>>From around the globe. It's the cube with digital coverage of AWS reinvent 2020 sponsored by Intel, AWS, and our community partners. >>Hi everyone. Welcome back to the cubes. Virtual coverage of AWS reinvent 2020 virtual. We are the cube virtual I'm John ferry, your host with my coach, Dave Alante for keynote analysis from Swami's machine learning, all things, data huge. Instead of announcements, the first ever machine learning keynote at a re-invent Dave. Great to see you. Thanks Johnny. And from Boston, I'm here in Palo Alto. We're doing the cube remote cube virtual. Great to see you. >>Yeah, good to be here, John, as always. Wall-to-wall love it. So, so, John, um, how about I give you my, my key highlights from the, uh, from the keynote today, I had, I had four kind of curated takeaways. So the first is that AWS is, is really trying to simplify machine learning and use machine intelligence into all applications. And if you think about it, it's good news for organizations because they're not the become machine learning experts have invent machine learning. They can buy it from Amazon. I think the second is they're trying to simplify the data pipeline. The data pipeline today is characterized by a series of hyper specialized individuals. It engineers, data scientists, quality engineers, analysts, developers. These are folks that are largely live in their own swim lane. Uh, and while they collaborate, uh, there's still a fairly linear and complicated data pipeline, uh, that, that a business person or a data product builder has to go through Amazon making some moves to the front of simplify that they're expanding data access to the line of business. I think that's a key point. Is there, there increasingly as people build data products and data services that can monetize, you know, for their business, either cut costs or generate revenue, they can expand that into line of business where there's there's domain context. And I think the last thing is this theme that we talked about the other day, John of extending Amazon, AWS to the edge that we saw that as well in a number of machine learning tools that, uh, Swami talked about. >>Yeah, it was great by the way, we're live here, uh, in Palo Alto in Boston covering the analysis, tons of content on the cube, check out the cube.net and also check out at reinvent. There's a cube section as there's some links to so on demand videos with all the content we've had. Dave, I got to say one of the things that's apparent to me, and this came out of my one-on-one with Andy Jassy and Andy Jassy talked about in his keynote is he kind of teased out this idea of training versus a more value add machine learning. And you saw that today in today's announcement. To me, the big revelation was that the training aspect of machine learning, um, is what can be automated away. And it's under a lot of controversy around it. Recently, a Google paper came out and the person was essentially kind of, kind of let go for this. >>But the idea of doing these training algorithms, some are saying is causes more harm to the environment than it does good because of all the compute power it takes. So you start to see the positioning of training, which can be automated away and served up with, you know, high powered ships and that's, they consider that undifferentiated heavy lifting. In my opinion, they didn't say that, but that's clearly what I see coming out of this announcement. The other thing that I saw Dave that's notable is you saw them clearly taking a three lane approach to this machine, learning the advanced builders, the advanced coders and the developers, and then database and data analysts, three swim lanes of personas of target audience. Clearly that is in line with SageMaker and the embedded stuff. So two big revelations, more horsepower required to process training and modeling. Okay. And to the expansion of the personas that are going to be using machine learning. So clearly this is a, to me, a big trend wave that we're seeing that validates some of the startups and I'll see their SageMaker and some of their products. >>Well, as I was saying at the top, I think Amazon's really trying, working hard on simplifying the whole process. And you mentioned training and, and a lot of times people are starting from scratch when they have to train models and retrain models. And so what they're doing is they're trying to create reusable components, uh, and allow people to, as you pointed out to automate and streamline some of that heavy lifting, uh, and as well, they talked a lot about, uh, doing, doing AI inferencing at the edge. And you're seeing, you know, they, they, uh, Swami talked about several foundational premises and the first being a foundation of frameworks. And you think about that at the, at the lowest level of their S their ML stack. They've got, you know, GPU's different processors, inferential, all these alternative processes, processors, not just the, the Xav six. And so these are very expensive resources and Swami talked a lot about, uh, and his colleagues talked a lot about, well, a lot of times the alternative processor is sitting there, you know, waiting, waiting, waiting. And so they're really trying to drive efficiency and speed. They talked a lot about compressing the time that it takes to, to run these, these models, uh, from, from sometimes weeks down to days, sometimes days down to hours and minutes. >>Yeah. Let's, let's unpack these four areas. Let's stay on the firm foundation because that's their core competency infrastructure as a service. Clearly they're laying that down. You put the processors, but what's interesting is the TensorFlow 92% of tensor flows on Amazon. The other thing is that pie torch surprisingly is back up there, um, with massive adoption and the numbers on pie torch literally is on fire. I was coming in and joke on Twitter. Um, we, a PI torch is telling because that means that TensorFlow is originally part of Google is getting, is getting a little bit diluted with other frameworks, and then you've got MX net, some other things out there. So the fact that you've got PI torch 91% and then TensorFlow 92% on 80 bucks is a huge validation. That means that the majority of most machine learning development and deep learning is happening on AWS. Um, >>Yeah, cloud-based, by the way, just to clarify, that's the 90% of cloud-based cloud, uh, TensorFlow runs on and 91% of cloud-based PI torch runs on ADM is amazingly massive numbers. >>Yeah. And I think that the, the processor has to show that it's not trivial to do the machine learning, but, you know, that's where the infrared internship came in. That's kind of where they want to go lay down that foundation. And they had Tanium, they had trainee, um, they had, um, infrared chow was the chip. And then, you know, just true, you know, distributed training training on SageMaker. So you got the chip and then you've got Sage makers, the middleware games, almost like a machine learning stack. That's what they're putting out there >>And how bad a Gowdy, which was, which is, which is a patrol also for training, which is an Intel based chip. Uh, so that was kind of interesting. So a lot of new chips and, and specialized just, we've been talking about this for awhile, particularly as you get to the edge and do AI inferencing, you need, uh, you know, a different approach than we're used to with the general purpose microbes. >>So what gets your take on tenant? Number two? So tenant number one, clearly infrastructure, a lot of announcements we'll go through those, review them at the end, but tenant number two, that Swami put out there was creating the shortest path to success for builders or machine learning builders. And I think here you lays out the complexity, Dave butts, mostly around methodology, and, you know, the value activities required to execute. And again, this points to the complexity problem that they have. What's your take on this? >>Yeah. Well you think about, again, I'm talking about the pipeline, you collect data, you just data, you prepare that data, you analyze that data. You, you, you make sure that it's it's high quality and then you start the training and then you're iterating. And so they really trying to automate as much as possible and simplify as much as possible. What I really liked about that segment of foundation, number two, if you will, is the example, the customer example of the speaker from the NFL, you know, talked about, uh, you know, the AWS stats that we see in the commercials, uh, next gen stats. Uh, and, and she talked about the ways in which they've, well, we all know they've, they've rearchitected helmets. Uh, they've been, it's really a very much database. It was interesting to see they had the spectrum of the helmets that were, you know, the safest, most safe to the least safe and how they've migrated everybody in the NFL to those that they, she started a 24%. >>It was interesting how she wanted a 24% reduction in reported concussions. You know, you got to give the benefit of the doubt and assume some of that's through, through the data. But you know, some of that could be like, you know, Julian Edelman popping up off the ground. When, you know, we had a concussion, he doesn't want to come out of the game with the new protocol, but no doubt, they're collecting more data on this stuff, and it's not just head injuries. And she talked about ankle injuries, knee injuries. So all this comes from training models and reducing the time it takes to actually go from raw data to insights. >>Yeah. I mean, I think the NFL is a great example. You and I both know how hard it is to get the NFL to come on and do an interview. They're very coy. They don't really put their name on anything much because of the value of the NFL, this a meaningful partnership. You had the, the person onstage virtually really going into some real detail around the depth of the partnership. So to me, it's real, first of all, I love stat cast 11, anything to do with what they do with the stats is phenomenal at this point. So the real world example, Dave, that you starting to see sports as one metaphor, healthcare, and others are going to see those coming in to me, totally a tale sign that Amazon's continued to lead. The thing that got my attention was is that it is an IOT problem, and there's no reason why they shouldn't get to it. I mean, some say that, Oh, concussion, NFL is just covering their butt. They don't have to, this is actually really working. So you got the tech, why not use it? And they are. So that, to me, that's impressive. And I think that's, again, a digital transformation sign that, that, you know, in the NFL is doing it. It's real. Um, because it's just easier. >>I think, look, I think, I think it's easy to criticize the NFL, but the re the reality is, is there anything old days? It was like, Hey, you get your bell rung and get back out there. That's just the way it was a football players, you know, but Ted Johnson was one of the first and, you know, bill Bellacheck was, was, you know, the guy who sent him back out there with a concussion, but, but he was very much outspoken. You've got to give the NFL credit. Uh, it didn't just ignore the problem. Yeah. Maybe it, it took a little while, but you know, these things take some time because, you know, it's generally was generally accepted, you know, back in the day that, okay, Hey, you'd get right back out there, but, but the NFL has made big investments there. And you can say, you got to give him, give him props for that. And especially given that they're collecting all this data. That to me is the most interesting angle here is letting the data inform the actions. >>And next step, after the NFL, they had this data prep data Wrangler news, that they're now integrating snowflakes, Databricks, Mongo DB, into SageMaker, which is a theme there of Redshift S3 and Lake formation into not the other way around. So again, you've been following this pretty closely, uh, specifically the snowflake recent IPO and their success. Um, this is an ecosystem play for Amazon. What does it mean? >>Well, a couple of things, as we, as you well know, John, when you first called me up, I was in Dallas and I flew into New York and an ice storm to get to the one of the early Duke worlds. You know, and back then it was all batch. The big data was this big batch job. And today you want to combine that batch. There's still a lot of need for batch, but when people want real time inferencing and AWS is bringing that together and they're bringing in multiple data sources, you mentioned Databricks and snowflake Mongo. These are three platforms that are doing very well in the market and holding a lot of data in AWS and saying, okay, Hey, we want to be the brain in the middle. You can import data from any of those sources. And I'm sure they're going to add more over time. Uh, and so they talked about 300 pre-configured data transformations, uh, that now come with stage maker of SageMaker studio with essentially, I've talked about this a lot. It's essentially abstracting away the, it complexity, the whole it operations piece. I mean, it's the same old theme that AWS is just pointing. It's its platform and its cloud at non undifferentiated, heavy lifting. And it's moving it up the stack now into the data life cycle and data pipeline, which is one of the biggest blockers to monetizing data. >>Expand on that more. What does that actually mean? I'm an it person translate that into it. Speak. Yeah. >>So today, if you're, if you're a business person and you want, you want the answers, right, and you want say to adjust a new data source, so let's say you want to build a new, new product. Um, let me give an example. Let's say you're like a Spotify, make it up. And, and you do music today, but let's say you want to add, you know, movies, or you want to add podcasts and you want to start monetizing that you want to, you want to identify, who's watching what you want to create new metadata. Well, you need new data sources. So what you do as a business person that wants to create that new data product, let's say for podcasts, you have to knock on the door, get to the front of the data pipeline line and say, okay, Hey, can you please add this data source? >>And then everybody else down the line has to get in line and Hey, this becomes a new data source. And it's this linear process where very specialized individuals have to do their part. And then at the other end, you know, it comes to self-serve capability that somebody can use to either build dashboards or build a data product. In a lot of that middle part is our operational details around deploying infrastructure, deploying, you know, training machine learning models that a lot of Python coding. Yeah. There's SQL queries that have to be done. So a lot of very highly specialized activities, what Amazon is doing, my takeaway is they're really streamlining a lot of those activities, removing what they always call the non undifferentiated, heavy lifting abstracting away that it complexity to me, this is a real positive sign, because it's all about the technology serving the business, as opposed to historically, it's the business begging the technology department to please help me. The technology department obviously evolving from, you know, the, the glass house, if you will, to this new data, data pipeline data, life cycle. >>Yeah. I mean, it's classic agility to take down those. I mean, it's undifferentiated, I guess, but if it actually works, just create a differentiated product. So, but it's just log it's that it's, you can debate that kind of aspect of it, but I hear what you're saying, just get rid of it and make it simpler. Um, the impact of machine learning is Dave is one came out clear on this, uh, SageMaker clarify announcement, which is a bias decision algorithm. They had an expert, uh, nationally CFUs presented essentially how they're dealing with the, the, the bias piece of it. I thought that was very interesting. What'd you think? >>Well, so humans are biased and so humans build models or models are inherently biased. And so I thought it was, you know, this is a huge problem to big problems in artificial intelligence. One is the inherent bias in the models. And the second is the lack of transparency that, you know, they call it the black box problem, like, okay, I know there was an answer there, but how did it get to that answer and how do I trace it back? Uh, and so Amazon is really trying to attack those, uh, with, with, with clarify. I wasn't sure if it was clarity or clarified, I think it's clarity clarify, um, a lot of entirely certain how it works. So we really have to dig more into that, but it's essentially identifying situations where there is bias flagging those, and then, you know, I believe making recommendations as to how it can be stamped. >>Nope. Yeah. And also some other news deep profiling for debugger. So you could make a debugger, which is a deep profile on neural network training, um, which is very cool again on that same theme of profiling. The other thing that I found >>That remind me, John, if I may interrupt there reminded me of like grammar corrections and, you know, when you're typing, it's like, you know, bug code corrections and automated debugging, try this. >>It wasn't like a better debugger come on. We, first of all, it should be bug free code, but, um, you know, there's always biases of the data is critical. Um, the other news I thought was interesting and then Amazon's claiming this is the first SageMaker pipelines for purpose-built CIC D uh, for machine learning, bringing machine learning into a developer construct. And I think this started bringing in this idea of the edge manager where you have, you know, and they call it the about machine, uh, uh, SageMaker store storing your functions of this idea of managing and monitoring machine learning modules effectively is on the edge. And, and through the development process is interesting and really targeting that developer, Dave, >>Yeah, applying CIC D to the machine learning and machine intelligence has always been very challenging because again, there's so many piece parts. And so, you know, I said it the other day, it's like a lot of the innovations that Amazon comes out with are things that have problems that have come up given the pace of innovation that they're putting forth. And, and it's like the customers drinking from a fire hose. We've talked about this at previous reinvents and the, and the customers keep up with the pace of Amazon. So I see this as Amazon trying to reduce friction, you know, across its entire stack. Most, for example, >>Let me lay it out. A slide ahead, build machine learning, gurus developers, and then database and data analysts, clearly database developers and data analysts are on their radar. This is not the first time we've heard that. But we, as the kind of it is the first time we're starting to see products materialized where you have machine learning for databases, data warehouse, and data lakes, and then BI tools. So again, three different segments, the databases, the data warehouse and data lakes, and then the BI tools, three areas of machine learning, innovation, where you're seeing some product news, your, your take on this natural evolution. >>Well, well, it's what I'm saying up front is that the good news for, for, for our customers is you don't have to be a Google or Amazon or Facebook to be a super expert at AI. Uh, companies like Amazon are going to be providing products that you can then apply to your business. And, and it's allowed you to infuse AI across your entire application portfolio. Amazon Redshift ML was another, um, example of them, abstracting complexity. They're taking, they're taking S3 Redshift and SageMaker complexity and abstracting that and presenting it to the data analysts. So that, that, that individual can worry about, you know, again, getting to the insights, it's injecting ML into the database much in the same way, frankly, the big query has done that. And so that's a huge, huge positive. When you talk to customers, they, they love the fact that when, when ML can be embedded into the, into the database and it simplifies, uh, that, that all that, uh, uh, uh, complexity, they absolutely love it because they can focus on more important things. >>Clearly I'm this tenant, and this is part of the keynote. They were laying out all their announcements, quick excitement and ML insights out of the box, quick, quick site cue available in preview all the announcements. And then they moved on to the next, the fourth tenant day solving real problems end to end, kind of reminds me of the theme we heard at Dell technology worlds last year end to end it. So we are starting to see the, the, the land grab my opinion, Amazon really going after, beyond I, as in pass, they talked about contact content, contact centers, Kendra, uh, lookout for metrics, and that'll maintain men. Then Matt would came on, talk about all the massive disruption on the, in the industries. And he said, literally machine learning will disrupt every industry. They spent a lot of time on that and they went into the computer vision at the edge, which I'm a big fan of. I just loved that product. Clearly, every innovation, I mean, every vertical Dave is up for grabs. That's the key. Dr. Matt would message. >>Yeah. I mean, I totally agree. I mean, I see that machine intelligence as a top layer of, you know, the S the stack. And as I said, it's going to be infused into all areas. It's not some kind of separate thing, you know, like, Coobernetti's, we think it's some separate thing. It's not, it's going to be embedded everywhere. And I really like Amazon's edge strategy. It's this, you, you are the first to sort of write about it and your keynote preview, Andy Jassy said, we see, we see, we want to bring AWS to the edge. And we see data center as just another edge node. And so what they're doing is they're bringing SDKs. They've got a package of sensors. They're bringing appliances. I've said many, many times the developers are going to be, you know, the linchpin to the edge. And so Amazon is bringing its entire, you know, data plane is control plane, it's API APIs to the edge and giving builders or slash developers, the ability to innovate. And I really liked the strategy versus, Hey, here's a box it's, it's got an x86 processor inside on a, throw it over the edge, give it a cool name that has edge in it. And here you go, >>That sounds call it hyper edge. You know, I mean, the thing that's true is the data aspect at the edge. I mean, everything's got a database data warehouse and data lakes are involved in everything. And then, and some sort of BI or tools to get the data and work with the data or the data analyst, data feeds, machine learning, critical piece to all this, Dave, I mean, this is like databases used to be boring, like boring field. Like, you know, if you were a database, I have a degree in a database design, one of my degrees who do science degrees back then no one really cared. If you were a database person. Now it's like, man data, everything. This is a whole new field. This is an opportunity. But also, I mean, are there enough people out there to do all this? >>Well, it's a great point. And I think this is why Amazon is trying to extract some of the abstract. Some of the complexity I sat in on a private session around databases today and listened to a number of customers. And I will say this, you know, some of it I think was NDA. So I can't, I can't say too much, but I will say this Amazon's philosophy of the database. And you address this in your conversation with Andy Jassy across its entire portfolio is to have really, really fine grain access to the deep level API APIs across all their services. And he said, he said this to you. We don't necessarily want to be the abstraction layer per se, because when the market changes, that's harder for us to change. We want to have that fine-grained access. And so you're seeing that with database, whether it's, you know, no sequel, sequel, you know, the, the Aurora the different flavors of Aurora dynamo, DV, uh, red shift, uh, you know, already S on and on and on. There's just a number of data stores. And you're seeing, for instance, Oracle take a completely different approach. Yes, they have my SQL cause they know got that with the sun acquisition. But, but this is they're really about put, is putting as much capability into a single database as possible. Oh, you only need one database only different philosophy. >>Yeah. And then obviously a health Lake. And then that was pretty much the end of the, the announcements big impact to health care. Again, the theme of horizontal data, vertical specialization with data science and software playing out in real time. >>Yeah. Well, so I have asked this question many times in the cube, when is it that machines will be able to make better diagnoses than doctors and you know, that day is coming. If it's not here, uh, you know, I think helped like is really interesting. I've got an interview later on with one of the practitioners in that space. And so, you know, healthcare is something that is an industry that's ripe for disruption. It really hasn't been disruption disrupted. It's a very high, high risk obviously industry. Uh, but look at healthcare as we all know, it's too expensive. It's too slow. It's too cumbersome. It's too long sometimes to get to a diagnosis or be seen, Amazon's trying to attack with its partners, all of those problems. >>Well, Dave, let's, let's summarize our take on Amazon keynote with machine learning, I'll say pretty historic in the sense that there was so much content in first keynote last year with Andy Jassy, he spent like 75 minutes. He told me on machine learning, they had to kind of create their own category Swami, who we interviewed many times on the cube was awesome. But a lot of still a lot more stuff, more, 215 announcements this year, machine learning more capabilities than ever before. Um, moving faster, solving real problems, targeting the builders, um, fraud platform set of things is the Amazon cadence. What's your analysis of the keynote? >>Well, so I think a couple of things, one is, you know, we've said for a while now that the new innovation cocktail is cloud plus data, plus AI, it's really data machine intelligence or AI applied to that data. And the scale at cloud Amazon Naylor obviously has nailed the cloud infrastructure. It's got the data. That's why database is so important and it's gotta be a leader in machine intelligence. And you're seeing this in the, in the spending data, you know, with our partner ETR, you see that, uh, that AI and ML in terms of spending momentum is, is at the highest or, or at the highest, along with automation, uh, and containers. And so in. Why is that? It's because everybody is trying to infuse AI into their application portfolios. They're trying to automate as much as possible. They're trying to get insights that, that the systems can take action on. >>And, and, and actually it's really augmented intelligence in a big way, but, but really driving insights, speeding that time to insight and Amazon, they have to be a leader there that it's Amazon it's, it's, it's Google, it's the Facebook's, it's obviously Microsoft, you know, IBM's Tron trying to get in there. They were kind of first with, with Watson, but with they're far behind, I think, uh, the, the hyper hyper scale guys. Uh, but, but I guess like the key point is you're going to be buying this. Most companies are going to be buying this, not building it. And that's good news for organizations. >>Yeah. I mean, you get 80% there with the product. Why not go that way? The alternative is try to find some machine learning people to build it. They're hard to find. Um, so the seeing the scale of kind of replicating machine learning expertise with SageMaker, then ultimately into databases and tools, and then ultimately built into applications. I think, you know, this is the thing that I think they, my opinion is that Amazon continues to move up the stack, uh, with their capabilities. And I think machine learning is interesting because it's a whole new set of it's kind of its own little monster building block. That's just not one thing it's going to be super important. I think it's going to have an impact on the startup scene and innovation is going, gonna have an impact on incumbent companies that are currently leaders that are under threat from new entrance entering the business. >>So I think it's going to be a very entrepreneurial opportunity. And I think it's going to be interesting to see is how machine learning plays that role. Is it a defining feature that's core to the intellectual property, or is it enabling new intellectual property? So to me, I just don't see how that's going to fall yet. I would bet that today intellectual property will be built on top of Amazon's machine learning, where the new algorithms and the new things will be built separately. If you compete head to head with that scale, you could be on the wrong side of history. Again, this is a bet that the startups and the venture capitals will have to make is who's going to end up being on the right wave here. Because if you make the wrong design choice, you can have a very complex environment with IOT or whatever your app serving. If you can narrow it down and get a wedge in the marketplace, if you're a company, um, I think that's going to be an advantage. This could be great just to see how the impact of the ecosystem this will be. >>Well, I think something you said just now it gives a clue. You talked about, you know, the, the difficulty of finding the skills. And I think that's a big part of what Amazon and others who were innovating in machine learning are trying to do is the gap between those that are qualified to actually do this stuff. The data scientists, the quality engineers, the data engineers, et cetera. And so companies, you know, the last 10 years went out and tried to hire these people. They couldn't find them, they tried to train them. So it's taking too long. And now that I think they're looking toward machine intelligence to really solve that problem, because that scales, as we, as we know, outsourcing to services companies and just, you know, hardcore heavy lifting, does it doesn't scale that well, >>Well, you know what, give me some machine learning, give it to me faster. I want to take the 80% there and allow us to build certainly on the media cloud and the cube virtual that we're doing. Again, every vertical is going to impact a Dave. Great to see you, uh, great stuff. So far week two. So, you know, we're cube live, we're live covering the keynotes tomorrow. We'll be covering the keynotes for the public sector day. That should be chock-full action. That environment is going to impact the most by COVID a lot of innovation, a lot of coverage. I'm John Ferrari. And with Dave Alante, thanks for watching.

>>from around the globe. It's the Cube with digital coverage of AWS >>reinvent 2020 sponsored >>by Intel, AWS and our community partners. Welcome back to the cubes. Virtual coverage of AWS reinvent 2020. It's virtual this year. We're not in person, so we're doing remote interviews. Part of the three weeks we'll be covering wall to wall a lot of great conversations. News to cover and joining me today Off Fresh off the news off Andy Jackson's keynote, We have two great guests here. Jason Kapoor, senior product manager for Accelerated Computing at A. W S and eight time Medina Chief business officer, Havana Labs, which was recently acquired by Intel Folks. Thanks for coming on, gentlemen. Thank you for spending the time for coming on the key. Appreciate it. >>Thanks for having us. >>J Town. So talk about the news, actually. Uh, computers changing. It's being reinvented. That's the theme from Andy's keynote. What did Andy announced? Could you take a minute to explain the announcement? What services? What ap What's gonna be supported? What's this about? Take a minute to explain. >>Yeah, absolutely. Yeah. So today >>we >>announced our plans to launch and easy to instance based on hardware accelerators from Havana labs. We expect these businesses to be available in the first time from next year. And these air custom designed for accelerating training off deep learning models, a zoo we all know like training of deep learning models is a really competition. Aly extensive task. Oftentimes it takes too long and cost too much. And we're really excited about getting these instances out of the market as we expect for them to provide up to 40% better price performance. Thani on top of the line GPU instances, >>a lot of improvements. Why did anybody do this? Why heaven or what's the what the working backwards document tell you? What is it customers looking for here is or specific use case? >>Yeah, absolutely. So, you know, over the years, uh, the use of machine learning and deep learning has, like, really skyrocketed, right? So we're seeing companies from all the way like 14, 500 to like start ups just reinventing their business models and use using deep learning more pervasively. Right. So we have companies like Pinterest, you know, you'd use deep learning for content recommendations and object detection to Toyota Research Institute that are advancing the science behind autonomous vehicles. And there's a consistent cream from a lot of these customers that are, you know, innovating in the deep learning space that you know the cost it takes to experiment, train and optimize the deep learning models. It's too high. And, you know, they're looking at us as one of their partners to help them optimize their costs, you know, bring them as well as possible while giving them really performing products and enable them to actually bring their markets, their innovations to market as soon as possible. Right? S o. Do you answer your questions straight on your wants? The working backwards. It's a feedback from customers that they want choice on. They want our help Thio lower. Uh, the amount of compute resources and the cost it takes to train the new planning models. >>Hey, Tom, why don't you weigh in here on Havana and now part of intel? What trends are driving this? What's the motivation? Were you guys fit in? What's your view on this? >>Yeah, So Havana was founded in 2016 to deliver a I processors for the data center and cloud for training and inference deep learning models. So while building chips is hard, building, the software and ecosystem is even harder. So joining forces with intel simply helps us connect the dots. Ever since the acquisition last year, we were able to significantly boost our armed. The resource is, and now we're leveraging inter scale in number of customers and ecosystem and partner support. >>So what's the name of the product? Is there a chip name got? Was it Gowdy is the name? >>Yes, the product is man angle. >>Okay. And so it's gonna be hardware. So it's the hardware software. What's involved? Take us through the product. >>Yes. So Gandhi was designed from the ground up to do one task which is training deep learning models. To do that well, we focus the architectural to aspect efficiency and scalability. The computer architectures is a combination of fully programmable TPC tensile process, of course, and a central g M n G. These DPC course are programmable Villa W seen the machines that we designed with custom instruction, set architecture, er and special functions that will developed specifically for a I. The Gandhi cheap integrates also 32 gigabyte off H B M to memory which makes it easy to port to. For GPU developers, Gandhi is unique in integrating 10 parts of 100 gigabit Internet rocky on cheap. And this is opposed to other architectural, which use proprietary interfaces. So overall, improving the cost performance is achieved through efficiency, namely higher utilization off the computer and memory resource is on cheap and the native integration off the rocky interfaces >>J Town. This is actually interesting, as this is the theme for reinvent. We're seeing it right on stage today. Play out again another command performance by Andy Jassy. Slew of announcements. How does Gowdy fit into the AI portfolio or Amazon strategy? Because what a town saying is it sounds like he's doing the heavy lifting on all this training stuff when people want to just get to the outcome. I mean, the theme has been, just let the product do what they do kind of put stuff under the covers and just let it scale. Is that the theme here is this. >>What does this >>all fit in? Take us through how this fits into the A, I strategy for Amazon and also what what what is Havana Intel bring to the table? >>Absolutely. Yeah. So with respect to our overall strategy and portfolio units, it's relatively straightforward, right? So we're laser focused on making sure we have the broadest and deepest portfolio off services for machine learning, right? So these range from infrastructure services specifically compute networking and storage all the way up to, like, managed and all services, which come with pre trained models and customers can simply invoke them using an A P. I call right eso. So from a strategy perspective, you want to make sure that we provide a customer to a choice, uh, enable them to pick the right platform for the right use case, help them get to the Khan structure they actually want, right eso with Havana. And you know, their acquisition with Intel, we finally have access to hardware software and the ability to kind of build out a ecosystem beyond what you know judicially is being used. Which is was a GP used right eso. So the engagement with with Havana, you know, allows us to take their products and capabilities, wrap it around, and easy to instance, which is what customers will be able to launch right on doing so. We're enabling them to tap into the innovation that Teton the rest of the Havana team are working on while having a solution that is integrated with the full AWS stack. Right? So you don't you don't have to rack in stock hard. Bring your data center thes. They're gonna be available standard. Easy to instances. You can just click and launch them. Get access to software that's already pre integrated and big den and ready to go right. Eso so it actually comes down to taking their innovations, coupling it with an AWS solution and making it too easy for customers together. I've been running with the respective training the deepened models. >>Well, here is the question that I want to get to. I think everyone's on everyone's mind is how is it Gowdy different or similar than other GPU? Specifically, you mentioned the software stack on the AWS What you get the software stack inside the chip. How is this different or similar? Two other GP use. And what's the difference between the software stack versus a traditional libraries? >>So from day one, we were focused on the software experience and we were mindful in the need to make it easy for developers to use the innovations we have in the hardware. Most developers, if not all of them, are using deep learning frameworks such as tensile flowing pytorch for building their deep learning models. So God is synapse AI software suite comes integrated and optimized for tensorflow and pilotage, so we expect most developers to be able to take their existing models and with minor changes to the training strips to be able to run them on Gowdy based instances. In addition, expert developers that are familiar with writing their own kernels will be provided with food too sweet for writing their own TPC kernels that can augment the Havana provided library. >>So that's the user experience for the developers, right? That's what you're saying >>exactly, exactly, and we will provide detailed guides for developers. In doing that, Havana will provide open access to documentation library software models and left toe Havana's kita and bi directional communication with the Havana developer community. All these resources will be available concurrently with the AWS Instances launch. >>Okay, so I'm a developer. How did I get involved? It's software on git hub I use the hardware is on Amazon, obviously, in their instances. It's a new instance. Take me through the workflow develop. I'm into this. I wanna I wanna get involved. What I what am I doing? Take me >>through? Yes, I think it s so If the developer is accustomed to using GPS for training the deep learning models three experience is gonna be practically the same, right? So they'll have multiple options to get started. One of them would be, for example, to take our deep learning, Um, it's or Amazon machine images that will come integrated with software from Havana labs. Right. So customers will take the deep Learning Army and launch it on an easy to instance, featuring the gaudy accelerators. Right? So when with that, they'll have, you know, the baseline construct off software and hardware available to get up and running with right, we'll support, you know, all different types of work flows. So if customers want to use containerized solutions, thes instances will be supported R E C s and E s services. Eso using containerized kubernetes you know, thes the solution will just work on. Lastly, we also intend to support these instances through sage maker eso. Just a quick recap on stage maker. That's a manage service that does end to end that provides end to end capabilities for training, debugging, building and deploying machine learning applications. Eso these instances will also be supporting sage maker. So if you're fiddling with sage maker, you can get up and running with this. This is fairly quickly. >>It sounds like it's gonna enable a lot of action and sage maker level. Then can that layer on the use cases? I gotta ask you guys quickly, What's the low hanging fruit use case applications for this product thing? This partnership, Because you know that's gonna be the first Traction said, What are some of these applications gonna be used for? What can we expect to see? >>So typical applications would be image classifications, object detection, natural language processing, the recommendation systems. You'll find reference models in our get up for that and will be growing at least a Z you can imagine. >>Okay, where can people find more info? Give us the data. Take him in to explain. Put a plug in for how What's all the coordinates? U r l sites support how people create, Um, how people get involved. The community. >>Yeah, so customers will be able to access information on AWS websites and also on Havana Labs website. So you will be kicking off a preview early next year. Eso I would highly recommend for customers to find our product pages and signed up for already access and previous information. Utah. >>Yes, and you'll find more information on Havana. A swell a Savannah's get up over time. >>Great announcement. Congratulations. Thanks for sharing the news and some commentary on it. This is really the big theme. You know what Cove in 19 and this pandemic has shown is massive acceleration of digital transformation and having the software and hardware out there that accelerates the heavy lifting and creates value around the data. Super valuable. Thanks for for doing that. Appreciate taking the time. Thank >>you so much. >>Yeah. Thanks for having >>us. Okay, this is the cubes coverage at 80. Best reinvent next three weeks. We're here on the ground. Will remote. We're live inside the studio. We wish we could be there in person, but it's remote this year. But stay tuned. Check out silicon angle dot com. Exclusive interviews with Andy Jassy and Amazon executives and the big news covering. They're all there in one spot. Check it out. We'll be back with more coverage after this break. Thanks for watching. Yeah.