 Hello, and welcome to theCUBE pod episode 53. I'm John Furrier with Dave Vellante this week. We're extracting the signal from the noise. We break down the top stories we're tracking, events we're going to, and of course, the hottest news in technology and in the enterprise, which is mainly cloud AI and everything around open source and data. So much going on, funding announcements to big news going around the internet. Dave, good to see you. Hey John. One of the next year of the pod, we had completed a full year an hour on to year two and a great audience developed, not huge numbers, but like big numbers, but still not huge numbers, not in the top 10 of podcasts, but getting a nice kernel of elite enterprise technologists who are really doing deep tech. So I got to say the results so far from the podcast, we did it every week. I think we missed one week because of travel or holidays, but pretty much a full year. And so this year we up our game on our packaging and get our format tweaked, great, great year. Great year. Yeah, we're getting good feedback. People seem to like it, listen to, and I love it. I get good John time once a week. Well, we had great feedback last night. I did an event in Palo Alto, a networking event put on by Mindshare PR, plugged for Heather Simmons. And we had a great, great cast. We had Shujata Banjirji. She is a VP. She would have been an SVP under VMware, VP of VMware Labs, part of Broadcom. Neema Badey, who's at NVIDIA, cube with cube friend and Ehab Theresey, who's the senior vice president at Dell Technology Core, working on all the systems stuff around AI. And boy was the turnout amazing. Founder of OCP. I'm sorry, the CEO of OCP was open compute project. And just a great conversation. And they're all loving the pod. They're like, they're all like builders and they're all loving the little nugget. So shout out to Shujata. He came on at MWC, he was great guest. He's deep tech. Well, the thing is, is that what's going on is that there's a lot of infrastructure going on that's going to enable the AI applications that was there. Chris, my burger was there from Newstack and Ray Wang was there. Just a ton of great people. And it was just fun to see the networking on the vibe of like dot com kind of vibe or web 2.0, when the magical moments of discovery and entrepreneurs and builders. And it's not just young Gen Z's. There's senior people too on their third four startup, multi-exited entrepreneurs and VCs all at NOLA's and Palo Alto. And like I said, the pod is getting some good traction and the feedback is, you know, get more structured and get some guests. We'll work on that. Chris is at Newstack now. I didn't know that. Or maybe I didn't know that. He's good. He's really good. He's working for like three or four publications, but great to see him. And you were in New York. Yeah. You were in a tour of the NYSE and you had the Amazon event. Yeah, I had, so Matt Wood, who's the Vice President of Amazon's AI, gave us, you know, his overview of their AI strategy. And then we met with a bunch of other leaders, guys in financial services, guys across industry. We got demos. We got robot demos. It was very cool. And then, yeah, I was at NYSE. I spent time with their head of technology. Met with him for about 45 minutes, which was really interesting. Got a tour of the facility. You know, saw what's going on down there. Spent some time, did a little pilot with those guys and their TV set up. So that was good. It was great, actually. Well, I put a little teaser out there on Facebook and LinkedIn. And so it got massive viral buzz. People were like, what's going on with the NYSE? Had our logo all over the floor. People thought we went public. I'm not going to go public. Maybe, yeah, right. But there is going to be a New York cube. And that's just a matter of time. We're looking at final details there, but we'll be partnering with NYSE and others. What was cool was when they flashed the SiliconANGLE logo. If you haven't seen it, go to John or my Twitter. They flashed the logo. SiliconANGLE logos were on every station. You know how they have the trading stations? So they had, it was like a cube. And then all four sides of the cube were three sides of the, whatever, the shape had SiliconANGLE red logo on there. And we were up where they ring the bell. And you could just see out over the show floor or the stock exchange floor. It was pretty cool. Great publicity, too, all the background. CNBC sets down there. So having the cube on the New York Stock Exchange floor will be a great addition. Hope to pull that off and other, and all the VCs down there, too, are reaching out as well. So I think we're going to have a great set up in New York. And there's, you know, this, the media landscape has changed. It's talking to the head of IBM media on this and a bunch of other folks here in the East, when they're on the West Coast. IDC had their big event here. There's a real need for deep tech analysis on business, the impact of business. And, you know, leaders that are in our network, the CEOs that we talk to every day and check in with, love the fact that we talk about the impact of the earnings and the stock price from a technology perspective, not just from sales. And I think that is a need. And I was just on the phone with IBM covering their master's application that they're launching the new features with. And even more and more analytics are coming into the sports side of things. And so the real tech audience wants the deep tech. And I think that's going to be a nice compliment to the New York scene, where we have a lot of financial analysts who read our stuff and do watch our videos come on theCUBE and are in our community. So I think this intersection of financial technology, deep tech. And you saw that at video day, but there are GTC event. You had a deep tech conference. I mean, the hardcore developer conference. You had financial Wall Street-like crowd coming in, the financial analysts coming in, really unpacking, because I think the competitive advantage for the financial side is going to be getting on the right wave and connecting the dots. And the smart money's going deep tech and getting in the weeds and really digging in to the technology shifts. Back in the day, when I started the Wall Street service with Alexa McClellan, who hired her out of Goldman when I was at IDC. So back then, the sell side analysts, remember the sell side analysts, they print the reports that the hedge funds read. And then the hope is that they trade through the firm. So Goldman Sachs will write a report or UBS will write a report. And the hope is that the fidelities of the world will then make trades through the idea generator. But back then, the Wall Street analysts, the sell side analysts, they could actually get paid on investment banking deals. It was the mother of all conflicts of interest. So you could take a company public and then the sell side analyst was incented essentially to write good things about the company. So really, really huge conflict of interest. And so the SEC said, well, that's got to stop. So I can't remember exactly when it was like sometime early 2000s, probably like around Enron, you know. And so what happened was after that, and by the way, the sell side analysts would make bank because they were, you know, they were making it up at the back end. You remember Mary Meeker and she's still around and she was like a rock star analyst and John Levinson from Goldman Sachs and a number of folks that just made a lot of dough with that little, you know, wink, wink. So the SEC blew it away. And then what happened was the sell side analysts, all of a sudden became like spreadsheet jockeys, right? And so almost like they lost their interest in like deep tech. And so it became watered down for a long time. What's happened now is, you know, Wall Street making a lot of dough and they've started to properly fund these sell side analysts and sell side analysts are getting really good. A lot of them kind of left over from those days that have, you know, now have wisdom and a lot of new names that really understand tech. And so they've reduced, they've eliminated that conflict of interest. So now sell side analysts has no incentive other than, you know, they like the stock. And so, but I've know, you saw it at the Broadcom session. I mean, those guys are, they know their stuff and they're really good. Many of them, not all of them, but many of them are really, really good. So the deep tech. The sell side just for the folks listening, sell side is the folks who should be unbiased opinion of a company based upon proprietary research of the company's security. So basically they're supposed to be independent. Yeah, and the buy side are much more about getting the story right, selling to hedge funds, getting to the alpha, whatever they do. I mean, I mean, they explain the buy side. Yeah, so buy side is making bets. They're like, those are the people who actually trade the stocks, right? So, and so, so the sell side, you know, their incentive is to write good research, do good research so that, you know, they make good calls. And then the buy side is obviously that's, their jobs depend on it, both sides depend on it. But the buy side is the ones who makes the bets. They're the hedge funds. So the sell side, sell side of the folks are the ones that you see on TV. The ones that they see, like on CNBC, they're the analysts, quote, analysts that are supposed to give an independent opinion from on the stock and the company. And that day, as we talked, I've saw some analysts on CNBC, I won't name names in our industry knows who they are. They're basically, oh, Intel's going to tail one. Meanwhile, they just lost $7 billion. So there's a real problem with analysts on the sell side right now. Not being tainted, but there's a challenge. If you're going to be an independent analyst, you're going to be on TV, you better be independent. And if you're getting paid by Intel to say good things about Intel and then a week later, they lose money. You're obviously not a research analyst. You're not a sell side analyst because you're obviously being paid by the company. So there's a real problem with analysts. You mentioned IPOs, that was solved by the SEC. But in the industry analyst community, there's a similar problem going on now with this so-called independent analysts on TV should have an unbiased opinion based upon research, not if they're getting paid. It's coming back again in our industry. This has been a big controversy, Dave. And we've been seeing it. Yeah, so you're right on on the sell side. And so guys like Tony Sakanagi, you might have seen him, he's Bernstein. He goes on TV all the time. He's really, really good. Aaron Rakers, I don't think he goes on TV but he's a really good analyst with Wells. The B of A semiconductor analyst is really, really good. There's a lot of really good ones. And they really study tech and they go deep. So it's been an interesting transition to watch. We used to have two Goldman Sachs analysts inside of IDC. We had like a $2 million deal with Goldman. It was an exclusive. We couldn't sell to any other sell side firms. And then after the 87 stock market crash, they canceled the deal. And so I started the service with Alexa. We hired her out of Goldman and then we crushed it at IDC today as a great. Well, there's also explained there's financial analyst and an industry analyst because I think it's important because I was referring to the industry analyst on TV saying. Yeah, so that's right. It's a similar model though, independent. Well, financial analyst is certified. They have a CF, most of them have a CFA, a certified financial analyst. And they're qualified, they've been certified and their job is to pick stocks basically and make calls. And they're usually really, really good writers. Like I said, they're increasingly technical because they got to be, this is a technical business that we're in, whereas the industry analysts are like Gartner, Forrester, IDC. They tend not to be really in tune with the financials. It's funny. I mean, I love going to financial analyst conferences because I think that, frankly, I think financial analysts are smarter than industry analysts. No offense to my industry analyst colleagues. But in fairness to industry analysts, industry analysts care much more about product and getting deep into the business of the company. I shouldn't say that business. They care more about like go-to-market type stuff and just kind of things that are less financially oriented. And so they tend not to care about income statements and balance sheets, which of course I care a lot about. But anyway, it comes down to transparency. So it's a New York's hot. Tell us about your Amazon. You went down there, I saw Matt Wood on LinkedIn posted some really good commentary. He went down and talked to the media. He talked to the analysts. You were there for an NDA discussion with AWS and Matt Wood, talk about AI. Was that NDA or can you talk about that or? I can't talk about it, but there was a lot of NDA on there, but I can talk about it. And so first thing is it was pouring sideways when I got there, I drive to New York because the weather sometimes holds up. Like Andy Thuray, if like got canceled, couldn't get in. And so he was kind of bummed, but there were a lot of local New York analysts there. There was a couple of people from Gartner. Doug Henshin was there from Constellation. He's a really good analyst and a number of the folks as well. It was probably only like, I don't know, 10 analysts. And so Matt Wood kicked things off with basically it was only there for an hour, but he took us through like a bunch of slides. He had this seven, it was seven or eight point, called it a journey or steps. He called it steps on AI adoption. It really wasn't like linear steps, but he took us through that, like started with training and then he made a big deal out of security and privacy, which there obviously, AWS has always been focused on security privacy designed in. And of course it was timely. They didn't bring it up. They didn't bring up that report. Did you see this? Just a quick aside, a review of the summer of 2023 Microsoft Exchange online intrusion from the China hackers. Yeah, I can see that. So they didn't bring it up. They should have. Maybe it was inside conversations, but they know they were cool. That does nothing to do with Amazon. That was Microsoft. I know, but they didn't bring it up because, but in a way I brought it up in my breaking analysis because I mean, if you're a CIO or if you're a CEO and you're relying on Microsoft and you read this report, you're like, whoa. Just to clarify, because it's a little non sequitur there. The Cyber Safety Review Board issued a damning report Tuesday on last summer's discovery on Microsoft's breach. That was the survey. And basically revealed a wide conspiracy between the Chinese government and other people. And they got sensitive authentication key and broke into Microsoft's managed accounts to steal massive sensitive information from the US government. And they just basically got hammered. They got taken to the woodshed. And I'm sorry, I didn't mean to go the first. No, no, no, let's stay on it. It's relevant because, okay. So it's unbelievable, John. Well, explain what you held up because I... So in 2023, there was a... Explain what it is in your holding your hand. In 2023, there was a hack from Chinese. It was linked to Chinese actors, but you're holding your hand over a board. I know, I'm gonna explain it. I'm gonna explain it. It was linked to Chinese state actors. And then the US government found out about it. It wasn't even Microsoft that found out. They had alert Microsoft. So the US government was like, WTF, we have to look into this. So the Cyber Safety Review Board was asked by the Secretary of Homeland Security to do a deep dive on this. So they published this back in late March. And I'll just give you a couple of poll quotes that I pulled out. By the way, we reported that. I reported on the podcast. Yeah, yeah. And SiliconANGLE wrote it up. Friday at four o'clock or something. They announced it in November. Remember that? They said it was the last possible minute before... They did such a poor job. No, they tried to sneak it on Friday before the weekend. They got eviscerated for that. So look at this poll quote, quote, the board finds that this intrusion was preventable and should never have occurred. The board also concludes that Microsoft's security culture was inadequate and requires an overhaul. And then the second poll quote, throughout this review, the board identified a series of Microsoft operational and strategic decisions that collectively point to a corporate culture that de-prioritized both enterprise security investments and rigorous risk management. And then the third takeaway was they cited best practice cloud examples that were kind of exemplary from Google, Amazon, and Oracle cloud infrastructure. And then they had a bunch of contributions from some other people. But so the reason I brought it up is because if you're a CEO, a board member, a CIO, a CISO, a P&L manager, and you're running your business on Microsoft, I would be like, whoa, time out, and we're betting our AI business on Microsoft? Hang on, we're at risk. And this report just eviscerated them and said they still haven't figured out what happened, why it happened, and they really didn't come clean for a long, long time. And so that should be a real cause for concern. And AWS, to their credit, they didn't really hammer on that. They just said, hey, we're focused on security. We designed it in. I think they did a good job of explaining that. It's funny to me that Charles Fitzy, he's always crappin' on Amazon for like, you see him this week, like searching for a compression algorithm for experience for AI. He's always crappin' on Amazon because they're maybe behind the race in AI. But he doesn't even bring this up. I mean, to me, this is the number one biggest concern is security. Yeah, I mean, this is really Microsoft puttin' their head in the sand. The other, this is so much backstory, but if you zoom out for a minute, okay, and look at Microsoft, they hired Charlie Bell away from Amazon in 2021. It was a big move. He was supposed to be the right hand, the Andy Jassy-era parent. And then they had this whole initiative of this secure future initiative that they were gonna implement, and they did, nothing really happened. So they're getting killed inside the industry by people throwin' a lot of darts at them, saying, hey, you guys, you know, they're supposed to implement this new thing and it just didn't work. Now, the thing that's interesting about this report that came out, again, we reported it, the minute it happened, we were like calling them out immediately. So we were all over it. The customers didn't have all the facts. So nobody knew. They hid the ball, and what was even worse is, they knew about it, okay? So they actually knew about it for long before. So again, the government doesn't trust Microsoft. So they issued a huge thing. Now, remember, we reported back in 2013 that Amazon won the deal to CIA. So this is a huge setback for the cloud of Microsoft. Azure team, you gotta be really lookin' at this and saying, look, this is a major, major, major setback. And if I'm Microsoft, and I'm Amazon, I mean, if I'm Amazon, I'm like, look it, we won the CIA, we have a GovCloud, Microsoft's foray into public sector just isn't working. So that's going to be really, really interesting to see. And to your point about the poll quote, they just don't have the corporate practices according to the government. But I got to tell you, this CSRB, which is the entity that gave that report, they are not happy. And then they, they was very much a threat and they knew, and they knew about it. The other thing that comes up is that OpenAI now has these co-pilots. So I don't know if you remember, but when we had Microsoft Ignite, I was hanging around the Seattle Westin, trying to get some stories from some folks there. And I overheard many security people saying, we are turning off co-pilot because they were launching it with default on. So this is- I reported that too, John. This is going to be a breaking analysis. This is going to be a huge blowback for Microsoft. So, you know, I'm surprised Amazon's not jumping on this because they're getting killed by OpenAI and Microsoft. Well, I know, I want to talk about that in a second too, but said Microsoft's failure to detect compromise of its cryptographic crown jewels on its own, relying instead on a customer. The customer was the government to reach out to identify anomalies. It got, it compromised the email accounts of the Commerce Secretary, Gina Ramondo, U.S. State, United States Ambassador to People's Republic of China, Nicholas Burns, Congressman Don Bacon. Can you imagine the private emails talking about China? I mean, just, it's just really incredible. And then, and then wait for one more thing. However, by the conclusion of this report, this review, Microsoft was still unable to demonstrate to this board that it knew how Stormo558, that's the actors, had obtained a 2016 MSA key, 2016 technology. And they were using, they were doing manually, manual updates to it. They hadn't even automated it. I mean, just really, really bad. No, no best practice. So wow. The report basically slammed them for dragging their heels and- But John- And delaying, because they publicly tried to bury this thing. And the thing that killed them on this one, the experts already knew it was sounding the bell, alarm bells on this. And this is the hackers getting into the government email accounts managed by Microsoft. So, it's unbelievable. So I think this can be, again, back to the setback with open AI giving Microsoft a huge cloud burst in the market share numbers and the vibe of the market. This is going to be a setback on the public sector. And I think Amazon CIA deal in the GovCloud will reap the benefits from this. But John, but here's the thing. If you look at the data, the spending data from our partners at ETR, open AI, open AI is the number one company in the whole survey, 1800 IT decision makers. Open AI is the number one in terms of spending momentum on their platform that percent of customers that are spending more net customers. It's like off the charts. And then Microsoft is like right there with them. Like they are off the charts. And then, but I'll give you some other data points. Anthropic is rocketing. Doesn't have the presence, doesn't have as many, you know, the responses in the survey, not nearly as many as an open AI. Databricks is also rocketing. And the other thing is Google and AWS are actually coming together. You know, I've reported for years on the cloud spending and AWS and Microsoft are by far the leaders in cloud spend in terms of the customer momentum. But in AI, Google and Amazon are coming together while Microsoft and open AI are like running away with it. So my point is that it's despite these reports like this, they're still kicking butt. And it's a real risk. And I think CEOs need to step back and CIOs and say, hang on, we have to do some risk assessment here because this is bad, bad, bad. The Dion Hinchcliffe from Constellation has got a survey going on around business uses. What are the questions? What is the primary LLM that you use today for your work in business? ChatGPT 41%, 41.7%. Cloud, which is the cloud family, that's which is Anthropic 25%. Gemini 16. Mistral, Lama, not registering up even under one. Basically zero on the chart. So that's interesting to see that. So let me give you some others like metrics here. So net score is the net percentage of customers. Basically, who's spending more? But what percent of customers spending more than less? You subtract the lesses from the mores and you get a net score. Open AI is at 80%. So say 80% of customers are spending more and out of a survey of 1800, 555 are saying we're using open AI. Compared to Anthropic, 77%, 76% net score, a huge net score, by the way. Anything over 40% is great, but only 89 responses. Microsoft, 74%, with 611. So you combine Microsoft and open AI, it's over 1,000 out of 1800. Databricks 69%, with 209. And then MetaLama, 60%, very strong, but only 87. So not as popular. AWS, 58%, with 370. And Google, 56%, with 340. And then, so all kind of coming together with this. But the other thing is Meta was ahead of Anthropic and Databricks last quarter. They flipped. Anthropic and Databricks are now ahead in terms of spending momentum. And then there's other guys like Oracle and IBM are in the chart, not as much momentum, but they got a pretty big presence in the marketplace. So the race is on, John. I have a question for you. I want to run something by you, something that I heard from Matt Wood. He kept talking about how model optionality and diversity is an advantage. And you know how a lot of people are saying that LLMs are just gonna get commoditized? You hear it all the time, right? You know what I'm talking about, right? Okay, so I asked him, I said, well, wait a minute, a lot of people think LLMs are gonna get commoditized. Where do you stand on that? And he said, definitively, no, I don't believe that at all. That model diversity, right tool for right job. And we're gonna play off of these other models. We're gonna use models to train other models. And that's gonna create, he didn't use this term, but essentially implied that's gonna create a moat for Amazon. I was like, hmm, and then I was on another call, same day with the head of strategy at a very large company, a tech company. And I asked that individual, what do you think about LLMs? He goes, you know, I've been around a long time, used to be a VC, I do a lot of strategy work. I think they're gonna get commoditized. VC gave it away, he's the wrong answer. Never ask a VC about a trend ever. If you're an entrepreneur listening to this, never ask a VC for advice. Because they're not entrepreneurial. Well, what do you think though? What do you think? Absolutely, they're not gonna get commoditized. Never, no way. So Amazon's assumption you believe is correct. Yeah, look at, the models are just the beginning of the new substrate of how the new architect is rolling. There is so much change going on in this market with AI and the infrastructure layer with data, everything's changing. Databases, Tesla's not even using data, databases to train their thing, he's video. They're using video images. We talked about that last night on the panel. All computer science tools and mechanisms are gonna radically change. All the sacred cows of computer science will evolve to the new normal, which is going to be AI infrastructure. Again, we talked about this last night, our Palo Alto media, too bad it wasn't recorded. LLM and foundation models will be a long tail power law which we publish of you. Look up the long tail of LLM, you'll see it. What's going to happen is API is going to connect the world. I'll continue to do that. LLMs will integrate with each other. They're going to talk to each other, they're going to have mechanisms for interacting, data interacting with each other and that's going to be the new model and that's why commoditization is not going to happen. It's going to be a value market, meaning whatever LLMs have value, other LLMs will seek them out and integrate with them for better value. So the combination of fusing data sets will be the model and absolutely will not be considered because a small model could be very valuable. If the data is valuable, data is only valuable in the eye of the beholder who wants to use it. So what does even that mean? It's going to get commoditized. Now, if you're going to argue anything, you can argue that open AI is going to get commoditized because it's vanilla. If you index the entire web, you're never going to be accurate in areas. So some areas you're not going to be accurate, hence hallucinations. So what's going to happen? It'll interact with other models. Smaller models will interact with larger models. So the power largest represents the size and scope or domain specific of the model. So now you're hearing specialty models. What does that mean? That means the data is specialized around something and that's what enterprises are going to have the value. So that's where the value is. It's going to be in the data and the workflow. So just like, let's take an example. I went out on detail last night on this. Let's take a semiconductor example. Is ARM a foundry? No. They design the chip and then ship it to Taiwan. So they actually don't even do that. They just have the software framework and then they give it to designers. They give it to designers. They give it to a standard and then ship it to TSM. Okay, so the thing is so tight. It's like they're getting the tape out in like a lap of time. Let me finish. Arms were in Broadcom. Broadcom does the same thing. So they design and then they get someone in a foundry to build it. In AI, what's happening is the people who design the value, the combination of the workflows or data could actually design a winning interaction and then ship it to say a foundry using quotes, meaning some sort of centralized compute farm, training farm, inference farm, reinforced learning farm, neural network, knowledge graphs, there will be someone like a cloud player or a hyperscale or maybe meta who will be processing and you can process it on the hyperscale clouds, on premise or edge, the compute side is going to get really, really robust there. So the value is going to come from designing AI and so people could use AI. So do you build AI in yourself or do you use AI? So that's really going to be the distinction. So Matt Wood is accurate on that. I would say that he's absolutely correct. Like Matt, models will not commoditize because what does it even mean commoditize? Well, it's okay. So that's part, well, what it means is that the value is going to flow up the stack. And that's what a lot of people want to believe but the second part of that question is, okay, if Matt Wood is correct, that models won't get commoditized, let's assume he's correct. And that becomes a key criterion, model optionality. Will other competitors to AWS be able to match that? In other words, last year at Ignite, Microsoft announced, I think some stuff with Cohere and you know, a couple of, I forget who, but, and I remember tweeting out, well, okay, so they're adding optionality. Will they be able to do so in a way that competes effectively with AWS or will AWS be able to build a mode around it? That's a wrong question, the wrong question. It's a question, I'm sorry, it's a question. Here's how I flip it. You can't say it's moving up the stack and being commoditized or abstracted away are two different things, okay? You can have commodity means comes across as no value or a low-switching cost. Well, that's what they're saying. They're saying low-switching costs. Low-switching costs race to zero. Well, depends on how valuable the model is. So again, some, if you're going to be obsolete, let's just say a model like OpenAI or Mestral. If they stop innovating or working on their model, they'll become obsolete, right? So in fact, I introduced a new term last night on the panel called tech obsolescence, not tech debt. Because if you have technical debt and you go out of business, you're dead. So there's no debt because it's irrelevant. So you can be irrelevant and be commoditized by not keeping up. Yeah, there's a changing landscape, land change, trying to keep trying to catch up and stay ahead and there, some are, say, losing ground. So what's happening is it's moving so fast. You're not being commoditized, you're being beaten. So commoditization is a whole nother conversation. By the way, I'm not defending the commoditization. No, no, no, I'm just trying to explain my thoughts because abstracting away the complexity is not a bad thing. If the models are valuable for the application, which is the top of the stack, that's where the value is. So the question is, the question of moving up the stack is about the application getting value. That's data and infrastructure capabilities, compute, big iron performance, low latency, real time. So as that infrastructure gets tooled, which is being done now, that's the bottleneck right now is infrastructure. So I think the commodity conversation is more about who stays ahead. So it's not being commoditized, it's becoming competitive. Well, and these guys are investing like crazy. I mean, it's like- Now Fitzgerald pointed out on Platform Economics today, on his blog that Anthropic is in bed with Amazon, obviously, but their alternative to open AI. He's reporting that some people are telling him that they're working on a competing version. Dude, that friggin' report came out last November. He's making it sound like he has some inside baseball. That's like old news. Olympus, right, Olympus came out- Explain. So Olympus, I think Reuters broke the story last November, like mid-November, around Thanksgiving, that AWS was working on a new foundation model that was world-class, two trillion parameters, I wanna say, called Olympus, and that their objective was to surpass Anthropic. I'm like, okay, he's making it sound like it's such a- Like, oh, they don't care about Anthropic. Like, well, no, that's just classic AWS. They did the same thing with Graviton, and they're doing the same thing with their silicon chips. They're just barraisers. That's what Amazon does. So it's not either or with Amazon. It's like, oh, snowflake or red ship. Yes. You know, so that is just a misunderstanding of AWS. But having to comment on some of the earlier comments, that abstraction layer, you know, Amazon's never been really that great at those abstraction layers, and that's what Q was supposed to be. And Matt Wood kind of laid that out. You know, Q's supposed to be the easy button. And I think they're trying to apply it, like Q for supply chain, Q for data, connectors to Jira and ServiceNow and Slack. And so it's gonna be interesting to see, John, will they actually build those apps for the people who don't wanna build them themselves, or will they allow their ecosystem to do it? And I think, again, the answer is yes, both. I mean, there's no brainer that, I mean, Amazon has to have their own model at some point. Well, they do, they have Titan, but people say Titan's met, you know, it's like, but they, of course- According to Alex Heath at the Verge, he's reporting that in his post on end of March, March 29th, Amazon has Rowett Prasad aggressively performing goals to beat- Yeah, he's the Alexa guy. He was the head of Alexa who, Jassy said, you're reporting to me, let's fix this. He's running their AGI group. Yeah, which is- Right, which is- It may not be AWS, it might be Amazon, because Amazon uses AI and all their products. I think you're right, John. Not just AWS. Well, you know Jassy better than anybody. I think Jassy said, hey, this is too important. You're gonna report to me. I think that's the guy, the guy you're talking about, used to run Alexa and Jassy, he reports to Jassy now. I mean, I was talking about this last night. I'm gonna get you, I want to get your thoughts on this riff here. I want to riff on something for a second. So last night I was talking to someone who talked about the meetup we were at, all these AI experts in the panel, AI experts in the audience. And one of them was like, hey, so it's really exciting seeing everyone's talking about pitching their companies and a lot of activity, entrepreneurial activity, and a lot of comparisons to the dot-com bubble. You think it's gonna burst? I pulled your line. Yeah, it's gonna burst at some point. When? I don't know, I have no idea, but it'll burst. And I said, it's a real good bubble though. It's a good bubble. I said, not a bad bubble. And then we started getting on to AI and who's gonna be the winners and losers. And I said, if you remember the web, the internet web days, there were web pages and search engines. And you had DNS, which ran, and that was government funded by the Department of Commerce. And it was like a public utility, the domain name system, which is what the URLs were based out of. If you think about the web and AI as an example, the comparison to the worldwide web early days and AI, there's some similarities. You use the web, if you were a business, by building web pages. And you had to do that. You had to handcraft HTML, get a domain name, host it, put content on it, and people could browse it and use the website and do stuff, browse for information, and then ultimately buy a product or do something, navigate to some content. And that was an easy way to use the internet. Then you had search engines like AltaVista and then Google came along, which were harder to do. So with AI, it's a similar thing. We're seeing people are using AI, like RAG and retrieval augmentation generation, using AI for chatbots, co-pilots, agents, all changing the game on the interface side. And then you have other players building real hardcore shit, like Microsoft, like Amazon. And that's the distinction. Are you building AI? Which means you got to hire specialty talent. They all probably make it 300 grand a year and you're getting poached by every other company. You got to have real intellectual property in algorithms and hardcore deep tech. That's hard. So you're going to have suppliers of AI to users who are going to use AI. So if you think about it, that's going to be very interesting dynamic, David. What do you think about that? Because in a way, we have AI for theCUBE, but we're not building machine learning algorithms and we're using AI tools. So we're like the webpage. So what the webpage was was, you took advantage of the technology. So there's users and buyers and people are trying to figure out where they go and it's okay to be a user because there's benefits there. And so data and workflows that you could use and get value. So you think about the role of AWS and I'll put Google there and just Microsoft scares the crap out of me now after reading this report. But like one of the things that I learned this week, and it's so true, you'll appreciate this. I think it was, yeah, it was Matt Wood. I think it was Matt Wood. He said, you know, it's not the greatest analogy of the world, but it's like Swiss cheese. What do you mean? Well, the data corpus is like Swiss cheese when you would build AI and you build a rag, for example. When there's data there, it's actually pretty good, but the problem is when there's no data there, it's like this big hole. And so what the rag does is the AI will start grabbing from different pieces and make stuff up. And so what you have to do is figure out either how to fill those holes or how not to go into those holes. And so that's the kind of thing that AWS does and they do it really well. The other thing that I learned, he said that a lot of times these models, they're really good at the beginning and then they have a U-shape quality curve and they get kind of crappy and then over time they get maybe better. He said, we're flattening that curve and that's their role. They're like hardcore technologists secure it, make it better, give me tools so that I can apply it to my own corpus of data like we've been talking about on the power law of Gen AI. Power law of Gen AI is completely playing out. You see it everywhere. So I got to ask you something, John. It's kind of related to what you're talking about before you meet up. I want to get this out there. I've been meaning to put this out in the Kube pod. I've been holding it since March. I have a friend who's like- You've been holding it in? Yeah, but I just haven't got around to it. I have a friend who's like deep inside AI, like three letter agencies and like a handful of AI experts in the world and he's one of them. Like maybe there's a hundred that are like this guy. So, and he's really down on full self-driving has been forever. He's like, Tesla's is bullshit. It's like, so he's always been saying that. No, of course I don't really know. I'm like, I'm excited about full self-driving, but okay. So back in March, when Tesla, I was reading about Tesla scrapping its old full self-driving code, rewritten stack and becoming a learning system. This person's that their thing is, it's got to be a learning system. Why do we not let kids drive until 16 years old? Cause they learn organically, you know, their brains mature, but all of a sudden we're going to get, you know, full self-driving. Not going to happen. So I send this note by a text and say, I don't really understand, you know, if this is going to work. He said, this is what I was explaining to you when the last time we talked and what they needed to do. The way they were doing it was amateurish. It was just a bandaid. And then he said, I've been pushing the rope with them, meaning Tesla and others to get them to do this for quite a while, but I'd like to see a new sensor architecture. But so far they know better. And so I'm going to put this out there. And if I don't really understand it, but I'm hoping that either you do or somebody in the community does, he said their sensor mapping and spatial reconstruction are human perception based. Instead of machine perception based, that's wrong and will give inherent bias to any data analysis. So not truly 100% adaptive or heuristic, not to sound heady, but building a machine perception system is beyond most people's understanding, much less capability. So I said, meaning Tesla's ingesting and analyzing human driver video to train the AI? My understanding, this is a step up from hard coding, a series of if this then do that in C++, but you're saying this is still adequate. So look, for example, talking down to me, when using sensor data to reconstruct, say a room, the sensor data has anomalies that are somewhat sensor specific, typical sensor systems change in quotes the data to fit a human perception of what it should look like. For example, straight walls, square corners, but any time you change data, you lose fidelity. So if you're creating an adaptive heuristic learning system, I've been preaching, it is better to let the learning happen with unprocessed data, not with data change to fit the human perception. He's like, again, talking down to me, you getting this or do we need a whiteboard? And I said, I thought this new Tesla approach in just actual video. You should have said get a whiteboard. Wait, wait, I've got it. Next time I'm in town. And I said, but I thought this new Tesla approach in just actual video and has the AI to quote, figure it out. Are you saying the Tesla sensors are providing associated metadata that has changed to fit the human perception? And he just sent back a smiley face, whiteboard needed son. So anyway, I caught, I didn't catch it, right? I obviously didn't grok what he was saying, but I get it out there. But I'm hoping that you or somebody in the community who's got more technical depth can explain, but I'm gonna go do the whiteboard session. It's hard to grok specifically the way you read that, but I would say that what I, the countries we had last night with some of the folks in the room who were under NDA with Tesla is how they're getting the data is not conventional data. It's not like your classic database. So they are feeding video into it to train the video and that there's a lot of AI agent technology being built. And this came up a lot. Agent Cockroff was there actually. So I am again, he and I were talking about how agents are being used in infrastructure. So the whole AI ops field is changing. Back to Tesla and as an example to other environments is a complete changeover is happening with AI in the sense that with agents, with AI agents, the ability, not just chatbots, it's like agents that can understand data, repeated processes, identify patterns, and look at these kinds of things. And then take action and generate a response based upon something else is so new and compelling. But the challenge is that before generative AI, code was written to fill the gap for those things. Meaning like we had to build mechanisms in software to handle, to manage things to be in real time and do things with generative, with AI agents. You can actually make the agents work on behalf of things that were built for the wrong reasons. So those other things that were built free generative can be disabled because generative agent, AI agent will take care of it. If the agent's good. So if you build the right agents, how to manage data ingestion, looking at certain processes that are repeating themselves. I mean, a trivial example would be like a bunch of help desk tickets. I want to kind of, I can, after a while, you pretty much know what you're going to see. You get formats and you can come in, you can train some an agent to fix those. You can look at things like microservices and cloud, for instance, and say, we don't need some of those things we had to build before because we now have agents who can do those things and generate things on the fly. And keep things in smaller controlled areas. And it's better management. So you'll see things that were built for reasons prior to generate AI that will have to go away because we don't need them anymore because an agent will handle it. So that's the big conversation that we've been having when some of these tech circles is, what are those mechanisms that are no longer needed for the generate AI generation? And I think that's just like to go to the web example. I don't need to go to the library anymore. I can just go to a web page. So the library is irrelevant to me. So there's going to be things that will go away because they don't, we're not needed anymore. Or not convenient. So I buy that. What I would say to that is, in order for that to happen, the system of agency, if you will, taking action, you've got to have a proper data foundation. Sounds like bromide, but you do. So I think behind that vision that you just laid out has to be some kind of fabric and a knowledge wrath that has access to coherent and unified metadata. That's not like hidden to use an Amazon example because I know Amazon better, but they've got metadata that's stuck in glue. They have other metadata that's stuck in data zone and data zones that they understand this. But to be able to surface that in real time and be able to take action on it. This is why we always use Uber for the enterprise because Uber basically had to write its own software to do this, to make all these different disparate data elements coherent so that you can take trusted action. Because which data do you trust? Which data are people actually have access to? Some people have access to that data. Well, it's kind of the same data over here, but it's in a different format. Which one do I use? So that's a really hard problem. But I think the vision that you're laying out is the future data platform. When we talk about the six data platform, it has that element as part of it. It's a system of agency. So it's a great vision. And I think it's actually going to happen within the next five years. We'll see. I mean, we'll track all this. The great thing about it is, I was saying to the folks last night at the panel and other people in the industry this week that it's a great time to be covering technology. And that's why I'm excited about the New York expansion for theCUBE is because there's such a thirst and appetite for technology, deep tech impact. Companies need help right now to move the needle on their business. They need real research. They need real advisory. They need real consultative approach. That's why Ascension is doing so well. So we're in the picks and shovel stages of AI, but problematically the developer community is so going crazy with demand. It's the infrastructure that's slowing everything down. And there's a huge debate about this. And I'm definitely hardcore on this because there's no doubt in my mind from what I see, the problem with AI right now is the infrastructure is just not there yet to fully turn the, move the needle in a big way. The activity on the developer side is booming. The solutions are coming online. You're seeing things hit the activity of building stuff with AI, using AI is off the charts. The RAG stuff and the retrieval stuff, co-pilots and chatbots is proof points that people are using their data and their knowledge of their workflows to write software and use AI to scale it. At lower costs, maybe what took 100 people now does 10 what took 10 now takes one. So like the human involvement to manage it is changing. So that activity is off the charts and scaling fast. The underlying infrastructure is the problem. The GPUs, how much is going to cost? What do I stand up? I mean, the edge is not existed. There's nothing on the edge right now. So again, this is a huge opportunity for the Dell's, the Eula package of the world, Amazon. I even think Meta will be a major supplier in this. So, Google. IBM too, IBM's the other one I'm excited about. I haven't been excited about IBM stock in 10 years. IBM is perfectly positioned. I mean, IBM, I won't say got lucky, but they're, you know, so you hang around the basket, you're going to get a rebound. In this case, IBM got a big rebound. They've been doing Watson for over a decade. Now it's prime time. They got all that DNA and muscle in consulting. They got business transformation and they got this Switzerland position where because they don't have a cloud and they have kind of everything else, they can actually be the monster connector of all the other stuff. So I won't call them an integrator, but they can just be a great solution provider to everyone and bring technology to the table. So they're going to sell Watson X to, they're going to provide Watson X to a lot of difficulty. So, I mean, I think, And red hat on the operation side. So cloud native goes mainstream with Watson X. It's a great combo. So they got to play at the Adobe conference last week, the Adobe summit and, and, but I guess my point being, I think the AI trade is going to expand beyond, I think it already is beyond just the picks and shovels. I mean, obviously co-pilots are an example, but I think, you know, Adobe with Firefly, I think ServiceNow, Salesforce, even though, you know, it's not really, I don't know, it's like lightweight AI, but still I think these big software suppliers are going to, going to do really well. And I mean, I think anybody with an install base and a brain and some resources is going to do really well with, with AI. Yeah. Well, speaking of AI, I heard that Amazon is doing great on the compute side. Obviously they're getting their teeth kicked in by open AI and the market perception. And I think they're going to catch up to them in a way so that, but they're still kicking ass on as a cloud provider on the IS side, infrastructure side. Azure, obviously the security breach that you pointed out is a really black eye for them. But on the compute side and Google on the compute side, I'm hearing people talking about this big time that they can't get the compute that they need. So I'm telling you right now, there's a cold chill going through the industry right now from developers and people trying to scale up AI. I need horsepower, I need compute. And that's why I think there's going to be a surge for servers again, clustered systems, AI systems. So again, I was talking about this with Dell and HPE again, all past the month that you're going to see a massive surge in server sales. Why? Cause people are building a new AI clusters, spine and leaf built into mega systems. It's all going to come together. So a whole nother data center architecture wave is coming back. Not because cloud's irrelevant. It's that cloud operations now supports data centers for AI horsepower. So the cloud guys have to get their act together and to provide the compute and GPUs. Otherwise an NVIDIA will step up to the table, a core we will step up to the table or GROC has the inference engine will step up to the table. So you're going to see a lot of action Dave on this. And next week at Google Cloud, we're going to unpack all this. Let me ask you this. So Matt Woods said, it was really a strong statement. He's like, look, we were the first to ship H 100s. We have 400 instances, EC2 instances. We will be the first to ship Blackwell. So we're really making that point strongly. But you remember we attended that luncheon with VAS that was hosted by VAS that GTC with the Genesis cloud. So for my breaking analysis, I wanted, cause VAS is kicking ass amongst those alternative clouds, those GPU clouds. So I reached out to VAS and said, can you give me a list of those guys? Cause I don't have them at the top of my head. So Genesis cloud was the one that we met, but there's core 42 core, we've lambda, Neville. And I'm sure there's others that are raising tons of money. What do you think about those alternative GPU clouds? I think that's huge. I talked about this year and a half ago. We call that tier two clouds. Remember we called it the cloud power loss. You think they're all going to make it though? There's a lot of them out there. Well, I'll tell you why the demand is so high. So first of all, I think there's a was we're in a secure and a security, a scarcity crunch with GPU. So one, having a managed service with some sort of elastic front end to it, whether it's pay as you go, managing costs is going to do well for any company's got that because right now the demand for GPUs are high. And until the horsepower comes on premise where people can build their own clusters, you're going to see managed services being used to the extent that someone can't get compute or GPU from Amazon, Azure or Google, they're going to go to Corweave or if they got a better price, they're going to go there at their price sensitive and not locked into the cloud, AKA using the stacks of services say at AWS. So that's that's that. I think Matt Wood and Amazon are going to absolutely get it right in the history. If we look back on history, I think the bet that Amazon's making is we will never yield on our ability to deliver infrastructure as a service. Everything that's come out of reinvent in the past for years has been leading up to this moment where Chick Custom Silicon is leading the way and they have the horsepower to do it both on training and inference. They might look like they're fumbling with them at the LLM level with bedrock, but I think they just set in the table and know that no matter what anyone does, the game will be determined in the later innings. And I think that's going to be a play from Amazon. Google is trying to catch up to stay in the game, but if they can't get compute, that's going to hurt Google. So I think Google's doing really well right now. They just got to get their act together on some of these SLAs from what I'm hearing in the field. So, and they're doing good. So there again, Gemini, I think we'll be embedded in. People are trash in Gemini, but there's uptake. So it wasn't a launch that the press freaked out on over some bad data, which is understandable, but they could have done a better job not defending Google. I'm just saying that you don't judge them on that. You judge them on how fast they can stand up the models and who's using it and can they deliver compute? And that's going to be Google. So clearly, Dave, this is going to be the end game. The later innings will determine the winner in this game. I have another source that I can't, it's anonymous source, but I met with this individual in New York City. We had drinks at the bar. And I was picking his brain on, I mean, they're deep into the tech. Pretty big AWS account, but they do Google. And I was asking about Microsoft and OpenAI. He goes, I won't go near it. He goes, I go, wow, we use it. It's a better product actually. He goes, I won't touch it. I won't touch what? And OpenAI. He goes, nope, nope, don't trust it. I'm like, really? He goes, yeah, I can't put my stuff into, I can't use it. It's banned in my company. We're big bedrock believers. I said, what about Google? Google shakes his head. He goes, they got the best tech. He goes, they just don't know how to sell. And he goes, they don't know how to service me, but their tech's unbelievable. So he's big. He said Amazon shop number one and Google for AI stuff. And so that's kind of interesting. Well, I mean, that is, that's a problem for Korean, George Korean. They have had challenges building all the infrastructure for their business operations. I mean, Google's sometimes their own worst enemy from what I can see, they got a great tech, but their discipline has always been big back end, not a lot of front end, you know, classic Google search kind of philosophy. But they've been trying over the years. It's just, it's not an overnight success. You just snap your fingers and you got instant sales operations, reporting, field operations, full go to market. It's hard. It's not easy. I mean, if anyone can do it, someone from an Oracle background like Korean, but, you know, think about that, the task he had. I mean, when Diane Green was there, she got the product stuff together, but it was just too much product up leveling and too much business operations. Amazon's a well-oiled machine compared to Google. Microsoft, I'm probably sure it has the best operational go to market stuff because they've got DNA in the enterprise for years and decades. So I would say Microsoft's probably at the best machinery from the customer standpoint. Amazon number two, Google number three. And if you put Oracle in the mix, I put them at one or two because Oracle's a well-oiled machine when it comes to sales at marketing. I'm so disappointed with Microsoft. I mean, I've been so high on them because of the data just says how much they're kicking ass. But then, again, just reading this report, I'm just freaked me out. We got to talk about Intel. Can we? Yeah. I mean, like I said earlier, I was really kind of, you know, seeing people on TV just a week ago, oh my God, Intel's tailwind. They're going to be great. Oh my God. After that foundry, after the foundry, like, Bruhaha. So the stat is they, you know, to their credit, they're going to split out the foundry business from, you know, the design business. And they said they had a last year a $5.2 billion operating loss or two years ago. And last year was $7 billion loss on $19 billion in revenue. And they're saying break-even is three years out. And they want to have 40% gross margins by 2030. I mean, it just underscores what a hard and really crappy business, the foundry business is, but it's strategically important. Lawyer texted me, John, because you know where he stands on arm and Intel. He said Qualcomm X Elite Win Arm Chip seems to have found a lot of favor with PC makers, boasting 75 trillion operations per second, double exclamation point in red. IDC's PC forecast is plus 2%. Therefore, x86 volume has to be declining. The only way volume to get Intel to volume is to take very large arm orders away from Samsung and TSMC. And you got TSMC obviously, the gold standard. Samsung just upped its investment in that Taylor, Texas plant to 44 billion. And, you know, you look at Intel, the vast majority of its foundry business, if not all of its foundry business is internal captive. And it's like, so unless they can take huge orders away from Samsung and TSMC, which is a very low probability, they can't get to volume, which means they can't get the cost, they can't get the yields. And so again, I asked the question, if de-risking the supply chain for the United States for its critical infrastructure is the objective. Would it perhaps be better to invest in TSM and Samsung, who are way further along? Samsung, you know, says it's gonna be profitable now. You know, it's had a couple of tough quarters, but big investments, they're gonna, 10x their operating profit. I just read, would it be better just to invest in them and say, you know, look, Intel, go design. Who needs Intel for foundry? I mean, this is a bit, this is an important question that I want to unpack. It's like taboo. But is the important criterion, John, that it's manufactured in the United States or does it have to be a US company? Like Toyota has 140,000 employees in the United States. So that's, I think, a question that has to be asked. That's a tough question. I think, first of all, you brought this up last pod. Whether it's Intel, Foundry or TSMC, it has to be someone in America. I think America should have some foundry. But the government bailing out Intel is not the answer. Intel's hemorrhaging, right? So again, the question is twofold. Foundries are needed. You got to build chips. That's a net. I do think that's a national security challenge. I do agree with that. Don't like the bailout. I don't think Intel might have been a good call there, but Intel's already like seeing massive financial hemorrhaging on this. And Bernstein's piling on too. They're already down 13%, one of the biggest shareholders. So Intel's licking their wounds. So the bailout thing bothers me with the government. I think they could have maybe looked at a different scenario. That thing's just a subsidy. Subsidies don't work. And I'm not a subsidy fan of subsidy. You got to have a turnaround plan with real meat on the bone. So that's one. Number two, you don't know what you don't know. For example, the market on semiconductors, for the people who know about semiconductors, and this is something that you got to be really careful of. You're going to be an analyst. You got to get it right. Not just say, you know, got no business being on TV saying Intel's going to be a great company. And then they just crash and burn. You shouldn't even be invited back. But if you took up the financials of the semis, the AI and allocation optimization is going to be critical because you have a supply chain problem. This is the key business problem. You got to deliver the chips and the foundries are going to be backlogged. So you can design all the chips you want. So there's a need for foundry. The question is it takes too much time to ramp them up. So this is a tough call. I don't know what to say. I mean Qualcomm, Broadcom, they're all building chips. They all got to have supply constraints. I mean, if you get the allocation right, you're going to have a good numbers because you'd be just shift. You know, the margin, you can manage the supply chain and the demand orders. And so I don't know, Dave, it's just like... I mean, history shows that number one in the market and a big market does really well. Number two does okay. Number three essentially fails. I mean, we're seeing that in cloud and the cloud's big enough and Google's got enough other business that it can sort of hide its deficient cloud business compared to the other two. But this is the classic case. You see it in so many markets. And if Intel can't become like a viable number two surpassing Samsung, but Samsung has to keep up with Apple. And it's got light years of experience in building advanced chips. And so not light years, but I mean, it's well ahead. So I think realistically, at least Pat is realistic. Like I'm not going to catch TSM. I can maybe be number two, but if he doesn't get to number two, it's going to be just a pouring money down a sinkhole. And that's really, I've said it before, Intel could go bankrupt. But the only reason, the only way they won't go bankrupt is subsidies. And I think you got to ask the question, is that the right use of taxpayer money? Also, they'll make like take Qualcomm for instance, Qualcomm has many different like platforms of chips. So the allocation of the resource to the different chips matter too, right? So it's just, it's just, it's a tough game. We'll see. I mean, we in talking to the Broadcom folks and Intel folks over the years on Qualcomm, the semi guys know the business. I mean, they're in the hardware game. Everything's about turn, next cycle, next front, next improvement, moving the units out there and being ready for the next generation. So we'll see. And again, Intel, we'll get to keep an eye on those guys, continue to keep an eye on it. The good news, Dave, we got it. We got everything right and we're not even getting access to Intel. So, you know, Intel is not really embracing the industry, engaging the industry analysts on this because I think they're just trying to feed the market PR. So it's interesting to see Intel's analyst relations strategy around this. I mean, I don't know. I mean, I think, again, lawyers been on this since 2011 when PC volumes peak, it had a huge influence on me because he educated me on this stuff. And I mean, we're doing another, we're doing a breaking analysis this month to unpack Nvidia. Obviously, we'll talk about Intel, ARM, alternative XPUs. He's somewhat down on the chiplets, as you know. I've been testing that, but I think there's a real market for it. So I'm going to push him on that. And I think he thinks that the edge opens up a lot of opportunities for specialized processors and XPUs. So that's going to be fun. Well, anyway, let's, we're getting down to the end here. I want to put a quick note out there. We're going to be at Google Next next week. Google Next is their cloud event. What's interesting, Dave, about this, is that Google Next is, they just had their event and they moved it up this year to make sure it's in Vegas this year, not San Francisco. Again, one of the companies that are moving, moving the event out of San Francisco because of all the issues there. But we have a great lineup. We got, we have a ton of leaders in the ecosystem. McKinsey, Accenture, Automation Anywhere, Quantify, Elastic, Deloitte, Google execs coming on, top, top execs coming on, PWC. We have ARM coming in, first an SVP of infrastructure coming in from ARM. So I'm going to talk about semi-conductor, got ARM coming on. We're going to have ACL coming on. We're going to, a bunch of CUBE alumni. And we're going to have J.O.T. Banzi, who's the CEO founder of Harness, who's also a VC. We got the chief product that was from Box coming on, Diego Dugakin, and we got the CTO, the field CTO of Databricks, Gabe Monroy, developer experience VP at Google Cloud, Bobby Allen's coming on. We're going to have a whole day dedicated to analysts. We're going to have five segments dedicated to analyst angles. Oh, nice. Analyst angles plural with an S. CUBE collective folks. So the CUBE collective are open, United artists of analysts, United analysts model is coming together nicely. We're going to have actually three dedicated segments to non-Cube research analysts. And of course we'll have our CUBE research analyst angle on there too. So you're going to see a lot of great sponsored content coming out from some of the best leaders. We're going to extract their stories and share them, but we're going to have great CUBE editorial and reporting and analysts coming on. So Google next should be great. And of course this month we get SAS and then next month the Red Hat Summit, Bumi World. I got two kids graduating from college, Horace A, Alter-Rig, Delta Tech World, IBM thing, Medica World, Memorial Day, Dave it's going to be summer and it's going to be like, what the hell just happened? I know. Hey, just a quick, I want a quick hit rubric filed for IPO and they're going on the NYSE. And it doesn't look great on paper, but I think it's actually fine. They have $6.3 billion valuation, they're losing money, but it's because they're transitioning from a perpetual model to an ARR model. They're ARR business, subscription business, growing like crazy. And so I actually think they're going to do really well. And I think we're going to be interested to see if Databricks goes, if Arctic Wolf goes, and then maybe eventually Cohesity, which by the way, speaking of Arctic Wolf, we're going to have RSA is coming up as well. And by the way, we've got official word now that at RSA, the Monday we're having an event. We're hosting an event with the New York Stock Exchange and Intel Capital and some two other companies with us we're going to be hosting an event, open policy, Intel Capital, and waiting here for a few others. The Cube will be headlining a party, invite only. It's going to be at the Lamar in San Francisco, the Monday happy hour from five to seven. If you're going to RSA conference and you're going to be there, send me and Dave a note or one of us a note and we will send you a link to sign up for a special VIP happy hour. If you're a company scaling up in the security space, got some general AI technology, we'd love to talk to you, have you sit and mingle with the Cube event. So great to have these happy hours, Dave. This is going to be, usually we go to the happy hours. Now we're having our own. That's going to be awesome, John. Yeah. All right. Well, let's sit for over an hour and have a good rap here. Again, Google next is going to be great next week. It's going to be, I'm expecting to see massive amounts of AI announcements, but we'll be squinting through those analysis. We're going to ask some questions about the infrastructure. We're going to have Mr. Lohimer back on. You know, he's awesome. He's former VMware Dave. And we're going to have all the infrastructure guys coming on as well. So we're going to, we're going to get them. We're going to get the stories. So see you next time. Thanks for listening. That's Q pod 53. Thanks for listening.