 Welcome back, everyone, to theCUBE's live coverage in San Francisco, California, theCUBE's live Google Next 2023 coverage. This is day two winding down. This is the analyst angle section where we go in and we break down, we analyze, we look at it, we kick it, we analyze, we critique it, and we see where the action is and where the needles are moving, where the game is being changed. I'm John Furrier, your host, Dustin Kirkland, contributing analyst with theCUBE. Welcome to the show. Thank you very much. We start with you, Joao, welcome back to theCUBE. Good to see you again. Thanks, Joao. As usual, great guest contributor appearance. Thank you. Rob Stretche, heading up our new CUBE Collective Analyst Team. Rob, as usual, great day today, a lot of great commentary. Yeah, it was great. Let's get into it. First question, are we drunk on Genevieve AI? Okay, I mean, I'm feeling it. I feel buzzed. Sure. A little intoxicated, a lot of action. Yeah, I don't know. Are we drinking too much Genevieve AI Kool-Aid? I'm going to say maybe we've had a few too many cookies and a little too much milk, okay? It's maybe not to the point of abuse, but we've been talking a lot about it. The thing is that this is a once in a lifetime, once in a generation change in the way we fundamentally approach technology. And that doesn't come along every day. John's been comparing this to the dawn of the internet in the late 90s. And I think that's apropos. I think there's really something here that is pivotal and will never be able to go back to writing code without an assistant, reading 1,000 page documents and manuals without something to help us summarize that and digest. Rob, what about you? You drinking the Kool-Aid? I think it's in everything that they're doing here, they've weaved Genevieve AI into every single platform. Duet is in everything from Looker, BigQuery, through all of the workspace apps. I think we are a little high on Genevieve AI, but I think to Dustin's point, maybe for a good reason for certain things, I think it applies more in certain areas than it does in other areas, but writing emails will never be the same, right? I mean, Genevieve AI is there to absolutely help with that kind of stuff. So Archie, we're going to ride the wave, hype wave? Are you drunk on AI too? I am actually, I have seen a couple of swings in the whole industry since 94.com as to why it happened and why some people got lost. It's a hype cycle. During the hype cycle, the warning to the vendors is that don't swing the pendulum to the new shiny object. Stay practical, do not shun your existing customers. They have real problems. The brown field is important. Yes, this time, this generated AI is real on the creational side. When the humans are involved, you can use these models to augment humans, right? But when it comes to automation, there's hallucination problem. You can't take these models close to the systems of a record or automation, which doesn't have human looking at the logs or machines, right? So you have to be careful there with the messaging that you don't want to alienate the CIOs of these companies which have a lot of brown field workloads. But having said that, the duet is everywhere. The demos were great. The narratives has shifted. I call the developers are not maze runners anymore. As a developer, I was like a maze runner. I need to take a look at the login, the identity, persistence, so many things I have to do. I have to do less of that now as a service services, as well as with a lot of help from AI to cook up the code routines for me. And refactoring demo was great from yesterday and today as well. Well, I'm kind of old, so I'll date myself. When I started in college in the 80s in computer science, I will never forget, I would say the older guys, hey, you're the punch card generation. And I never once touched a punch card. And people don't even know what punch cards are. That's how the mainframes were coded. Dustin, this is a punch card moment where you don't talk about that maze running. That's old. Now the new way is going to be completely different. And by the way, structured languages at that time were awesome. And that's just changing. I think it's easier and easier. Yeah, it's fundamentally changed. And that's the thing, we'll never go back to punch cards. We'll look at it as antiquated. We kind of nostalgically think about, I don't know, Romans running around in togas day to day. That's the way we think about punch cards. It's this anachronistic sort of thing. We're going to look back, I think, at a moment before AI was able to generate really intelligent answers to queries, really high quality tests for code. I want to flip this on its head though and ask you a question, John. So, you know, if maybe there's too much discussion around an AI here at this Google event, is there maybe not enough focus on AI certain somewhere else? You mean like, I don't know. AWS? Yeah. Well, I mean, I think the question everyone's asking is, did AWS miss the gen of AI? And I know that's come up a lot. Again, I've talked on theCUBE about it all the time, Rob, you know I've been my rant here. Amazon Web Services definitely missed the gen of AI wave as we know it in the hype. Okay, so start talking about a lot of things that are going on. But remember what's happened in the past 18 months. Amazon's been on a growth rate that's been second to none in terms of revenue growth, client growth, just thundering performance and everyone knows we've been documenting that. Internally, I've had conversations with Andy Jassy, Adam Sileski, Swamy, all the top people and some of the product managers. They knew about gen of AI. It's not like they weren't doing it. They were doing stuff. They just didn't see it as an important point at that time. And open AI and Microsoft pulled off the marketing hype CUBE of the century, which is by organizing a press event and doing the 10 billion, whatever the number was, was a 10 billion dollar investment. That became such a shiny object, if you will, even though it was not, but that woke up mainstream. Once that shoe dropped, they were flat footed. And that's what happened with Amazon. So it wasn't like a bad decision. I asked Adam Sileski last year before re-invent about data and platforms. He actually brought up LLMs, unaided awareness on that part. He said, yeah, LLMs, he had it in his talk track, but it wasn't like we're investing in it because they're the number one. So there's really no story there other than the fact that they missed gen AI and then they quickly pivoted and said, we are not going to get caught flat footed and they don't want to pull a Microsoft and miss a big wave. And so we will see what happens at re-invent, Rob. And we're going to see if AWS can step up to the plate and, but remember, that's a few months away. This shit's changing so fast. Yeah. Well, it'll be interesting to see because they actually promoted Swamy to the S team as well as part of this over the last couple, I think it was last month. And if you start to look at it, like you said, they've had LLMs for a long time. I mean, Titan underneath, Alexa has been there forever. And that's came out of like nuance and the people who built that. So it's, but to your point, the usability of it other than for voice was missing. Amazon has historic run and the key point of the story with AWS, and this is may or may not be well known for the folks that's just jumping in now, back when Amazon Cloud was growing like a weed because of a major factor. If you wanted to provision a data center viewer startup, like Dropbox, Airbnb, Twitter, they all started on Amazon because your choice was to provision by gear as cheap as 10 grand to 100 grand just to get the app up and hosted. And then so it was easy and no brainer to put the credit card down and go to the only game in town, AWS. Okay, so that's why they won. And then they took over the enterprise and natural growth. But today, Amazon's weak link. Their only, I think their blind spot is if they miss that startup wave, with right now, this culture, if you take someone in the dorm who's 22 and say, here's the console for AWS and here's the console for Google Cloud. I guarantee you, nine out of 10, I'll be like, what's AWS? Okay, I can back up you over here. The experience matters actually, Google being BDC company from DNA, they know experience that I think better than the. Do you disagree with my statement? That if I. No, actually I do, partially I disagree. Okay, AWS is a good contender. They number one right now. Yeah, I don't know. Actually no, AI as well, AI as well. Because we're shifting, right? In that shift, they're not going to let it go as easily as we think they will. Because a lot of, okay, look at this. Google is doing AI, of course, but a lot of ecosystem around it, a lot of startups are doing AI as well. VM where we saw last week, they are training their SDKs and their code actually snippets and CLIs. They have a model, so that can sit in AWS as well, in Azure and in Google. So Google will train their own sort of Google stuff, which when we talk about the creation side of things, right? Yep. Enabling practitioners of Google Cloud, they will do that. And Amazon will do theirs. I think the. The question, I mean, still. The benchmark. Okay, go ahead. I think Google still has to push harder and keep pushing if they want to get that, you know, like number two seat. Okay, Rob, you worked at AWS, you worked at Google, Dustin. Amazon's number one, they've been growing. They have a lot of large enterprises. They are the number one cloud. They have a lot of stuff that they've solved. Now, they're there. So they're in a good position. They're not hurt, believe me, they're kicking ass. So what's the benchmark for AI dominance today? What's the scoreboard? How do you measure whether someone's winning or losing? Because they did more press releases? Yeah. For a moment. I think AWS has a challenge that they can overcome potentially. And that challenge is that each service that's built there is an individual service. Unlike here where they're building solutions, GWT is integrated into everything here. Maybe too much so, but it's in every product. The way that AWS services, even within the services, like S3 has three engineering teams. Integrating AI into S3, if you were going to go and do that and say, hey, I want to be able to find my objects, it's going to be tough. I think that they have a tougher road to have as pervasive in AI than Google does or Microsoft for that matter. I think Amazon could still be an AI infrastructure dominant player with bedrock in that space, but I don't think it will be as pervasive as it is with Google or Microsoft. Doesn't which it could. Yeah, I'd break it into kind of two pieces. One is the marketing piece. And in that one, I agree with you to a large extent, AWS has missed the marketing hype cycle for AI. Several questions, whether that matters or not. Go look at the run rate and the business and I'd say, no, it doesn't matter all that. They're doing just fine. The longer tail is a different question. But Microsoft nailed it, Google nailed it. As we've said over and over and over again, much of this AI stuff is old. Alexa's a great example. We've been asking Alexa for directions and recipes and that's AI. Duet is kind of just a really fancy clippy. We had clippy in Microsoft word years ago, right? Now, what's changed is that, wow, I mean it's just so much more powerful than, hey, can I help you put a period on into that sentence? So there is a marketing angle. The big piece though, I think you've nailed it, which is if seven out of 10, eight out of 10, any number north of five out of 10, if a majority of kids out of college, startups getting funded are doing the majority of their work on TPUs, on Google Cloud, GPUs, Google Cloud, more so than Microsoft or Amazon, that fundamentally changes the landscape for a decade or more. And if you look at this market, and by the way, Amazon never had competition before like this when that startup wave was the proxy, was the canary in the coal mine, for who will win, obviously they grew with them. If Amazon loses the startups, I think they might lose the generational game of the next gen cloud and that's a risk. Now I'm just saying, we'll see they have startup programs, so that's an area we're going to watch it reinvent. The next question comes down to, what does the scoreboard look like for winning in AI? Is it production workloads? Is it revenue? Because right now the developer market, to me, is where that focus should be, because that's like, to me, that has always been in this market, at CNCF and open source, the developers are the ones voting with the collective intelligence of their code. And that's the de facto standard body in my mind of what gets adopted. If those developers are not coding on AWS, they won't be there. And that means the enterprises will follow the consumer developer market, which is open source. So that's to me the dots that kind of are weird to connect, like you can't just say we're going to put a service and use our power of our machine. If the developers don't like it, it won't hunt, the dog won't hunt. So I believe that Amazon's got to focus on the developers better than they are now while doing the enterprise game, because they're doing okay at that, they're not hurting. I think developers, masons of digital empires, right? They lay every brick and build these things, right? So AI is in hype cycle, lot is very real as well. It's not just fake hype like Web 3 or crypto, you know? Those were just fake, you know? So this is real and definitely AWS has to put their act together. There's no doubt about that. I'm sitting here watching you and you guys from this seat and you guys were over here too when you're doing the interviews. Behind us is the hardwareverse. It's where the TPUs are. People are getting selfies next to it. I mean, come on. We are in a world where speed matters, speeds and feeds. Amazon's got great infrastructure. They're always optimizing the physics from James Hamilton's days, the Peter DeSantis here at Google, they got great pedigree on infrastructure making, you know, from cooling and power to IO transfer. So the question is, what's the future look like for the next gen cloud for AI? Is it TPUs versus GPUs? Is it going to be embedded in the software? That's going to be a very interesting question because the dominoes will fall and storage comes up only once in the entire conference ride. No one says we're worried about it. I was going to say that made my head explode in a positive way today at least. I mean, when storage was brought up, but it just shows you how buried some of the other announcements were. Like that Ally, Alloy, DB, Omni can be deployed on any Docker container platform. It's a Docker container. It can be on bare metal. It's running on laptops at people's houses. The fact that Google is really pushing it all in on that and really trying to decouple and bring their kit other places, we talked about it with the Google, the distributed cloud, Google distributed cloud where CSPs are running it. And I think to me that's a big win for them is that they're taking a different approach. They learn from the failures of outposts. And they're taking a very different approach than even Azure Stack, which is done okay. Well, we're going to have a few months to prepare for re-invent. So this team here and all of our analysts and our collective, we're going to be looking at all the corners of the landscape of AWS and we're going to be inspecting where they're going to infuse AI, how that's going to play out. Because we will see how they play their hand, Rob. When their cards come down to re-invent, we're going to see what they're going to be holding relative to their GNAI capabilities. Well, I want to shift gears. Oh, you want to continue? Yeah, actually, yeah. I think on that note, the channel conflict, like you guys talked about that yesterday evening, right? The channel conflict is more here on Google side because they are cooking up the sovereign cloud and also cloud in your data center, like equivalent of outpost, you know, if you will. So two different flavors of infrastructure. For countries, they want to be sovereign, for companies, which they want to have to keep everything private. So it collides with VMware, it collides with Red Hat, it collides with so many other- We talked about that yesterday, Jeff, and I think I've always said, Dave, and Dave and I talked about this years ago when we started theCUBE, EMC had, VMware had an ecosystem. Ecosystems is about enabling them to create value to make money, right? Get me some distributions, my products, get some customer value, get paid for it, and when you start having conflict, sometimes it's good conflict, we thought about yesterday, and I think when the tide is growing, growth hides a lot of the barnacles, if you will. When the tide pulls out, you can see where the barnacles are and who's going to close on, who's swimming naked. So, but that's a different, I think Google's fine. I mean, to me, the thing is, is that what I'm thinking about all week here now is the AI cloud was my thing yesterday about what I'm thinking about that day. Today I'm thinking about this wave of abstraction, and every wave, whether it's punch cards, web, it was a game of simplifying, reducing the steps it takes to do something, whether it's productivity, like you said, the maze versus simplicity, and making it easy to use, right? So simplification, reduce the steps it takes to do something. So the area is, the question on it for you guys is, where do you see that formula hitting? Yeah, I see data analytics, there's all these areas of storage where you can reduce, abstract away steps. Yeah, I got strong feelings on that one. I've got a personal mantra that's written across my whiteboard at home, and it's been there for years, and it's two words, reduce complexity. That's something that as a product manager, and as a leader and as executive, I've always challenged the teams that work for me to do. Anytime we can reduce, remove unnecessary complexity, especially. Now, I think where we're at right now in the cycle is an incredibly complex set of solutions. I'll draw a very quick analogy, since we're here at the Google Cloud event, Kubernetes, Kubernetes, Kubernetes, we spent years, I spent years working on Kubernetes, incredibly complex, and way too many people worry about Kubernetes they need to. Meanwhile, in the last four years, I think Google has done a really nice job of abstracting that away into Cloud Run, Cloud Build, write some code, deploy it, and it runs, hey, guess what, there's a Kubernetes underneath it, but who cares, you don't have to know, and I think that's a good example of reducing complexity. Where I think we'll go over the next six months, and I think some of the winners will emerge, will be those who are able to reduce the complexity of, I don't know, how many APIs did we have to learn the names of? I mean, it's amazing all those announcements in the keynote, but the fact that you've got to hook up your Vertex AI to your Palm, to your, you know, you get some Lama models, and a lot of that is going to have to be combined. We're the areas, we're the areas that you can jump right out and just jump off the page, we're the areas that are screaming for simplification and ease of use. Multi-Cloud, right? Multi-vendor, Simplicity, Multi-Cloud, Simplicity. Remember, Google was supposed to be the Multi-Cloud, you know, where's that narrative? So Google is known for like shunning the old narrative and moving on to the new narrative is very fast. I still want them not to do that, keep the core intact. Okay, you love open source, stay there. You still want to be open and enable multiple clouds, stay there, do not shun old narratives. Like I have, actually I'm doing analysis at home which I'll publish tomorrow end of the day, that what was said in last two, three years, which was not even mentioned this year by Google. So, yes, that's the- They're going to pull the plug, Rob, final word here, get the word in, officially, where's the spots? I think it's all about how you pick your models, what's the right model, how do you know which framework to use, and all of the different stack that goes on there to Dustin's point and you know, sorry, Jeep's point, it's still super complex. I think it's really somewhere it has to get simpler. It's the new enterprise model, it can't be the old one which will solve complexity with more complexity. That's the old way, that's the old enterprise way, software, everyone knows that, don't do that, it's making it simpler, that's the cloud way. Just briefly, we talked to the people who train the partners, they are telling us that it's very complex problem to solve. Training partners on the new stuff, it's very hard. Guys, thanks so much, again, sorry, Jeep, thank you for being a great contributor. Dustin, thank you for coming on for the first time. We're going to do more of this and appreciate your contribution and of course, we've got theCUBE collective. Rob, are you going to put a quick plug-in for the collective? Yeah, so we're really ramping up, starting to ramp up things and get things off the ground and we have a lot of really interesting services that will be coming around that will be partnering with that collective to be able to provide, so really excited. And Sue, I'm super excited because theCUBE is a unique point, we talk to great people, we get the data, we want to share that with you and having a community that can work together to roll it all up is something, it's been a big mission of ours for many, many years. Dave and I and the team, so guys, thanks so much for a great analyst angle. And we're going to end day two, we'll see you tomorrow, day three in theCUBE. Keep it right there, look for us tomorrow at theCUBE.net and SiliconANGLE.com. Thanks for watching.