Cloud AWS re:Invent Coverage | Andy Jassy, AWS

Clips
News
More from Cloud AWS re:Invent Coverage

Andy Jassy

President & CEO

Amazon

Three insights you might have missed from AWS re:Invent

The AWS re:Invent gathering in Las Vegas last week provided several newsworthy announcements, including the cloud giant’s entry into the world of AI f...

Andy Jassy returns to AWS re:Invent as Amazon beefs up its AI chops

Amazon Web Services’ re:Invent conference in Las Vegas dominated the news this week, as the cloud giant debuted credible new artificial intelligence m...

Exclusive: Amazon CEO Andy Jassy reveals AWS’ strategy for building the enterprise AI platform

Amazon.com Inc. Chief Executive Andy Jassy made an unscheduled appearance at the AWS re:Invent cloud conference Tuesday to announce the company’s bid to reshape the world of artificial intelligence models. Today, in an exclusive interview with SiliconANGLE’s theCUBE, Jassy doubled down on how important he views his company’s AI initiatives.“This is the biggest change since the cloud and possibly the internet,” said Jassy (pictured, far right, with theCUBE Research’s John Furrier, left, and Dave Vellante). “I think every single customer experience we know of is going to be reinvented with AI

play_circle_outline The AI Revolution in Cloud Computing: Data Organization, Infrastructure Modernization, and Building Blocks

play_circle_outline Revolutionizing Infrastructure with AI: AWS' Vision for Generative Applications in Customer Experiences

play_circle_outline Impact of large-scale AI clusters and infrastructure

play_circle_outline AI's role in transforming SaaS applications and customer experiences

play_circle_outline Amazon's role in the cloud market and SAS opportunities

play_circle_outline Unveiling Amazon's Cloud Dominance: Opportunities in SaaS, Challenges of AI Clusters, and Serverless Shift

play_circle_outline Serendipity in innovation experimentation

play_circle_outline The Rise of Cloud Technology: Cost-Effective AI Model Training in a Shifting IT Landscape

play_circle_outline Impact of AI on cost-effectiveness and productivity

play_circle_outline Rise in Corporate Model Building: Enhancing Customer Experiences and Societal Benefits

Info
Transcript

Andy Jassy, AWS

Andy Jassy

President & CEO Amazon

Andy Jassy, CEO of Amazon, discusses the connection between AI and cloud growth, emphasizing the importance of organizing data for AI usage. He mentions the challenges of building generative AI applications and the necessity of modernizing infrastructure for AI integration. Jassy highlights the difficulty of building AI clusters on-premises and the need for cost-effective compute solutions. He discusses Amazon's innovative services like Connect and the potential for AI to transform industries. Jassy envisions a future where customer experiences are driven by ... Read more

explore Keep Exploring

What is the connection between the growth of AWS and the rapid development of artificial intelligence (AI)? add

What are some examples of software services that have been built from the ground up with AI and are highly scalable and cost-effective? add

What are some key market segments that have proven to be large and successful despite competition, and how has Amazon's experience with sellers and scale influenced their approach to developing new services like serverless and Aurora DSQL? add

What are you hoping for as we document the next wave of cloud growth changing the world? add

What are some potential benefits and opportunities for the future of technology, specifically in terms of productivity, cost-effectiveness, and advancements in AI? add

bolt Powered by CUBE AI

Andy Jassy, AWS

search

>> Hello. Welcome to theCUBE special presentation. I'm John Furrier, Dave Vellante 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. You're just going to be doing media hits all week. Great to see you.

Andy Jassy

>> It's great to be with you guys. This is a tradition. We've done this a lot of years together.

>> Great.

Dave Vellante

>> Right.

>> I love the fact that you're doing videos and you came on stage. The energy in the room completely changed when Matt introduces the OG, the godfather of cloud. Then just the crowd reaction was pretty strong. It was really ... You can really feel the energy. Then that 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. My first question is, do you miss it? You're back.

Andy Jassy

>> Well, first of all, I thought Matt did an awesome job with the keynote. It's a thrill to be back. I love re:Invent. There's not a week I like better during the year. I'd miss not being here for a few years. I was honored to be back. One of the things I love about re:Invent is that ... Actually the thing I love most about it is the people, just being around our customers. We work so hard all the time, and we see all the warts and all the things that we think we can be doing better for customers that we want to deliver for them. You're heads down. Then to be around all your community partners and your customers and your partners and have them excited. We hear stories. One of my favorite parts about re:Invent every year, and I've heard this at least three times in the last 24 hours, is that people come here to learn from each other and to learn about what they can be doing differently, and to be inspired to go back to their own businesses and change their customer experiences. You get teams that come together and they see a bunch of services released or they hear what others are doing, and they go to the bar and they say, "We're doing this. Let's go." They go back and they make a change. That's what this is about. It's trying to help people change their business.

>> It feels like the levels of the original re:Invent we went to in 2013, our first. It's now been our 12 years. But the vibe feels like the old school re:Invent when it was a smaller group, but the energy was high. We're on this AI generation. We can see some of the cards you guys are playing, the infrastructure advancements. But now with the AI, I have to ask you, you've operationalized the cloud and did all the greatness, turned the world on fire, created disruption, okay, taught people who are working backwards, documents, they operationalized everything. Now with AI, is that the same plan that you see? Because these builders are going to go faster, they're going to have the AI at their back. How do you look at this next wave of cloud?

Andy Jassy

>> Well, I think they're very much connected. I mean just perspective-wise, I mean we felt like we grew the AWS business really quickly. It took us about eight or nine years to build an AWS annual run rate of about $4 billion or so. If you look at AI, that'll happen in just a couple of years. So it's growing incredibly quickly, but the reality is they're very much connected, because, first of all, if you want to use AI, you have to have your data organized and architected in such a way that you can access it. Try doing AI from a mainframe. It's nearly impossible. So you really need to have your infrastructure modernizing the cloud and your data accessible to run AI. Then the reality is that I think ... You've heard us talk for years about every application is compute. Virtually every application is storage. Almost every application is database and analytics and content. Another one of those core building blocks is going to be inference. Every application is going have some generative AI and inference infused in it. And so, it's very much a building block. I also think that you don't show up and just have one big service that everybody uses that has all the features and, presto, people can do it. You have to build the right set of primitives and building blocks that people can stitch together. That's what we've been doing over the last number of years with SageMaker and with Bedrock and with our own chips that we think are going to help people be much more price-performant in their training and their inference. You'll continue to see that from us.

Dave Vellante

>> When you think about the impact that Amazon has had on the industry broadly, it's pretty remarkable. John mentioned working backwards, customer obsession, two piece of teams, the flywheel concept, what I call Jassy's law, there's no compression algorithm for experience. Two questions. Has that changed? Is there a compression algorithm for experience in this AI era? Are there new operational frameworks that are emerging as a result of this new technology?

Andy Jassy

>> Well, I think that the models ... it's not like people started working on these big transformer models two years ago. People were working on these ... A lot of us were working on them for a long time, and the first several versions of the models just weren't that interesting. Then all of a sudden really OpenAI's GPT-3, the intelligence level, just kind of popped off the chart, and that opened up all sorts of possibilities. I feel like the first real application that just exploded was ChatGPT. But if you think about ChatGPT, it's kind of a pretty thin user interface on top of a model. It's remarkable. I think one of the things that people don't realize, I was talking about this in the keynote, is to build great generative AI applications, it's not fast actually. If you're doing a software development project, we can get on a whiteboard and map it out. Of course, there are differences, but it largely functions as you design it. Whereas with generative AI apps, it's very iterative. You think you're going to make a big advancement in what you've done, and then it turns out you don't. You think you're making a tiny change, and it gets much better. And so, it's very iterative. Then it's not just the model. I think people very often trick themselves with a good model that they think they're there, but they're really only about 70% of the way there. Applications, it don't really work well for users if there's 30% error rates or wonkiness. And so, the UI really matters. The fluency of the messaging really matters. The latency really matters. The cost efficiency really matters. So all of these things, it very much applies that there's no compression algorithm for experience. What we tell ... A lot of companies, before the pandemic, companies were on this march to modernize their structure, move from on-premise to the cloud. Then the pandemic hit, we got into survival mode. Then there was an uncertain economy and everyone was cost-optimizing. Now as people start to spend again, people are asking us, should we modernize our infrastructure or should we do generative AI? Of course, the answer is yes, but if you don't take the low-hanging fruit of modernizing your infrastructure, you're actually not in a position to take advantage of AI. But you have to pick a few initiatives that really will change your business and get that muscle to build generative AI apps, because it's not fast to be great at it.

>> Andy, I want to get your thoughts on the infrastructure side, because we've had this question on theCUBE, about building AI clusters. People try to do it on-prem. Cloud has advantages. Also on theCUBE, the word blast radius has come up a lot, which I know is an Amazon term that James Hamilton and the team uses. It's hard to build large-scale infrastructure. Dave Brown and I talked about that. Talk about the dynamics of the infrastructure at scale, because you guys see stuff at scale that others don't. Things do break. Building a system power the kind of apps that are coming. Certainly the developers are going go crazy with GenAI, certainly with Bedrock and some of the models, but to run it all, you need the horsepower.

Andy Jassy

>> Yeah. Well, as you know, we've been building very large-scale infrastructure and running really the largest workloads in the world for a really long time. I think that when you look at the ultra-large training clusters that are being run, they're not simple. When you're running a cluster of a thousand chips, it's very different than when you're running a cluster of a few hundred thousand chips, like we're going to run for Anthropic and their future models training on us.

>> Yeah.

Andy Jassy

>> I also think that one of the biggest inhibitors, in my opinion, for generative AI applications to be as broad as they're going to be, some of it is skillset and experience. But a good piece of it is the costs need to keep getting lower, and the cost of the compute, and chips and the compute so you can do training much more cost-effectively. Optimizing inference, which again is the cost of the compute, but also some efficiencies. We've done a lot of inventing there. And so, as we are building very large-scale generative AI applications inside Amazon, because we're building about a thousand of those apps already, as well as running them for large customers, we're getting really good experience, even ... If you think about training, if you guys have looked at HyperPod in SageMaker, it's a radically different experience in what you can do on the networking side. Then if you take HyperPod and you combine that with what we're going to be able to do with training them with UltraServers and UltraClusters. It's just a different level of skill to be able to train your workers .

>> I mean from a replication standpoint, I want to try to do that on-prem. I mean-

Andy Jassy

>> Really hard. Really hard.

Dave Vellante

>> I want to ask you about some of the things you're doing in amazon.com, and I want to talk about the market. We love to talk about the market. So if you look at the market and you just focus on the big three hyperscalers, maybe throw in Alibaba, IaaS and PaaS, Amazon's got over 50% of the market, and it's maybe a couple hundred billion. John, years ago, when you and I first met in New York, wrote a piece, A Trillion Dollar Baby.

>> Yeah.

Dave Vellante

>> So if you add in SaaS-

>> It's actually a trillion now....

Dave Vellante

>> and all the ... What John calls fake cloud and the professional services, it's-

>> Well, on the numbers standpoint, just to put it down.

Dave Vellante

>> But you add all that in and it's about almost 900 billion now. So then it's growing at, whatever, 20%. So you undercounted, a trillion-dollar baby.

>> Well, bundle in their SaaS .

Dave Vellante

>> So here's my question. So you're doing some really interesting things, and you don't participate really in the SaaS. I think our SaaS business is probably as big as yours. But you actually have things like Connect, and you're doing-

Andy Jassy

>> I was going say Connect is in that category.

Dave Vellante

>> Yes, yes. It's bigger than ours. But you're doing some other really interesting things, applying AI internally that you could potentially and will, I'm sure, point to the external world. Is that how we should think about your up-the-stack opportunity?

Andy Jassy

>> Well, I mean what's interesting, I mean something like Connect as an example, it was built from the ground up to be a call center, a software service that was built on the cloud, that was highly scalable, was really cost-effective, but also was built from ground up with AI. If you look even in the last week at the features that Connect just launched, they continue to iterate in a very fast clip. And so, we have a bunch ... I'll say supply chain is another area that we think we can be very effective and we have a lot of experience just like customer service there. But I also believe that AI's going to open up all sorts of new SaaS opportunities and software-as-a-service opportunities. I've been saying this for a long time, I've told you guys this too, which is that I think every single SaaS company and application that we know of will be reinvented with what's available in the cloud. I think that's doubly true when you think about what AI allows.

Dave Vellante

>> Yeah. And that's a partner play. You just mentioned it.

Andy Jassy

>> Yeah, that's a partner play. Well, I mean these market segments are so large. I always got a chuckle over the years that we would launch something in some area and people would theorize it was the end for these companies. It just never has happened. They're really large market segments with lots of winners. Sometimes our customers will insist that we have an offering in a certain area, they really want it it, but we also have lots of experience that even when we offer something, a lot of our partners in that same area are very successful on top of it.

>> Yeah. I mean you guys at Amazon have a great experience with sellers, right?

Andy Jassy

>> Right.

>> I mean you guys have scale. So to me, as you guys grow, you are in rarefied air. I think one of the things that's coming out this year is the maturity of services at Amazon is an economic force. So you're seeing things at scale. This is scalable apps, and now with all the horsepower, basically high-performance computing. You're seeing new applications emerging. You once said in theCUBE, when I asked you, if you had to build Amazon again, what would you do? You said serverless. I think what year that was. But serverless has been popular. You mentioned inference. So will there be these next-level apps that are going to take advantage of the high-performance computing that are going to be inference lists or database lists? What do you call that? Serverless is serverless. Inference would be database lists or ... Because-

Andy Jassy

>> No, I think you'll still have databases. Databases serve a really important role even with inference. But I mean I think ... I remember that conversation we had where you asked me if we were going to build it over again. I said I would do a serverless, and you were incredulous about that.

>> Yeah.

Andy Jassy

>> But I think one of the things that's been really interesting, if you've looked, we have metronomically, over the last five years, taken all of our analytics services and given people serverless opportunities. Then same thing on the database side. We were blown away at how much traction Aurora Serverless has gotten the last few years. It was a really important part of how we thought about building Aurora DSQL. Aurora DSQL, in many ways, I think it's a great AWS and Amazon example of how we think about building, which was we were working on this concept when I was still in AWS, which was we knew that customers wanted a multi-region database, that relational database. They wanted strong consistency. They wanted low latency. They want high availability. They wanted SQL compatibility. And they wanted it to be easy to run. There were a couple options out there, but they were really only good at a couple of those things, and all the ideas we had were also only good at two or three of those things. We kept getting down the road in the working backwards process and ripping it up because we just felt like we weren't really solving the problem. And so, it forced us to radically rethink how we did ... Verne went through in his keynote a bunch of the components that are in DSQL.

>> There were a lot.

Andy Jassy

>> There's a lot ... Right. I mean it's really ... It's very inventive. Then we solved for the end. We solved for all those components, and it's totally serverless. And so, I think that the large-scale databases in the future will also be serverless.

>> Well, serverless actually set the table in those 10-year Lambda anniversary and reverting on that. But, as you guys say, that inference is the next building block. I mean that means something in your world, right?

Andy Jassy

>> Yeah.

>> I mean a building block is significant. It's like adding another major component. What are the implications? Because serverless still ran on servers. Databases are everywhere. So I see this as a developer touch point, an application touch point. What are the implications of inference as a building block? It's not just a service, it's a building block. Can you just share your vision of why as a building block it's going to be impactful?

Andy Jassy

>> Well, I mean inference really is the implementation of models of generative AI, if you think about it. I mean you work on these models, you train these models, you try to be able to make predictions, and then the actual predictions are the inferences. And so, what it really is another way of saying is that generative AI is going to be a very significant part of every application. We talk about generative AI. In the first year, year and a half of what was happening with generative AI, it was so breathless and there was so much hand-waving. We have tried to be as clinical as we can be. We knew what we were building. We were building, like we always do, a bunch of primitive building blocks and components that customers can stitch together to build great generative AI and inference. But we have always felt like we were on this path in AWS to be, I don't know, a couple hundred billion dollar-plus annual run rate type of business, and that was before AI. That was before generative AI. I think that every single customer experience we know of is going to be reinvented with generative AI. Then I also think it's going to open up all sorts of opportunities and applications that we just didn't really dream were possible before. That's one of the interesting things inside the company is when you start to see how the models work and you start to build great experiences, it opens up this explosion of ideas that people just didn't think was possible before. And so, to me, the fact that everything can be reinvented and a lot of things invented for the first time with inference and generative AI is a huge opportunity for customers to build great customer experiences.

>> The innovation experimentation creates serendipity. You mentioned being on whiteboards. I mean we're in this really experimental breakout moment. Final question, as you come back to AWS, the home for you that you built, the house that you built, the OG, the godfather of cloud, what's it feel like? Then what do you hope happens as we document the next wave of cloud growth changing the world? What's your-

Andy Jassy

>> Well, first of all, it's great to be back with you guys. I do remember the first time we met in that dank conference room in New York City.

>> Yes.

Andy Jassy

>> I felt like you guys got the cloud in a way that nobody else had at that point. I continue to watch what you guys do and say and think and theorize. You guys are very on it. So I always enjoy spending the time. I would say also, I mean I appreciate nice words, but, as you know, anything that you build that has success, it's a group of people from the start. So I've been lucky to be part of this team. What I hope for in the next several years is that ... I mean, remember, as fast as the cloud has grown, it's still about 85 to 90% of the worldwide global IT spend is still on-premises, which I think is insane. I am very confident that you fast forward 10 to 20 years from now, that equation is going to flip, and it will allow people to get more done for less money, to invent a faster clip, and to get better productivity from their engineers, which is their scarce resource. So one is we hope that we help people make that transformation. Then when I think about the opportunities, this is a golden age in technology. I mean we don't have ... This is maybe the biggest change, for sure, since the cloud and probably since the internet with what's available and what the opportunities are in AI. And so, I'm hopeful that we make it much more cost-effective for people to be able to train models, to be able to do inference at scale. It's part of why people are so excited about training. This is just going to make it much more cost-effective to do more with less.

>> .

Andy Jassy

>> I would've told you, if you'd asked me 12, 15 months ago where most companies would operate in those three layers of the stack we talked about, I would've assumed almost everybody was going to only operate in the middle and leverage existing frontier models. But I really now strongly believe from our own experience that those with technical competence are going to do a lot of their own model building as well. So we're going to try and make this much easier for people to do. Then we're still early with respect to what the customer experiences are. I mean how long have we been talking about autonomous driving? I mean I remember talking about this with you 10 years ago and it's still not here in any broad way, although we're getting close. But there are going to be so many experiences that are going to be completely different for us, for our kids, for our kids' kids that are going to be great for society. I'm optimistic that we'll be underneath a lot of them.

>> Yeah. I appreciate the compute and all the advancements. It's the renaissance for software. Again, congratulations. Thanks for having us and great to see the throwback Cube reunion.

Andy Jassy

>> It's great to be with you guys. Thank you.

>> Thanks for coming back.

Andy Jassy

>> Yeah, thanks.

>> All right. Andy Jassy here at theCUBE special broadcast. I'm John Furrier, Dave Vellante. Thanks for watching.