 Welcome back everyone to theCUBE's coverage here in Las Vegas on location for AWS re-invent. Amazon Web Services annual user conference. This is theCUBE's 13th year covering re-invent. We've seen the trials, the tribulations, the growth step by step and now this year more than ever a major step function of a change is happening with generative AI. It's up and down the stack. You're seeing applications, user experience, all changing, all driven by data and scale and price performance of infrastructure, new middleware layers, new content concepts. We're here with Thomas Sautterstrom who's the Enterprise Strategy at AWS. He's more of a legend. He worked at JTBL, a lot of great stuff there. Built a chief technologist organization at AWS now on to helping customers re-architect their future. Welcome to theCUBE, great to have you. Thank you, John. It's a pleasure. I mean, we talked about this, we met on the bus a few years ago and finally it's happened. You mentioned you've been at all the re-invents. I was actually part of creating re-invent as a customer. So I've been at all of them too. And this one is 10 times bigger in every aspect than the first one in 2012. You've had a front-row seat on both sides with now Amazon. A lot of cultural change even within re-invent, the big speculation this year for Adam Sileski was his keynote, what is he going to do? Given all the hype on AI and generative AI in particular, and he delivered in classic AWS fashion re-invent, they save all their goods for re-invent, they angle conference, they lay it out there and they go, okay, game on, you're move, other people. But also with customers. I mean, there's opportunities to rethink and what the big takeaway for me and what we've been reporting is, okay, you got three layers of this new generative stack, there's going to be a feeding frenzy for developers and new ways to write applications. But yet under the covers is going to be some new dynamics happening with the technology. What is, you're in the middle of this with AWS, you're advising customers, the biggest brands and companies on changing, what's your take on this? What are you talking about? So if you look at the high level, if you look at the trends, as you know, that's what I did at NASA, Jet Propulsion Laboratory, look at what are the future trends. And now, look at re-invent last year, what were the trends? It was much bigger data, make it easier to access, but not yet easy to understand. This year continues that, much bigger data, and you saw a lot of announcements of the big data warehouse redshift and how it makes it faster. I love the idea of an unlimited Aurora, unlimited, that's a big word, but yeah, it is. And then, so you have a lot of data. Now, it makes it easier to access. You saw a lot of serverless, which of course means you don't have to worry about the details of the cloud. And you also, it's faster and easier to access. And then, so now you have a lot of data. Now what do you make, how do you make it easier to understand? And that's where Q comes in. So in 2017, we built an intelligent digital assistant at Jet Propulsion Laboratory. We used Alexa to tie into thousands of contracts. It was stored in Oracle databases, pulled it out, and it was 10 times faster than it was to actually type it, just to speak it. And, but we had to build all that heavy lifting inside ourselves. Now, with Bedrock, you don't have to. So I think the big trend is that you are getting Q to be this intelligent virtual assistant for everything. It's at the code, how to design it, how to code it, build test cases, deploy it. But Werner Vogels this morning said something really key. He said, if I could quote it correctly, AI recommends professionals decide. So it still is where humans have to be in the loop. We can't let it run amok. And Swami also talked about human intellect scaling as well. We have data in our heads as well, so we can think too. Yeah, and the other thing that I really like about this big picture is when you look at internet of things, all of our sensors, sensors on ships, in mines, in space, and you tie and you push the AI smaller and push it out to the edge. And now you have generated, you can just ask it questions. Can you imagine what the future looks like? You can ask anything about your enterprise and have a conversation with the data. So that's, I'm not sure if people missed the exciting things about the data. Yeah, the data thing is huge. Well, two things that jumped out of me this year I want to get your reaction to. One is the sheer performance of compute to generate images like just six months ago was incredible, night and day difference, one. Number two, moving from a component-based system of stuff to more productivity-driven, simple, composable elements for data. Exactly, I love that. And so somebody said like with the S3 infrequently accessed data zone, 10 times faster, 50% cheaper. So when you say, well millisecond access, that's darn fast. So when you need a generative AI, you need the data. That's the S3 express zone that was announced. By the way, he started that out in the keynote. So if you're going to reinvent something, start with storage, because how do you reinvent storage? And they did it. And I asked people about feedback and stuff. Why would you talk about storage? Why is that so exciting? And then came Q and all this like, okay, now it makes sense. If you can reinvent storage, you can reinvent everything. But that's an example of the reinvention going on. I want to get your thoughts on this, because you've been involved in a lot of these big projects with NASA, Jet Propulsion Lab, and Amazon. And they've been, cases where you can get your arms around where like the supercomputing capabilities or space, great stuff, you need compute. Now it's just, I won't say general purpose, but democratizing the access, what Q does, is allows the low code, no code environment, and then this new developer layer with foundation models, where inference becomes the app. Where inference is the velocity and the latency of answers, or things getting done on behalf. So questions are answered in low latency. Not packet latency, but like. Exactly. And what I like about what AWS is doing, you can think of it as a steady hand of innovation for enterprises and startups. It's not going to go wild predicting things that's not going to happen. It's not going to go crazy. But it's thinking about the underpinnings, the machine. And the machine here is innovation. So what I think this reinvent, more than anything, it's democratizing innovation. Because you can get all these developers to reuse what other people are doing, but also insert their own. And what's the next programming language? Voice. Yep, it's natural language. Yeah, yeah. So I'll get your thoughts on the data pieces. I think, I mean, I look at this, I go, this is going to be a completely script flip moment where data flips the script. Data management, data intelligence. So how do you unify data? And there's a couple of different schools of thought here. You unify it with a central lake and you create intelligent mechanisms around it or you distribute it and make it kind of on its own. What is the common theme here? Because like security, you got to get your arms around the data. You do. And the developers have to code in line now with data, almost as a data developer now that's native into their workflow. So whether it's voice or low code, no code. The programmer, the developer, needs to have access to all that stuff built in from day one, compliance, security, governance, almost embedded from day one. So three points on that. One is for a large enterprise. If you are a chief data officer out there, don't give up. You're going to get a lot of love. We hired them in the enterprises. We talked to 80 executives one-on-one in the last six months. And the chief data officers weren't able to produce results because 70% of their time is spent on people issues. It's my data, you can't have it. So we recommend an organizational data mesh. Look backwards, think work backwards. What's the outputs? What's the input? And then how do you make this work? And so train everybody on data. So when you look at generative AI, all of a sudden the CEO see there's something tangible here, something I can understand and need. How do I make it happen? So they give it to whoever is this AI person and they need that data. So all of a sudden the investment will go back into creating an actionable data strategy. And AWS will help to make it easier with the data lakes like Redshift, but also all this data from all the sensors. So I think what we're going to see is we're going to see a data integration expert, new title, just like prompt engineering is a new title. But I think that AWS, there was a lot of- How about a data engineer? The data engineer is going to be crucial. But it's going to be data engineer and data integration engineer. Because you get so much different types of data. And it's got to be fast. And their technical roles, this is like an architect. They're engineering, it's an engineering position. Not a data science or database administrator kind of thing. So the people who get a lot of love, the chief data officer finally will deal with data. The data engineer is going to be super important. The network engineer is going to be incredibly important. Especially when you tie an internet of things with all the various networks. 5G, soon 6G, it's a wonderful world to be in. Yeah, and you brought up a good point I want to double click on. The time to value the CEO sees data. So data now can reveal some instant benefits versus some pie in the sky, proof of concept, purgatory, some sort of test. Where are we? They forgot about it, person left. Instant value that you can say we can go further. So important. How can you get to action? You'll think big, start small, act now. And one of the things I've been talking to these CEOs about is can you have a conversation with your data? No, how do I do that? I go to my CIO and he tells me. So generative AI light is something called QuickSight Q. And you can now have a conversation with your data. AWS partner with NFL, National Football League, with sensors so they know everything about completion rates and touchdowns. And you can just have a conversation with it. It's a fabulous thing. So now when you have QuickSight BI, this new tool, and Q, you can really have a conversation with the whole thing. So the new user interface is going to be natural language. Which is wild. Discovery, invention, recreation, reinvention, all happening at once. Now, next quick hit here for you to think about and I'd love to get your opinion on this. The role of synthetic data. There's a huge opportunity. I know a little bit about it. Bill Vassily had a quick conversation in the hallway. I'll see you talk to Bill. Well briefly, but he then enlightened me on where the value is, whether it's you're trying to simulate packages and we just had another guest talk about how you can look at scratches, like defects, you don't want to know how to be a scale defects. These are things that you really can't scale but synthetic data can fill the role. What is synthetic data and why is it good and why is it going to be important? So if you think about, let's say you're building a spacecraft or an airplane. So you want to use machine learning to figure out where the problems are and perhaps use something called reinforcement learning to figure out how to fix it. So now you take that airplane and you bang it with a hammer, you take a picture, bang another with a hammer, take a picture a billion times. Eh, it doesn't work. Instead, what you can do is do synthetic data which creates those faults. Now you use machine learning or reinforcement learning to figure out what happens when those faults happen. So as far as creating a resilient anything, it's fantastic. Now, really big. If you do a digital twin of this physical environment and you trust that they're the same, you can now start experimenting with a digital twin and injecting the synthetic data to the faults even in the metaverse, insert people and only deploy it once it's perfect. So this is great for simulation and also predictive scenario analysis on anything. Exactly, so you build in resilience from the get-go. See impact on the edge devices. What does synthetic data mean for the edge? So I think, this is my opinion now, I think what is happening is the edge is becoming, calling it cloud continuum. Today we think of the cloud as data centers and we talk about regions availability zones. And the devices out there are something you feed. The devices will be part of this cloud and the machine learning will squeeze down into something that can fit and work on those end devices, whether it's a camera or a sensor and just sit there and do its job and connect back to the cloud when it needs to. And the AWS Greengrass is a fantastic mechanism for that. Great, I mean, I really appreciate your time. I know you're super busy and I really appreciate you coming on theCUBE. I have to ask you about the super edge or the far edge, space is the ultimate edge. Okay, we're just having three hours here. So, quick hit on the space. Because again, you can't do break fixed in space. You know, and it's congested and contested right now with a lot of people and there's the security channels up there, you got space junk. You got a lot of ability to launch satellites. So, I talk a lot about the new space economy. Of course, I came from jet propulsion NASA. And the new space economy is really about, if you look at from 15 years ago to now, each component is about 5% of what it costs then. So, the barrier to entry of cost is greatly reduced. So, now you have a business of creating a launch capability and just take hitch a ride. And so, everything has gotten smaller. So, you're seeing 20,000 satellites come up here in near term and most of them are going to provide internet access like Hyperwell and awesome. But also, they will help us look at Earth from space to help protect the planet, understand climate change. And of course, there's business to be had. We're already seeing satellites identify illegal fisheries, identify wildfires and before they start, as they start to smoke. And now, all of a sudden, you can put out a drone or something automatically to put it out. What if you could move that up into space? Now, it's from three minutes, it's three seconds. And AWS has already extended the cloud into space. We've already run with partners. Is it a region? It's availability zone. One day there will be a region on the moon and on Mars. Today, we have machine learning on the satellites. We even put Alexa into space. We put a snowball into the International Space Station. So it's, and the idea here, the big, big, big news is if you could treat these satellites like you treat the computers on Earth, you just update them on demand. What changes the network protocol? That's why those networking engineers give them love. I got to get a lot of love. So, in summary, a lot of love going to the network engineers, especially space, data CDOs, chief data officers, and humans. And for space, this is wild. Government has funded space. We're seeing commercial funding more and government funding less. And that's a big change. Well, I know you got to go. We'd like to do another hour on space. We should do. We should do. Let's schedule that up. I appreciate your time, Tom. Thank you for coming on the queue. Always a master class with you and great insights. And I think everything you just put it together perfectly and good luck on continuing to educate the customer base on how to transform with GNI. Yeah, as an enterprise architect, we have a lot of fun to do that. A lot of love here in theCUBE. We're bringing more love and great content to you. Of course, this is theCUBE's coverage. Back to you in the studio in Palo Alto for the live stream. We'll be right back here on location in Vegas for re-invent after this short break.