have to be a billion dollar company now? I remember it used to be a hundred million, yeah, they had to have a hundred million to do IPO. Now is it a billion? Is it a 10X in this AI? I think what you're going to see is you're going to see a fundamental change in how people are. The zero interest rate period is over. I think you're going to see a more pragmatic entrepreneurial story coming public. I think you're going to see real growth companies. There'll be some hype out there on the AI side, but I think in the next 18 months, we're going to figure that out, sure. But like in general, the vibe here in Silicon Valley is you can't run a business and think like get customers and at least have a strategy. If you're going to go for the growth, you better not just think of it like it's free money. You got to work on the fundamentals on the go-to-market aspect of the product market fit. I think that's the big observation that's happening here is that if you're in a series A or later, series A, B or C, if you don't have a solid field go-to-market growth plan, then that's not going to fly. You won't get funded. So those days of having that, not having that in place are over. You're going to be more fundamental when you go public. People have obviously freaked out about interest rates, 5% interest rates. But remember, during the dot-com run-up from whatever, 96, 97, 98, 99, even into 2000, interest rates were 7% to 8%. Now the difference was the Clinton administration who raised taxes were paying down debt. They were, they actually balanced the budget for one or two years. So now the difference is you got this massive 33, 34 trillion dollar debt and you're paying huge debt service. So that's the difference that compresses sort of the enthusiasm rightfully. So I think that has to be addressed. And of course, nobody's talking about it. And I don't want to really get into it, but that's kind of not the focus of our Q-pod. We like to talk tech, but yeah, so. I mean, we have a lot of, Adobe had their conference this week, they had that big blowout with pigment acquisition didn't go through. They're talking about Chinerva AI. You have this whole. I was invited to that, but I couldn't go. I wanted to go. I really like Adobe. I like what they're doing in AI. I think they're applying AI very intelligently. I heard the conference was really strong. I know I was talking to Andy Turai. He said it was pretty good. It was good. I followed the whole event, it was great. Followed all the keynotes, followed all the tweets. I obviously didn't want to go to Vegas. So I've tapped out, but we did cover it. We had a cube alumni there. We had cube collective resources there. We have a whole analysis piece coming out. So yeah, I mean, I think Adobe is really, they're trying hard. I mean, it was a good event, I'm sure, but because they have a great product, but Adobe's in the crosshairs right now. I think there, Dave, they had the big Figma deal went south on it. Remember that? Yeah, I do. I almost think that was a blessing in disguise though. Well, I mean, we'll see if they can cross the catapult. I mean, you know, there was a joint decision to end the merger. Remember that? I was, that's what I know. I felt like that, remember the Figma deal was announced, was consummated really prior to the chat GPT hype. And I think there was a period where they, that Figma wasn't worth what Adobe was offering. So I don't know. I think it'll- There was regulatory issues big time, so. For sure. And so that's why they both said, all right, forget it, let's tap out, but we need to add it again. I think Adobe with their assets could be a great Gen AI company. The question is, do they have the culture? That's what I was looking at and trying to read through the tea leaves there. Not sure yet how I feel about it, and more to dig in on that. But yeah, all good. And also, you know, saw the big news on Crypto Exchange King, Sam Bankman-Freed 25 years. You think that was the right call? I mean, to put him away for that long? I mean, not that I'm sympathetic to Sam Bankman-Freed, but you know, I think this, I think reasonable people could say, well. You know what I'm saying? It's a moment in time where you say, hey, you know, this is not how we do business. But my thinking is that they probably gave him the sentence and making him appeal out, something will go downstream, but it sends a message. If you're a Crypto Bro, and you're thinking about taking a shortcut, you know, don't fuck over people like he did. I mean, he basically didn't admit any wrongdoing, but you know, he'll be out in 2048, Dave. I think a lot of that was showed, I mean, he took the stand. He really showed no remorse. He's kind of a weird dude. You know, people say he's on the spectrum. So that couldn't have helped him. The tweets were pretty hilarious. SPF going to hold H-O-L-D in jail. Going to hold him. He's going, hold him. That's 25 years. Oh, it's beautiful. Which is, for those of you who don't know, it's like the crypto hotshot. Hold him. Don't sell, hold him. Someone else from SPF is going to come out of prison firing an all cylinders on the web five. Oh, that's brutal. The entertainment is strong on this tweet streams. So. But I mean, you know, didn't, I mean, I know people's money was frozen for a while, but didn't people get their money back eventually? I mean, I know they could have sold and the money was locked up. And I'm sure that that affected a lot of people. I think what Madoff did was, well, Sam Bankman Freed was pretty bad what he did, but the outcomes for people with Madoff were a lot worse. And apparently anthropic investments doing extremely well. Yeah, there you go. Right. Hit the sell that off. So. Right. He hit assets, he could liquidate and people got their money back or at least, you know, well, I guess you could say that they would make a lot more now with crypto rising the way it has, but I don't know. It's a philosophical question, but he got what he got, deserved it. Not to change topics, but I do want to get your thoughts because I just finished watching the Apple dynasty about the Patriots. And. Yeah, what's that? Have you seen that yet? I have. What do you think about it? Did you think Belichick was? I thought it was a Belichick hit piece, you know, I think from hard. I mean, it's it's a isn't it wasn't it a craft production? I was surprised, John, that the players like Matthew Slater kind of threw Belichick under the bus. I wasn't surprised that Amondola. Amondola didn't like Belichick. But I thought Matthew Slater, who was the captain of the team and he really kind of threw Belichick under the bus. They made Belichick look really bad. I think people forget the great things that Belichick did. They made him look like a total tool the way his camera angles were. He was like leaning back, looking, looking fat and evil, like he was like, he's an old school coach, not in touch with the players. Brady's the king of all things. And, you know, and so. Well, I didn't know what I didn't know because I wasn't living there because you were close to that. I didn't know how it was unraveling so much earlier in the process. I had no idea that Belichick really didn't want her around with the Garoppolo tire. So I had no idea that was the case there. So that's, that's the amazing thing about, you know, Brady, a couple of amazing things. I mean, first of all, don't forget, if Malcolm Butler doesn't intercept that pass, then Brady's just, you know, dynasty over, right? If, if there wasn't that strip sack against Maddie Ice in Atlanta, you know, I mean, football, they did an onside kick on that game and also, but by the way, but they also highlighted nothing was the whole Malcolm Butler sitting out in the Super Bowl. Well, again, that's, they really, hey, I don't forgive Belichick for that. He had his reasons. He didn't answer it. He was sort of all live. I've answered that. I mean, he, they really made him look like a tool, but he got to remember to all that stuff's edited. So it was easier to make him look like a tool. It was a Brady, Brady piece for sure. But, but I mean, it was also interesting, Kraft saying the reason he kept Belichick was the masterful defensive scheme he did against LA that year, the Rams. And by the way, the Rams were a better team. They had a better offense and a better defense. And he outcoached, what's his name? The young guy. I mean, if you're, if you're the, if you're the owner, you keep Belichick, right? I mean, I mean, that's, it's a tough, I mean, he, he did have to hold them together. But at some point you got to keep the coach. Well, in retrospect, no, but yeah. I mean, I don't think, I think at the time, and I'll tell you the sentiment around this region at the time was, okay, let Brady go. People were kind of pissed at Brady for leaving. And so, you know, a lot of people say, ah, he's dead to me. And then of course he just goes and wins the Super Bowl with Tim Pabetti. That was, that was, that was a dagger. But I mean, the first three Super Bowls, I mean, Brady's amazing, right? But, I mean, talk about team though. I mean, that, that team with Brusky and, and, and William McGinnis and, and those guys, Troy Brown. I mean, the consummate team, remember for Super Bowl, choosing to be introduced as a team. So, and, and Belichick, they bought into Belichick. And I think the last few years, the young kids didn't. They were like, screw this guy. He's old, he's out of touch. But I, I thought it was unfair. I thought they, they didn't do a really good job of talking about the dynasty. It was all about deflate gate and spy gate and all that other nonsense that was, you know, okay, they did it, whatever. But that's, but that's not why that team won. They, the team won because it was a consummate team. Everybody bought in Belichick was an awesome coach. He had great coaches around him that bought into his system. Of course they never became much outside of his system. But anyway, that's a shame to see Belichick, you know, I don't know, I think he's put his greatest coach of all time at risk these last few weeks. I think they could have done a better job. Belichick did get slammed. All right, well, we got the big NCAA brackets. How you, how's your brackets, Dave? Oh, I had Kentucky going all the way. They lost in the first round to Oakland. I mean, I'm like, I'm at school. The Kansas loss was a big one for me. Yeah, that was, I'm rooting for the Carolina because Caroline, my daughter is a senior at UNC. They made it to the final game two years ago. Didn't even make the NCAA last year. NC State, I had, I should have picked them because I had my hunch was to pick them to the final four because they were hot. They won the five straight games in the ACC. There were on fire. Then they'd love how they just got into that last game, beat Oakland was really an awesome. So we'll see how they do against Marquette. Houston, I got Houston going, going deep. I had Houston going to the final four. Yeah, me too. And I have Purdue going there too. So Gonzaga Purdue is going to go down. So we'll see. I think NC State can have a good ride in. We'll see how Arizona does. They're a wild card now. Caleb Love on Arizona used to play for UNC. I don't know if you know this, but he left the team because he got into a scuffle with the RJ Davis. He went after RJ Davis's girlfriend. And there was a huge locker room problem at UNC around this whole issue. So he had to leave the team or left the team on that. That's why they had a terrible year last year. Huge dynamic. Caleb Love was a great player. I had. You can't go after the same group. You can't go after your teammate's girlfriend. That's like code. Oh, right. I think that was locker room problems, Dave. You think that's going to cause locker room problems? Oh, my God. That's crossing the line. That's crossing the line. That's a big time problem. You just don't do that to bros. I had Arizona because my daughter went there. So I like Arizona. If I had one daughter, I went to Ole Miss. So I always root for them and not that they were in this tournament, but football. They had a good football season. Arizona, go U of A. Bear down. So you got NC State and UNC. We got, speaking of North Carolina, we've got SAS Innovation Innovate Event coming up at the Aria Hotel next month. The week after Google Next, they're in Cary, North Carolina. What a great private company SAS is. Last year was our first time going to SAS Innovate. I was killer. We got MongoDB local on May 2nd. I got my, I have two kids graduating from college, Tyler and Caroline, Northeast and UNC. We got the Red Hat Summit, Boomi World, Alterics in Vegas. Delta Tech World. Delta Tech World, Informatica World, IBM Thinks, Just Go Live, Snowflake, Click, Databricks, HB Discover. HB Discover. I mean, the list goes on and on and on. It's going to be a fun run. Browstrike, Hall. Let me ask you a question, Dave. What are you looking for the next few months? As we look at the research, I saw your research agenda we're publishing on theCUBE. By the way, great job on theCUBE research, publishing that agenda, research agenda. But what are you looking at? What are we going to be talking about in the next few months? I mean, the real, the thing that I'm looking at is, and I said in my predictions, AI has to be, 2024 has to be the year of AI ROI, or it's going to constrict funding. And I think what I'm seeing in the data is that expectations for ROI are getting pushed. The macro, the spending is getting pushed into the second half. So I think people are realizing, they're getting a reality check. If you look at what they're doing with generative AI, it's pretty basic. I mean, it's amazing, but it's really not game changing yet. You hear some pockets of productivity boost, and I think the potential is there. But in terms of the real dollars that are going to loosen up the CFO purse strings, that's not happening yet. It's getting pushed out. And I think it's going to be a 2025 and even 2026 story that this thing really kicks in. And I think the big question is, okay, everybody wants to know, when does the bubble burst? Is it like the.com? And I think because funding levels are so much higher here, you know, we may get a soft landing for the AI bubble burst. It may not be that bad. We may hit productivity, you know, and the bubble bursting may not be as painful as it was in the past. At least I hope that's the case. Well, I'm going to tell you right now, it's an opening day in the MLB today. So you got Major League Baseball kicking off, March Madness with theCUBE season kicking in. The CUBE community is doing great. Want to have a shout out to our 14th year doing theCUBE. Can't believe it's been 14 years and a lot of still fun going on. And we got the Red Sox tickets this year again, right? You do baseball in March. It's crazy. We got a bunch of tickets this year. The years at early for baseball. It's early. It's definitely early and the Red Sox haven't done much on the off season. I'm really kind of disappointed. It's like two years in a row. They really have just like- Is it home game or is that Seattle? Yeah, they don't, nothing home in March. The first home games in April, opening day there in Seattle. All right, well, Dave, we'll call it a day. I know you wanted to take good Fridays to the stages of the cross and Devers had a solo home run. Red Sox got a one-nothing lead. So it's streaming right now. That's what you know. I think it was 2013. I think they were not projected to be last and they ended up winning the World Series. So you never know baseball. All right, well, Dave, I have a great weekend and a great pod, 52, one year in. I got to say, I think I like the feedback we're getting from people. I got a great hat tip from a financial analyst and Bloomberg and a bunch of other folks on our pod that love the deep dive. So we'll continue to do deep dive and bring in some guests next year. Have a great weekend. Thanks, John. You too. You have 52 in the books. Thanks for listening, everybody. Go to siliconangle.com and keep that net. Check out thecuberesearch.com. New site we just converted over for wikibon.com. That's all the research. That's the heavy lifting content. See you next time.