(upbeat music) >> Hello and welcome back to theCUBE. We're here in Palo Alto, California. I'm your host, John Furrier with a special guest here in the studio. As part of our Cloud Native SecurityCon Coverage we had an opportunity to bring in Jon Turow who is the partner at Madrona Venture Partners formerly with AWS and to talk about machine learning, foundational models, and how the future of AI is going to be impacted by some of the innovation around what's going on in the industry. ChatGPT has taken the world by storm. A million downloads, fastest to the million downloads there. Before some were saying it's just a gimmick. Others saying it's a game changer. Jon's here to break it down, and great to have you on. Thanks for coming in. >> Thanks John. Glad to be here. >> Thanks for coming on. So first of all, I'm glad you're here. First of all, because two things. One, you were formerly with AWS, got a lot of experience running projects at AWS. Now a partner at Madrona, a great firm doing great deals, and they had this future at modern application kind of thesis. Now you are putting out some content recently around foundational models. You're deep into computer vision. You were the IoT general manager at AWS among other things, Greengrass. So you know a lot about data. You know a lot about some of this automation, some of the edge stuff. You've been in the middle of all these kind of areas that now seem to be the next wave coming. So I wanted to ask you what your thoughts are of how the machine learning and this new automation wave is coming in, this AI tools are coming out. Is it a platform? Is it going to be smarter? What feeds AI? What's your take on this whole foundational big movement into AI? What's your general reaction to all this? >> So, thanks, Jon, again for having me here. Really excited to talk about these things. AI has been coming for a long time. It's been kind of the next big thing. Always just over the horizon for quite some time. And we've seen really compelling applications in generations before and until now. Amazon and AWS have introduced a lot of them. My firm, Madrona Venture Group has invested in some of those early players as well. But what we're seeing now is something categorically different. That's really exciting and feels like a durable change. And I can try and explain what that is. We have these really large models that are useful in a general way. They can be applied to a lot of different tasks beyond the specific task that the designers envisioned. That makes them more flexible, that makes them more useful for building applications than what we've seen before. And so that, we can talk about the depths of it, but in a nutshell, that's why I think people are really excited. >> And I think one of the things that you wrote about that jumped out at me is that this seems to be this moment where there's been a multiple decades of nerds and computer scientists and programmers and data thinkers around waiting for AI to blossom. And it's like they're scratching that itch. Every year is going to be, and it's like the bottleneck's always been compute power. And we've seen other areas, genome sequencing, all kinds of high computation things where required high forms computing. But now there's no real bottleneck to compute. You got cloud. And so you're starting to see the emergence of a massive acceleration of where AI's been and where it needs to be going. Now, it's almost like it's got a reboot. It's almost a renaissance in the AI community with a whole nother macro environmental things happening. Cloud, younger generation, applications proliferate from mobile to cloud native. It's the perfect storm for this kind of moment to switch over. Am I overreading that? Is that right? >> You're right. And it's been cooking for a cycle or two. And let me try and explain why that is. We have cloud and AWS launch in whatever it was, 2006, and offered more compute to more people than really was possible before. Initially that was about taking existing applications and running them more easily in a bigger scale. But in that period of time what's also become possible is new kinds of computation that really weren't practical or even possible without that vast amount of compute. And so one result that came of that is something called the transformer AI model architecture. And Google came out with that, published a paper in 2017. And what that says is, with a transformer model you can actually train an arbitrarily large amount of data into a model, and see what happens. That's what Google demonstrated in 2017. The what happens is the really exciting part because when you do that, what you start to see, when models exceed a certain size that we had never really seen before all of a sudden they get what we call emerging capabilities of complex reasoning and reasoning outside a domain and reasoning with data. The kinds of things that people describe as spooky when they play with something like ChatGPT. That's the underlying term. We don't as an industry quite know why it happens or how it happens, but we can measure that it does. So cloud enables new kinds of math and science. New kinds of math and science allow new kinds of experimentation. And that experimentation has led to this new generation of models. >> So one of the debates we had on theCUBE at our Supercloud event last month was, what's the barriers to entry for say OpenAI, for instance? Obviously, I weighed in aggressively and said, "The barriers for getting into cloud are high because all the CapEx." And Howie Xu formerly VMware, now at ZScaler, he's an AI machine learning guy. He was like, "Well, you can spend $100 million and replicate it." I saw a quote that set up for 180,000 I can get this other package. What's the barriers to entry? Is ChatGPT or OpenAI, does it have sustainability? Is it easy to get into? What is the market like for AI? I mean, because a lot of entrepreneurs are jumping in. I mean, I just read a story today. San Francisco's got more inbound migration because of the AI action happening, Seattle's booming, Boston with MIT's been working on neural networks for generations. That's what we've found the answer. Get off the neural network, Boston jump on the AI bus. So there's total excitement for this. People are enthusiastic around this area. >> You can think of an iPhone versus Android tension that's happening today. In the iPhone world, there are proprietary models from OpenAI who you might consider as the leader. There's Cohere, there's AI21, there's Anthropic, Google's going to have their own, and a few others. These are proprietary models that developers can build on top of, get started really quickly. They're measured to have the highest accuracy and the highest performance today. That's the proprietary side. On the other side, there is an open source part of the world. These are a proliferation of model architectures that developers and practitioners can take off the shelf and train themselves. Typically found in Hugging face. What people seem to think is that the accuracy and performance of the open source models is something like 18 to 20 months behind the accuracy and performance of the proprietary models. But on the other hand, there's infinite flexibility for teams that are capable enough. So you're going to see teams choose sides based on whether they want speed or flexibility. >> That's interesting. And that brings up a point I was talking to a startup and the debate was, do you abstract away from the hardware and be software-defined or software-led on the AI side and let the hardware side just extremely accelerate on its own, 'cause it's flywheel? So again, back to proprietary, that's with hardware kind of bundled in, bolted on. Is it accelerator or is it bolted on or is it part of it? So to me, I think that the big struggle in understanding this is that which one will end up being right. I mean, is it a beta max versus VHS kind of thing going on? Or iPhone, Android, I mean iPhone makes a lot of sense, but if you're Apple, but is there an Apple moment in the machine learning? >> In proprietary models, here does seem to be a jump ball. That there's going to be a virtuous flywheel that emerges that, for example, all these excitement about ChatGPT. What's really exciting about it is it's really easy to use. The technology isn't so different from what we've seen before even from OpenAI. You mentioned a million users in a short period of time, all providing training data for OpenAI that makes their underlying models, their next generation even better. So it's not unreasonable to guess that there's going to be power laws that emerge on the proprietary side. What I think history has shown is that iPhone, Android, Windows, Linux, there seems to be gravity towards this yin and yang. And my guess, and what other people seem to think is going to be the case is that we're going to continue to see these two poles of AI. >> So let's get into the relationship with data because I've been emerging myself with ChatGPT, fascinated by the ease of use, yes, but also the fidelity of how you query it. And I felt like when I was doing writing SQL back in the eighties and nineties where SQL was emerging. You had to be really a guru at the SQL to get the answers you wanted. It seems like the querying into ChatGPT is a good thing if you know how to talk to it. Labeling whether your input is and it does a great job if you feed it right. If you ask a generic questions like Google. It's like a Google search. It gives you great format, sounds credible, but the facts are kind of wrong. >> That's right. >> That's where general consensus is coming on. So what does that mean? That means people are on one hand saying, "Ah, it's bullshit 'cause it's wrong." But I look at, I'm like, "Wow, that's that's compelling." 'Cause if you feed it the right data, so now we're in the data modeling here, so the role of data's going to be critical. Is there a data operating system emerging? Because if this thing continues to go the way it's going you can almost imagine as you would look at companies to invest in. Who's going to be right on this? What's going to scale? What's sustainable? What could build a durable company? It might not look what like what people think it is. I mean, I remember when Google started everyone thought it was the worst search engine because it wasn't a portal. But it was the best organic search on the planet became successful. So I'm trying to figure out like, okay, how do you read this? How do you read the tea leaves? >> Yeah. There are a few different ways that companies can differentiate themselves. Teams with galactic capabilities to take an open source model and then change the architecture and retrain and go down to the silicon. They can do things that might not have been possible for other teams to do. There's a company that that we're proud to be investors in called RunwayML that provides video accelerated, sorry, AI accelerated video editing capabilities. They were used in everything, everywhere all at once and some others. In order to build RunwayML, they needed a vision of what the future was going to look like and they needed to make deep contributions to the science that was going to enable all that. But not every team has those capabilities, maybe nor should they. So as far as how other teams are going to differentiate there's a couple of things that they can do. One is called prompt engineering where they shape on behalf of their own users exactly how the prompt to get fed to the underlying model. It's not clear whether that's going to be a durable problem or whether like Google, we consumers are going to start to get more intuitive about this. That's one. The second is what's called information retrieval. How can I get information about the world outside, information from a database or a data store or whatever service into these models so they can reason about them. And the third is, this is going to sound funny, but attribution. Just like you would do in a news report or an academic paper. If you can state where your facts are coming from, the downstream consumer or the human being who has to use that information actually is going to be able to make better sense of it and rely better on it. So that's prompt engineering, that's retrieval, and that's attribution. >> So that brings me to my next point I want to dig in on is the foundational model stack that you published. And I'll start by saying that with ChatGPT, if you take out the naysayers who are like throwing cold water on it about being a gimmick or whatever, and then you got the other side, I would call the alpha nerds who are like they can see, "Wow, this is amazing." This is truly NextGen. This isn't yesterday's chatbot nonsense. They're like, they're all over it. It's that everybody's using it right now in every vertical. I heard someone using it for security logs. I heard a data center, hardware vendor using it for pushing out appsec review updates. I mean, I've heard corner cases. We're using it for theCUBE to put our metadata in. So there's a horizontal use case of value. So to me that tells me it's a market there. So when you have horizontal scalability in the use case you're going to have a stack. So you publish this stack and it has an application at the top, applications like Jasper out there. You're seeing ChatGPT. But you go after the bottom, you got silicon, cloud, foundational model operations, the foundational models themselves, tooling, sources, actions. Where'd you get this from? How'd you put this together? Did you just work backwards from the startups or was there a thesis behind this? Could you share your thoughts behind this foundational model stack? >> Sure. Well, I'm a recovering product manager and my job that I think about as a product manager is who is my customer and what problem he wants to solve. And so to put myself in the mindset of an application developer and a founder who is actually my customer as a partner at Madrona, I think about what technology and resources does she need to be really powerful, to be able to take a brilliant idea, and actually bring that to life. And if you spend time with that community, which I do and I've met with hundreds of founders now who are trying to do exactly this, you can see that the stack is emerging. In fact, we first drew it in, not in January 2023, but October 2022. And if you look at the difference between the October '22 and January '23 stacks you're going to see that holes in the stack that we identified in October around tooling and around foundation model ops and the rest are organically starting to get filled because of how much demand from the developers at the top of the stack. >> If you look at the young generation coming out and even some of the analysts, I was just reading an analyst report on who's following the whole data stacks area, Databricks, Snowflake, there's variety of analytics, realtime AI, data's hot. There's a lot of engineers coming out that were either data scientists or I would call data platform engineering folks are becoming very key resources in this area. What's the skillset emerging and what's the mindset of that entrepreneur that sees the opportunity? How does these startups come together? Is there a pattern in the formation? Is there a pattern in the competency or proficiency around the talent behind these ventures? >> Yes. I would say there's two groups. The first is a very distinct pattern, John. For the past 10 years or a little more we've seen a pattern of democratization of ML where more and more people had access to this powerful science and technology. And since about 2017, with the rise of the transformer architecture in these foundation models, that pattern has reversed. All of a sudden what has become broader access is now shrinking to a pretty small group of scientists who can actually train and manipulate the architectures of these models themselves. So that's one. And what that means is the teams who can do that have huge ability to make the future happen in ways that other people don't have access to yet. That's one. The second is there is a broader population of people who by definition has even more collective imagination 'cause there's even more people who sees what should be possible and can use things like the proprietary models, like the OpenAI models that are available off the shelf and try to create something that maybe nobody has seen before. And when they do that, Jasper AI is a great example of that. Jasper AI is a company that creates marketing copy automatically with generative models such as GPT-3. They do that and it's really useful and it's almost fun for a marketer to use that. But there are going to be questions of how they can defend that against someone else who has access to the same technology. It's a different population of founders who has to find other sources of differentiation without being able to go all the way down to the the silicon and the science. >> Yeah, and it's going to be also opportunity recognition is one thing. Building a viable venture product market fit. You got competition. And so when things get crowded you got to have some differentiation. I think that's going to be the key. And that's where I was trying to figure out and I think data with scale I think are big ones. Where's the vulnerability in the stack in terms of gaps? Where's the white space? I shouldn't say vulnerability. I should say where's the opportunity, where's the white space in the stack that you see opportunities for entrepreneurs to attack? >> I would say there's two. At the application level, there is almost infinite opportunity, John, because almost every kind of application is about to be reimagined or disrupted with a new generation that takes advantage of this really powerful new technology. And so if there is a kind of application in almost any vertical, it's hard to rule something out. Almost any vertical that a founder wishes she had created the original app in, well, now it's her time. So that's one. The second is, if you look at the tooling layer that we discussed, tooling is a really powerful way that you can provide more flexibility to app developers to get more differentiation for themselves. And the tooling layer is still forming. This is the interface between the models themselves and the applications. Tools that help bring in data, as you mentioned, connect to external actions, bring context across multiple calls, chain together multiple models. These kinds of things, there's huge opportunity there. >> Well, Jon, I really appreciate you coming in. I had a couple more questions, but I will take a minute to read some of your bios for the audience and we'll get into, I won't embarrass you, but I want to set the context. You said you were recovering product manager, 10 plus years at AWS. Obviously, recovering from AWS, which is a whole nother dimension of recovering. In all seriousness, I talked to Andy Jassy around that time and Dr. Matt Wood and it was about that time when AI was just getting on the radar when they started. So you guys started seeing the wave coming in early on. So I remember at that time as Amazon was starting to grow significantly and even just stock price and overall growth. From a tech perspective, it was pretty clear what was coming, so you were there when this tsunami hit. >> Jon: That's right. >> And you had a front row seat building tech, you were led the product teams for Computer Vision AI, Textract, AI intelligence for document processing, recognition for image and video analysis. You wrote the business product plan for AWS IoT and Greengrass, which we've covered a lot in theCUBE, which extends out to the whole edge thing. So you know a lot about AI/ML, edge computing, IOT, messaging, which I call the law of small numbers that scale become big. This is a big new thing. So as a former AWS leader who's been there and at Madrona, what's your investment thesis as you start to peruse the landscape and talk to entrepreneurs as you got the stack? What's the big picture? What are you looking for? What's the thesis? How do you see this next five years emerging? >> Five years is a really long time given some of this science is only six months out. I'll start with some, no pun intended, some foundational things. And we can talk about some implications of the technology. The basics are the same as they've always been. We want, what I like to call customers with their hair on fire. So they have problems, so urgent they'll buy half a product. The joke is if your hair is on fire you might want a bucket of cold water, but you'll take a tennis racket and you'll beat yourself over the head to put the fire out. You want those customers 'cause they'll meet you more than halfway. And when you find them, you can obsess about them and you can get better every day. So we want customers with their hair on fire. We want founders who have empathy for those customers, understand what is going to be required to serve them really well, and have what I like to call founder-market fit to be able to build the products that those customers are going to need. >> And because that's a good strategy from an emerging, not yet fully baked out requirements definition. >> Jon: That's right. >> Enough where directionally they're leaning in, more than in, they're part of the product development process. >> That's right. And when you're doing early stage development, which is where I personally spend a lot of my time at the seed and A and a little bit beyond that stage often that's going to be what you have to go on because the future is going to be so complex that you can't see the curves beyond it. But if you have customers with their hair on fire and talented founders who have the capability to serve those customers, that's got me interested. >> So if I'm an entrepreneur, I walk in and say, "I have customers that have their hair on fire." What kind of checks do you write? What's the kind of the average you're seeing for seed and series? Probably seed, seed rounds and series As. >> It can depend. I have seen seed rounds of double digit million dollars. I have seen seed rounds much smaller than that. It really depends on what is going to be the right thing for these founders to prove out the hypothesis that they're testing that says, "Look, we have this customer with her hair on fire. We think we can build at least a tennis racket that she can use to start beating herself over the head and put the fire out. And then we're going to have something really interesting that we can scale up from there and we can make the future happen. >> So it sounds like your advice to founders is go out and find some customers, show them a product, don't obsess over full completion, get some sort of vibe on fit and go from there. >> Yeah, and I think by the time founders come to me they may not have a product, they may not have a deck, but if they have a customer with her hair on fire, then I'm really interested. >> Well, I always love the professional services angle on these markets. You go in and you get some business and you understand it. Walk away if you don't like it, but you see the hair on fire, then you go in product mode. >> That's right. >> All Right, Jon, thank you for coming on theCUBE. Really appreciate you stopping by the studio and good luck on your investments. Great to see you. >> You too. >> Thanks for coming on. >> Thank you, Jon. >> CUBE coverage here at Palo Alto. I'm John Furrier, your host. More coverage with CUBE Conversations after this break. (upbeat music)

(bright upbeat music) >> Hello, everyone. Welcome to SiliconANGLE News breaking story here. Amazon Web Services expanding their relationship with Hugging Face, breaking news here on SiliconANGLE. I'm John Furrier, SiliconANGLE reporter, founder, and also co-host of theCUBE. And I have with me, Swami, from Amazon Web Services, vice president of database, analytics, machine learning with AWS. Swami, great to have you on for this breaking news segment on AWS's big news. Thanks for coming on and taking the time. >> Hey, John, pleasure to be here. >> You know- >> Looking forward to it. >> We've had many conversations on theCUBE over the years, we've watched Amazon really move fast into the large data modeling, SageMaker became a very smashing success, obviously you've been on this for a while. Now with ChatGPT OpenAI, a lot of buzz going mainstream, takes it from behind the curtain inside the ropes, if you will, in the industry to a mainstream. And so this is a big moment, I think, in the industry, I want to get your perspective, because your news with Hugging Face, I think is another tell sign that we're about to tip over into a new accelerated growth around making AI now application aware, application centric, more programmable, more API access. What's the big news about, with AWS Hugging Face, you know, what's going on with this announcement? >> Yeah. First of all, they're very excited to announce our expanded collaboration with Hugging Face, because with this partnership, our goal, as you all know, I mean, Hugging Face, I consider them like the GitHub for machine learning. And with this partnership, Hugging Face and AWS, we'll be able to democratize AI for a broad range of developers, not just specific deep AI startups. And now with this, we can accelerate the training, fine tuning and deployment of these large language models, and vision models from Hugging Face in the cloud. And the broader context, when you step back and see what customer problem we are trying to solve with this announcement, essentially if you see these foundational models, are used to now create like a huge number of applications, suggest like tech summarization, question answering, or search image generation, creative, other things. And these are all stuff we are seeing in the likes of these ChatGPT style applications. But there is a broad range of enterprise use cases that we don't even talk about. And it's because these kind of transformative, generative AI capabilities and models are not available to, I mean, millions of developers. And because either training these elements from scratch can be very expensive or time consuming and need deep expertise, or more importantly, they don't need these generic models, they need them to be fine tuned for the specific use cases. And one of the biggest complaints we hear is that these models, when they try to use it for real production use cases, they are incredibly expensive to train and incredibly expensive to run inference on, to use it at a production scale. So, and unlike web search style applications, where the margins can be really huge, here in production use cases and enterprises, you want efficiency at scale. That's where Hugging Face and AWS share our mission. And by integrating with Trainium and Inferentia, we're able to handle the cost efficient training and inference at scale, I'll deep dive on it. And by teaming up on the SageMaker front, now the time it takes to build these models and fine tune them is also coming down. So that's what makes this partnership very unique as well. So I'm very excited. >> I want to get into the time savings and the cost savings as well on the training and inference, it's a huge issue, but before we get into that, just how long have you guys been working with Hugging Face? I know there's a previous relationship, this is an expansion of that relationship, can you comment on what's different about what's happened before and then now? >> Yeah. So, Hugging Face, we have had a great relationship in the past few years as well, where they have actually made their models available to run on AWS, you know, fashion. Even in fact, their Bloom Project was something many of our customers even used. Bloom Project, for context, is their open source project which builds a GPT-3 style model. And now with this expanded collaboration, now Hugging Face selected AWS for that next generation office generative AI model, building on their highly successful Bloom Project as well. And the nice thing is, now, by direct integration with Trainium and Inferentia, where you get cost savings in a really significant way, now, for instance, Trn1 can provide up to 50% cost to train savings, and Inferentia can deliver up to 60% better costs, and four x more higher throughput than (indistinct). Now, these models, especially as they train that next generation generative AI models, it is going to be, not only more accessible to all the developers, who use it in open, so it'll be a lot cheaper as well. And that's what makes this moment really exciting, because we can't democratize AI unless we make it broadly accessible and cost efficient and easy to program and use as well. >> Yeah. >> So very exciting. >> I'll get into the SageMaker and CodeWhisperer angle in a second, but you hit on some good points there. One, accessibility, which is, I call the democratization, which is getting this in the hands of developers, and/or AI to develop, we'll get into that in a second. So, access to coding and Git reasoning is a whole nother wave. But the three things I know you've been working on, I want to put in the buckets here and comment, one, I know you've, over the years, been working on saving time to train, that's a big point, you mentioned some of those stats, also cost, 'cause now cost is an equation on, you know, bundling whether you're uncoupling with hardware and software, that's a big issue. Where do I find the GPUs? Where's the horsepower cost? And then also sustainability. You've mentioned that in the past, is there a sustainability angle here? Can you talk about those three things, time, cost, and sustainability? >> Certainly. So if you look at it from the AWS perspective, we have been supporting customers doing machine learning for the past years. Just for broader context, Amazon has been doing ML the past two decades right from the early days of ML powered recommendation to actually also supporting all kinds of generative AI applications. If you look at even generative AI application within Amazon, Amazon search, when you go search for a product and so forth, we have a team called MFi within Amazon search that helps bring these large language models into creating highly accurate search results. And these are created with models, really large models with tens of billions of parameters, scales to thousands of training jobs every month and trained on large model of hardware. And this is an example of a really good large language foundation model application running at production scale, and also, of course, Alexa, which uses a large generator model as well. And they actually even had a research paper that showed that they are more, and do better in accuracy than other systems like GPT-3 and whatnot. So, and we also touched on things like CodeWhisperer, which uses generative AI to improve developer productivity, but in a responsible manner, because 40% of some of the studies show 40% of this generated code had serious security flaws in it. This is where we didn't just do generative AI, we combined with automated reasoning capabilities, which is a very, very useful technique to identify these issues and couple them so that it produces highly secure code as well. Now, all these learnings taught us few things, and which is what you put in these three buckets. And yeah, like more than 100,000 customers using ML and AI services, including leading startups in the generative AI space, like stability AI, AI21 Labs, or Hugging Face, or even Alexa, for that matter. They care about, I put them in three dimension, one is around cost, which we touched on with Trainium and Inferentia, where we actually, the Trainium, you provide to 50% better cost savings, but the other aspect is, Trainium is a lot more power efficient as well compared to traditional one. And Inferentia is also better in terms of throughput, when it comes to what it is capable of. Like it is able to deliver up to three x higher compute performance and four x higher throughput, compared to it's previous generation, and it is extremely cost efficient and power efficient as well. >> Well. >> Now, the second element that really is important is in a day, developers deeply value the time it takes to build these models, and they don't want to build models from scratch. And this is where SageMaker, which is, even going to Kaggle uses, this is what it is, number one, enterprise ML platform. What it did to traditional machine learning, where tens of thousands of customers use StageMaker today, including the ones I mentioned, is that what used to take like months to build these models have dropped down to now a matter of days, if not less. Now, a generative AI, the cost of building these models, if you look at the landscape, the model parameter size had jumped by more than thousand X in the past three years, thousand x. And that means the training is like a really big distributed systems problem. How do you actually scale these model training? How do you actually ensure that you utilize these efficiently? Because these machines are very expensive, let alone they consume a lot of power. So, this is where SageMaker capability to build, automatically train, tune, and deploy models really concern this, especially with this distributor training infrastructure, and those are some of the reasons why some of the leading generative AI startups are actually leveraging it, because they do not want a giant infrastructure team, which is constantly tuning and fine tuning, and keeping these clusters alive. >> It sounds like a lot like what startups are doing with the cloud early days, no data center, you move to the cloud. So, this is the trend we're seeing, right? You guys are making it easier for developers with Hugging Face, I get that. I love that GitHub for machine learning, large language models are complex and expensive to build, but not anymore, you got Trainium and Inferentia, developers can get faster time to value, but then you got the transformers data sets, token libraries, all that optimized for generator. This is a perfect storm for startups. Jon Turow, a former AWS person, who used to work, I think for you, is now a VC at Madrona Venture, he and I were talking about the generator AI landscape, it's exploding with startups. Every alpha entrepreneur out there is seeing this as the next frontier, that's the 20 mile stairs, next 10 years is going to be huge. What is the big thing that's happened? 'Cause some people were saying, the founder of Yquem said, "Oh, the start ups won't be real, because they don't all have AI experience." John Markoff, former New York Times writer told me that, AI, there's so much work done, this is going to explode, accelerate really fast, because it's almost like it's been waiting for this moment. What's your reaction? >> I actually think there is going to be an explosion of startups, not because they need to be AI startups, but now finally AI is really accessible or going to be accessible, so that they can create remarkable applications, either for enterprises or for disrupting actually how customer service is being done or how creative tools are being built. And I mean, this is going to change in many ways. When we think about generative AI, we always like to think of how it generates like school homework or arts or music or whatnot, but when you look at it on the practical side, generative AI is being actually used across various industries. I'll give an example of like Autodesk. Autodesk is a customer who runs an AWS and SageMaker. They already have an offering that enables generated design, where designers can generate many structural designs for products, whereby you give a specific set of constraints and they actually can generate a structure accordingly. And we see similar kind of trend across various industries, where it can be around creative media editing or various others. I have the strong sense that literally, in the next few years, just like now, conventional machine learning is embedded in every application, every mobile app that we see, it is pervasive, and we don't even think twice about it, same way, like almost all apps are built on cloud. Generative AI is going to be part of every startup, and they are going to create remarkable experiences without needing actually, these deep generative AI scientists. But you won't get that until you actually make these models accessible. And I also don't think one model is going to rule the world, then you want these developers to have access to broad range of models. Just like, go back to the early days of deep learning. Everybody thought it is going to be one framework that will rule the world, and it has been changing, from Caffe to TensorFlow to PyTorch to various other things. And I have a suspicion, we had to enable developers where they are, so. >> You know, Dave Vellante and I have been riffing on this concept called super cloud, and a lot of people have co-opted to be multicloud, but we really were getting at this whole next layer on top of say, AWS. You guys are the most comprehensive cloud, you guys are a super cloud, and even Adam and I are talking about ISVs evolving to ecosystem partners. I mean, your top customers have ecosystems building on top of it. This feels like a whole nother AWS. How are you guys leveraging the history of AWS, which by the way, had the same trajectory, startups came in, they didn't want to provision a data center, the heavy lifting, all the things that have made Amazon successful culturally. And day one thinking is, provide the heavy lifting, undifferentiated heavy lifting, and make it faster for developers to program code. AI's got the same thing. How are you guys taking this to the next level, because now, this is an opportunity for the competition to change the game and take it over? This is, I'm sure, a conversation, you guys have a lot of things going on in AWS that makes you unique. What's the internal and external positioning around how you take it to the next level? >> I mean, so I agree with you that generative AI has a very, very strong potential in terms of what it can enable in terms of next generation application. But this is where Amazon's experience and expertise in putting these foundation models to work internally really has helped us quite a bit. If you look at it, like amazon.com search is like a very, very important application in terms of what is the customer impact on number of customers who use that application openly, and the amount of dollar impact it does for an organization. And we have been doing it silently for a while now. And the same thing is true for like Alexa too, which actually not only uses it for natural language understanding other city, even national leverages is set for creating stories and various other examples. And now, our approach to it from AWS is we actually look at it as in terms of the same three tiers like we did in machine learning, because when you look at generative AI, we genuinely see three sets of customers. One is, like really deep technical expert practitioner startups. These are the startups that are creating the next generation models like the likes of stability AIs or Hugging Face with Bloom or AI21. And they generally want to build their own models, and they want the best price performance of their infrastructure for training and inference. That's where our investments in silicon and hardware and networking innovations, where Trainium and Inferentia really plays a big role. And we can nearly do that, and that is one. The second middle tier is where I do think developers don't want to spend time building their own models, let alone, they actually want the model to be useful to that data. They don't need their models to create like high school homeworks or various other things. What they generally want is, hey, I had this data from my enterprises that I want to fine tune and make it really work only for this, and make it work remarkable, can be for tech summarization, to generate a report, or it can be for better Q&A, and so forth. This is where we are. Our investments in the middle tier with SageMaker, and our partnership with Hugging Face and AI21 and co here are all going to very meaningful. And you'll see us investing, I mean, you already talked about CodeWhisperer, which is an open preview, but we are also partnering with a whole lot of top ISVs, and you'll see more on this front to enable the next wave of generated AI apps too, because this is an area where we do think lot of innovation is yet to be done. It's like day one for us in this space, and we want to enable that huge ecosystem to flourish. >> You know, one of the things Dave Vellante and I were talking about in our first podcast we just did on Friday, we're going to do weekly, is we highlighted the AI ChatGPT example as a horizontal use case, because everyone loves it, people are using it in all their different verticals, and horizontal scalable cloud plays perfectly into it. So I have to ask you, as you look at what AWS is going to bring to the table, a lot's changed over the past 13 years with AWS, a lot more services are available, how should someone rebuild or re-platform and refactor their application of business with AI, with AWS? What are some of the tools that you see and recommend? Is it Serverless, is it SageMaker, CodeWhisperer? What do you think's going to shine brightly within the AWS stack, if you will, or service list, that's going to be part of this? As you mentioned, CodeWhisperer and SageMaker, what else should people be looking at as they start tinkering and getting all these benefits, and scale up their ups? >> You know, if we were a startup, first, I would really work backwards from the customer problem I try to solve, and pick and choose, bar, I don't need to deal with the undifferentiated heavy lifting, so. And that's where the answer is going to change. If you look at it then, the answer is not going to be like a one size fits all, so you need a very strong, I mean, granted on the compute front, if you can actually completely accurate it, so unless, I will always recommend it, instead of running compute for running your ups, because it takes care of all the undifferentiated heavy lifting, but on the data, and that's where we provide a whole variety of databases, right from like relational data, or non-relational, or dynamo, and so forth. And of course, we also have a deep analytical stack, where data directly flows from our relational databases into data lakes and data virus. And you can get value along with partnership with various analytical providers. The area where I do think fundamentally things are changing on what people can do is like, with CodeWhisperer, I was literally trying to actually program a code on sending a message through Twilio, and I was going to pull up to read a documentation, and in my ID, I was actually saying like, let's try sending a message to Twilio, or let's actually update a Route 53 error code. All I had to do was type in just a comment, and it actually started generating the sub-routine. And it is going to be a huge time saver, if I were a developer. And the goal is for us not to actually do it just for AWS developers, and not to just generate the code, but make sure the code is actually highly secure and follows the best practices. So, it's not always about machine learning, it's augmenting with automated reasoning as well. And generative AI is going to be changing, and not just in how people write code, but also how it actually gets built and used as well. You'll see a lot more stuff coming on this front. >> Swami, thank you for your time. I know you're super busy. Thank you for sharing on the news and giving commentary. Again, I think this is a AWS moment and industry moment, heavy lifting, accelerated value, agility. AIOps is going to be probably redefined here. Thanks for sharing your commentary. And we'll see you next time, I'm looking forward to doing more follow up on this. It's going to be a big wave. Thanks. >> Okay. Thanks again, John, always a pleasure. >> Okay. This is SiliconANGLE's breaking news commentary. I'm John Furrier with SiliconANGLE News, as well as host of theCUBE. Swami, who's a leader in AWS, has been on theCUBE multiple times. We've been tracking the growth of how Amazon's journey has just been exploding past five years, in particular, past three. You heard the numbers, great performance, great reviews. This is a watershed moment, I think, for the industry, and it's going to be a lot of fun for the next 10 years. Thanks for watching. (bright music)