You've been a warrior, holy cow, over 40 plus interviews exclusive with Andy Jassy, which we're going to reveal here on the Cube. So stay with us on this podcast. You'll hear exclusive interview with Andy Jassy, who came back after a two-year hiatus from coming to AWS with the show he built, reinvent. Obviously he turned the world on fire, disrupted the old guard, created the thing that is cloud computing, godfather, and trained the entire business community on how to do cloud right, how to work backwards, and how to create these high level services. So Andy Jassy here on the Cube. Stay tuned, that'll come on shortly. You know, we're here just grinding away, getting the stories. The full team is warfare, getting all the puzzle pieces, squinting through all the news and getting all the analysis. And of course, the Cube team here on a what I call very agile setup. We are in the press area and we've been going full throttle for four days, so the team's been amazing. Andrew, Christian, Alex, Anderson, great job guys again pulling the Cube off. But we're totally stoked because it's just been on fire here at reinvent. And the news around the internet, Intel, we're going to get to Veeam financing and all this good stuff. But AWS dropped its holiday charm on the industry in classic, classic AWS reinvent fashion, going back to how it used to be.
The holiday, all these gifts are coming. I want to unpack, I want to play with all the new new action. It really was a sign of Matt Garman's strong leadership. You Jassy trained him well on how to do these keynotes. So everyone was looking at Matt Garman and saying, hey, will he have it? He brought it. Oh, he totally did. Surprise guest Jassy comes out, completely changed the energy in the room. He just delivered, 'cause he's a pro, he's been doing it forever. Up until two years ago, this was his show and he was the master at laying down and he handled the big announcements. And then also flexed a little bit about, you know, subtly saying without saying the cliche, we've been doing AI before AI, but he was flexing with data. Here's what we've done at Amazon, here's what we're doing at AWS.
Goodies. It was back to the brown hardware. Dave Brown was part of that Monday. I don't know if you saw that Dave Brown on Monday night, they gave a master class on storage and JBOD and how they're using Nitro. And you know, there was amazing thing on Monday night. Monday nights are just they're always awesome. The crowd was packed, nobody was on their phones, and it was like hardcore geeking out and everybody was just like focused on what Dave Brown and then Peter was saying. I mean, it was just so Monday night. I'm in the hallway and I'm going to another event. I missed the keynote, I had a meeting, but I was with someone from a big company that's first time here at reinvent. And we're walking through the main exchange there and it was just a mass rush of people.
Concert? No, no, it's a keynote. Like now on a Monday night. And this company, they have like one keynote they hope everyone shows up. But no, no, that's like one of four keynotes. And so that was kind of a pause moment for someone in the industry to get a feeling of what this conference is all about. Like there's keynotes that are packed and a stampede overflow rooms for every single one of them. Peter DeSantis, that's the hardware, the one that James Hamilton used to do, that's all about what's under the hood. The keynote main one, all the announcements, Matt Garman, Andy Jassy, Swami does the whole data with Ruba and partners. And then obviously Werner today with the developer piece. So every single one of them is intentionally driven on a script that's packed and it's always packed and it's so detailed.
See AI and analytics coming together because they're using the same data and then right into the weeds. And people eat it up. And it was great to see Andy back because I think Andy and Matt are teamed up. It's like two power players saying, look, we're going to win this game. Let's put the game on. Andy's competitive, so's Matt Garman. The whole team's here. Swami's a great leader. You got Swami heading up AI. Andy back on stage saying, I'm on the team, coach, put me in, Matt. I'm here to help. And Matt's like, I accept all this, no problem. He didn't come out and do like the Santa Claus wave and the Macy's Parade. I mean, he was out there for 20 plus minutes going deep into the tech, taking us through all the Nova models, right? Introducing, no gimmick at all. He's like, I'm on the team. We as a Amazonian cult are going to win this. And so that was great to see that.
Just Q&A, right? They come in, they say for an hour, let's talk Q&A. And then Adam, Andy and Adam, their answers were always long, you know, and pretty detailed, which was fine except you didn't get as many questions in. Matt was like, he reeled off so many questions and home run. He probably got like, it was a home run derby. Yes, it was like the All-Star Game. Ortiz, boom, there's another one. Boom, there's another one. You know, he didn't waste a lot of time, he didn't go off topic. It was really, really good. And the favorite moment too, obviously for us, is that Andy took time out of his schedule, the guy is super overbooked, to come down and recognize the Cube and sit down with us for an exclusive interview. You're going to hear that here first.
You wrote the trillion dollar baby, which was epic. And I think that you know, I think he got that we got the cloud at the time. We've gotten to know Andy over the years as you know, me and you have been tight with him. And so you know, he's also very competitive, he's also sports. I like hanging out with Andy. But I wanted to try to stump him on something. So before we went on camera, this is good, tell the story. I went on camera, I'm like, okay, I got some, I'm gonna school Andy right now. Okay, 'cause he's like a schoolyard boys flipping baseball cards or like we used to do, you know, arm wrestling. Okay, I'm gonna one up, yeah, right. So I say to Andy, Andy, this is our 12th reinvent. I think we've been to more reinvents than you. Of course we have.
Jassy, because it was kind of a fun comment just to kind of needle a little bit on that. But said, but like, hey, you know, we got you. No, no, no, I'll just delist you. Yeah, I'll just de-invite you. And the other thing too, he said, and people will hear this, he said, yeah, when we first introduced S3 in 2006, people like, what are you doing? Like this makes no sense. And I remember you know, being there, I and Flore were like, oh my God, this is going to change everything. This is incredible. And because we saw storage networks try to do it and do it wrong. And then you know, so that we were on it obviously early. And so that I think we made a bet early on on this ecosystem. Yeah, it's pretty obvious we've been on Amazon since the beginning. Let's get to the Andy Jassy interview a little bit later, so stay tuned. You're going to hear that. We'll do a little intro on that, but let's get some of the news.
Gelsinger essentially retired, quote retired, canned maybe by the board, some say. Oh no question, he got fired. I mean, they let him retire. They said, you got a choice, I'm sure. Retire or you're going to get fired. I mean, there's no way that happened that fast and that wasn't a firing. Right, but it was his choice to how it was played, right? I'm gonna retire versus get fired tomorrow. So yeah, I mean, take us for your analysis because again, we've been on this. So if you listen to this Cube, you know that Dave and I have been talking about this for a long time. Other analysts that are out there with cold takes were like, oh Intel's got a tailwind. You know, putting lipstick on that. Intel's been trying to market its way out of the problem, but you can't.
Intel's problem started in 2012 when PC volumes peaked. And you can go back before that when the real problem started when they passed on the iPhone because they said there's no market there and there's not enough business and we can't make money. So ARM stepped in and said, okay, we'll do it. Well, the reason why those two points are really milestones are really important is because when PC volumes peaked, that set in motion ARM volumes overtaking PC volumes and they did. And so in semiconductor manufacturing, volumes win. And we've been saying this over and over and over. Everybody knows Moore's Law. We talk about Wright's Law. I'm not going to go into Wright's Law, but look it up.
Split off foundry. We wrote a piece of special breaking analysis, Flore, saying no, now is the time. That years ago was the time, but they're going to run out of money, right? So unless we said at the time, unless Intel has unlimited capital, which they don't, they're going to go bankrupt. And so that's what was happening here. They look at the numbers and they said, this is not going to work. We have to do something different. So now I'm not sure what's going to happen. They're talking about fiscal responsibility. Fiscal responsibility is they got to figure out how to spin out foundry so that they won't continue to be an albatross on the design business. And I think here's my scenario. I think private equity probably has some interest. My scenario, which probably is not going to happen, is that I think that Apple and...
Spin out foundry and then recap it and then save Intel. And it's going to take 10 plus years to do that. And I say as a quid pro quo, the US government should get its foot off the necks of big tech and big tech should step up because they have skin in this game. But I asked Matt Garman in the Q&A what he thought about that. I said, is it important to have US-based manufacturing? Yes. Is it important that that is a US headquarters company? He said, no, I'm not sure about that. And then, would you be willing to fund it as I just described? And he said, I'm not sure I'm willing to fund it. So they don't want to do that. Obviously, all these guys are essentially getting a free ride.
VMware, Annapurna are definitely up there. Those two have to be right. Nothing else is even close. What else do you put there? You know, PWC saved IBM, Sun, and Oracle with Sun was very interesting, that was an IP transfer. But yeah, not even close to what Annapurna is throwing off. It gave them Nitro, it gave them Graviton, Inferentia, Trainium. I mean, their whole playbook of silicon, amazing acquisition. They're getting good at it too. So free ride in the sense that they had the foresight to buy that company and get off the x86 crack and do that. That's what I mean by free ride. And the ARM model being fabulous and the fact that TSMC has done all the dirty work.
We save Intel. And is it important to save Intel or do you just basically say, do what happened to GlobalFoundries, say give up on advanced manufacturing and then give it all to TSMC? Which look, Frontline just came out with a new series or a new episode on the rise of Xi Jinping. Basically he says, I'm gonna take Taiwan either by force or by coercion. It's going to happen. So something's got to give in the United States. Another Cold War or a war could potentially be there. It could be a hot war. This could be a national security. Foundry has to be in the US. I think so.
Can they get across the finish line? What time frame could this be up and running? And does the market shifts? Does it matter? The time frame is literally 10 to 15 years. And who's got the stomach for that? And then Matt, when he was giving the answer to me, he was like, Intel's on a path. And I was like, I made a face, you know, sitting in front row. And he looks at me, goes, well, you know, they got troubles. They are not on a path. They are on a path that's a spiral downwards because of foundry. And so something's got to give there. I feel like it's really, really important that the government and these big tech companies step up and save Intel. But nobody's got, why do they want to spend billions of dollars doing that? It's a corporate development play right now. To me, foundry, Intel, it's all corporate development.
A $50-60 billion company with this $80-90 billion valuation. So you know, something's got to give there. Maybe private equity takes them. If I were private equity, I'd take them and say, okay, let's do what GlobalFoundries did. We'll just make chips for cars and TVs and make money. Marvell had a pretty good quarter. No, Marvell had a great quarter. Talk about Veeam raising a $2 billion round ahead of an IPO. So really interesting here. Let's get the scoop on Veeam.
Veeam was owned by these two Russian entrepreneurs for years and they would have these crazy parties in Vegas and they were awesome. They own VM backup and they were brilliant marketers. They named the company Veeam and they started in...
Insight for $5 billion just before COVID, so this is 2020. But it was all kinds of convoluted, you know, Russian stuff back then. Russia gate, all this stuff was going on. So Insight had to clean that up and Anand Eswaran, the new CEO, he had to come in and clean that stuff up. And they couldn't even do business with the US government at the time. Now they have a huge US government business. Well, anyway, fast forward to 2024. Veeam just announced on December 4th that they raised $2 billion. That Insight is selling $2 billion investment in Veeam at a $5 billion post. So I did the math on this. Back when I was meeting with Veeam and they were giving me all their metrics, I came up with...
Crank up its growth a little bit. I think that's going to be a rocket IPO. Oh yeah, I think so too. This company definitely gets cleaned up. The founders, it is cleaned up. It's cleaned up. The founders cashed out. They get a 29% EBITDA. Insight is good. I mean, they're going to take this thing monster IPO. So they just took $2 billion off the table and now they're playing with house money. This is a beautiful move. I mean, remember, private equity used to be just suck all the cash flow out, just leave the carcass for Broadcom or something. There's a growth playbook with later stage right now and this is one great example of it.
Well, I mean, look at the market's changing. And again, if I look at AWS's reinvent this year, coming back here and set up to Andy interview, they really, and I wrote a post on this which I ended up nailing the keynote...
The community and the heroes specifically, which was that is a huge gesture to the community. Second one, I love developers. Laptop sticker. Developers are near and dear to my heart. The quote I had in my article. So again, he's reiterating and he's been consistent on that by the way since he's been CEO. I've been pressing him, don't lose the developers. He's like, we will not lose the developers. We will not lose the founders. He's hardcore on that. So those were two opening sequences. And then in the keynote, we didn't hear about agents until an hour and a half in 'cause he went right to security and then he went right to the infrastructure core jewels. And that's where their competitive advantage is. So you know, Andy Jassy and these guys don't like to play their cards if they don't have to, but they played a lot here, Dave. They showed at least me from what I...
Win the big artillery is infrastructure and then software innovation at the Bedrock model, SageMaker layer. And they're going to create an environment for the best place to do generative AI apps. And I think they're saying, hey, we're going to play the long game because we don't see those apps coming yet. I mean, yeah, there's some stuff out there and there's some use cases, but as we've been saying on the Cube, until the infrastructure is done, it's way too costly, not predictable. And they've done, they're doing the work under the covers where that needs to be done. And like I said, the old days was spin up some servers, some EC2, load my app and run on the web, my web app on Amazon. Doing that kind of ease of use with large scale supercomputer clusters is freaking hard. Dave Brown...
Chance that that entire on-prem market, good creator except for use cases that are connected to the cloud. So it's going to be a very interesting situation to look at. Certainly not the story you're hearing from Dell and HP and Lenovo. You know, I love the AI factory positioning. I think companies will have AI factories. But to me, I think what's going to end up happening is the reality will set in like we just can't do stuff at the scale of the A+ hyperscalers. And I don't put Microsoft in that category. Well, so I put them in the B category. Well, they are hyperscaler. I put them in the A category of M&A. Wait, wait, wait, okay, this is interesting. Okay, so you, but they are hyperscaler. I...
So here's the thing. The narrative has been Amazon's behind in AI. Why is that? Why? Because people are looking at their LLMs and going, you know, Titan and Truly and Olympus, no, and we'll see Nova, we'll see how that plays. But what everybody's missing is who's making all the money in AI right now? It's picks and shovels. It's infrastructure. What AI apps are out there? But outside like OpenAI, stick with me on this. So who's making money? Nvidia is making money. Who's number two in AI? I'd say Broadcom or Amazon's right there, right? The hyperscalers, Google, I mean, these guys are picking shovels and or you're making AI technology or you're selling ads, Google and Meta. Okay, so what did Amazon do that Microsoft couldn't do? So Amazon...
Have that these other guys don't have. They have Inferentia and Trainium which is low-cost training. So they could say, okay, let's train on Trainium. Yeah, it's not as powerful as the GPUs from Nvidia, but it's good enough to start learning and build a clustered scenario and building the tools around it. And what's really one of the most important things in AI is networking. You know something about that exactly. So they're now able to train models on their own infrastructure without having to rely on Nvidia's InfiniBand. So they're doing it at much lower cost and they're building. So the point is, they're on the S-curve whereas Microsoft couldn't do it because they couldn't get enough GPUs. They didn't have an Inferentia, Trainium equivalent. And so what'd they do? They...
For a while, Google and Amazon have really, really strong infrastructure, great tech, both players. And you know, Microsoft, so now they're just a B player. Now you got OpenAI and Sam Altman running around, who knows what shell game he's playing. And the Anthropic piece, we were talking about this recently. Anthropic looking pretty like a pretty good move. I think the Anthropic collaboration, if they can get access to what Amazon's doing at that cost and then can be trained closer to the hardware, that could be. And if they're collaborative, which we've been seeing here, and that's the feedback we've been seeing, that they're very collaborative. You had dinner with the execs there. So I mean, we're going to see at the end of the day, there's picks and shovels. Here's the AI landscape. People selling...
Judged as an AI player is having an LLM or a foundation model. And by the way, I don't think that's the definition. The dirty little secret is everyone has the same corpus of data on the internet, so it's just who's got it better. So that's going to get commoditized when people level up. So data feeding a data feeder and the picks and shovels will make money today. Okay, we're a consumer of AI, running a RAG, we're doing all converting all this. So we're a consumer of AI. So is JPMorgan Chase. Will they start making their own AI? So the Palantirs of the world have a nice strategy. They're clearly saying we will build and be an AI provider of technology. So they hired the PhDs, they got all the top guys. But most people are going to be consuming the AI and then buying picks and shovels.
Blatantly clear that Amazon's taken their Graviton playbook, like Graviton to x86, and they're bringing that to Inferentia and Trainium. And you know, we'll see about the models. So they're reducing their reliance on Nvidia for GenAI training and inference. So here's this. You talked about the Cube research. So George Gilbert, he goes deep. The in-house Inferentia team says that they're already running Claude 3.5 Haiku, which is the smallest Claude model, on Amazon infrastructure and they're coming soon to 3.5 Sonnet, okay, which is the leaderboard favorite right now. And they got a roadmap for running all models. And the Trainium team says they have the software...
Training and they, you know, we'll see about the Nova and house models. The point is that they were able to develop all this tooling, they put them on the S-curve, and that is what's going to help them win. How were they able to do this? Inferentia, Trainium, and Nitro. Yeah, right. That's the secret sauce that they have that allows Amazon to play in this game. And they I think asserted their leadership in AI finally. I mean, I think the playbook is they're going to have more compute than any of the cloud. Amazon's playbook is going back to the roots. They won by a very simple formula: be the best and the easiest place to stand up infrastructure to run apps. So just it's AI apps now. So they're targeting, they're not going to try to fall into the shiny new toys. Hey, one of the best models. So you know, Nova might not be a big hit, big announcement, but it might not be there long term. We'll see.
Playing their cards and their competitive strategy is clearly infrastructure. That's their bread and butter. That's where they have, they're on the curve, multiple curves now. And you look at all the people they're getting good at certain stuff. You heard the SageMaker guy saying, we're getting good at this. We're getting, Dave Brown, we're getting good at building silicon. You talked early on about clustered systems, right? Back years ago. And yeah, and so but AWS has the infrastructure chops to follow in Nvidia's footsteps relatively quickly. And you know, Microsoft doesn't. They have GPU constraints. Model distillation in Bedrock, that's going to be a big one. S3 table buckets is huge. But before we go there, the key for that I see for Google and Amazon is they're able to...
In the hole. But to your point about S3 managed Iceberg tables, we'll see. I reached out to Snowflake for a comment on this and I haven't seen anything yet. I might see it. I didn't see it by this morning because this, you know, we've been talking all last summer about Snowflake and Databricks and what this all means. So imagine, imagine here's a scenario. We heard today, so okay, infrastructure they playable. Let's get in some of the weeds and how, why they're going to make it easy to build apps. Today, if I'm a user and people who are listening can relate to ChatGPT, you take a Word document, okay, the context window, how many tokens can I in my context window? You go to ChatGPT, everyone's kind of done this. You cut and paste a big long document. You take actually a document and you put it into ChatGPT. It could be a 20-page research paper, whatever, something big, okay, and it digests it and you like...
Primitive at an S3 bucket that's got 20 petabytes, okay, or you know, running Bedrock guardrails on with reasoning at S3 buckets that has a billion objects, okay. So just the concept of saying, oh, just point to it, yeah, that is the same as that ChatGPT. That is revolutionary. So I think when you start to see that kind of under the covers activities now, the application developers can just take advantage of that by basically stitching data sets together. That's why the data conversation, that's why the S3 buckets things are huge. The table formats, cleaning up a lot of that infrastructure gets them building that tooling to your point about SageMaker. I think that's where they win. So again, I...
Love with, I'm a model, I'm gonna build a model. I think that's just going to happen. Models will be there. And I think that's not going to be the benchmark. The benchmark is what apps are hitting. And if you ask yourself what AI apps are actually out there right now, OpenAI, I mean ChatGPT, like I mean, it's the LLM, right, that you see and rudiment. Salesforce, actually Salesforce, you know, claims that Agentforce was a big driver in the quarter. Okay, but are they apps that where they're consuming AI to do something or are they actually doing AI and building an app? They're apps that are applying AI to their AI or someone else's AI. Both. Yeah, this is again, this is where you start to squint through. So AI gets embedded in everything. I asked the question to some folks offline, how do you tell a player...
Know where you are. You can't say you're real AI when you're not. So you're using AI to do something of benefit, that's cool. It's like a website. If you're in the web, you know, got a search, I got a look how cool it is. No, but that's the way it is. So let's get into the Andy Jassy preview. Before we do that, Bitcoin hit $100,000. We have to at least give some props to that. But couldn't hold it. Couldn't hold $100,000. Yeah, we were right on this. Remember, remember? See, you know, right, I do remember. Well, there was a couple things that was interesting on this one. Was his election year, Trump. Two is MicroStrategy, Saylor has been talking about this and I kind of like what I like his narrative. I'm not going to lie. I think Saylor's got it. It's like Bitcoin's a small percentage of store of...
Like gold. I'm like, wait a minute, isn't gold a store of value? But anyway, go Bitcoin. But you know, Bitcoin, we'll see what Ethereum and Solana come. That's a whole another conversation. We'll say that for another part. But let's get into the Andy Jassy exclusive. Again, we had a chance to get with Andy. Shout out to him. Let's set this up. He, we met with him down on the show floor. We set up some production there and we got this exclusive. We're the only outlet, news or any media, that has the exclusive one-on-one with Andy Jassy. We only had 15 minutes. We did our best to pepper him with some questions, but we wanted to get some thoughtful answers out. And we didn't want to ask the classic, hey, I saw Apple on stage. We just want to try to get at the heart of how he's thinking, why he came back, what's his mood, is he leaning in, and what he thinks about the...
Interview. Hello, welcome to the Cube special presentation. I'm John for Dave for exclusive Cube coverage with Andy Jassy, the CEO of Amazon, former CEO of AWS. Andy, great to see you. Thanks for having us in your little den here. Used to be doing media hits all week. It's great. It's great to be with you guys. This is a tradition. We've done this a lot of years together. I love the fact that you're doing videos and you came on stage. The energy in the room completely changed when Matt introduced the OG, the Godfather of cloud. And then the crowd reaction was pretty strong. It was really, you can really feel the energy. And then this just continued. You dropped the mega announcement of having your own model. But just in general since then, it's just been a great event. And my first question is, do you miss it?
