HPE Discover 2024 | Harnessing AWS to build game-changing applications

Clips
News
More from HPE Discover 2024

Art Baudo

Principal Product Manager, Amazon EC2

AWS

AWS Cloud Services: Enabling businesses to mitigate costs and enhance security

As Amazon Web Services Inc. continues to lead the cloud market, the company is continuously looking to innovate the industry to help solve problems across the world while delivering a reliable and safe cloud to its users. While resources are becoming more and more expensive, users are seeing the cloud as the best way to mitigate their workloads without breaking the bank or compromising security, with the help of AWS Cloud services.AWS’s Art Baudo talks with theCUBE about AWS cloud services.“The cloud offers a lot of security as well

play_circle_outline AWS Steps Up Game with U7i High Memory Instances: Meeting Demand for Performance-Driven Apps

play_circle_outline Maximizing Enterprise Value with SAP Workloads and AI: Performance Options for Large Memory Instances

play_circle_outline AI deployments and tools like SageMaker driving innovation

play_circle_outline Game-changing applications in drug discovery, image recognition, and more

play_circle_outline Exploring the Diverse Innovations in Cloud Services and Server Options for Every Business Need

Info

Harnessing AWS to build game-changing applications

Art Baudo

Principal Product Manager, Amazon EC2 AWS

search

>> Welcome back to HPE, discover this conversation. I'm John Furrier, host of theCUBE here in the Palo Alto studio. We're going to have a conversation with AWS harnessing the AWS Cloud to build game changing applications. Art Baudo is here. Welcome back to theCUBE. Good to see you.

Art Baudo

>> It is nice to see you too, John. It's good to be here.

>> So principal product marketing manager at EC2, you're on the front end of all these new applications that are coming online. We're seeing a lot of on-premise two cloud migrations and the talk of every conversation we've been in theCUBE for the past six months to a year has been, "I need more horsepower, where we got more scale on the workloads. Next gen clouds happening, generative AI is happening on the consumer side." You guys are providing a lot of that horsepower. HPE is seeing a lot of action here. Matt Garman was at SAP Sapphire talking about the Amazon products supporting a lot of memory. Take us through what you guys are talking about here with HPE?

Art Baudo

>> Yeah, so I'm really happy to be here to talk to the folks at HPE Discover, and we have a lot of other events happening here as well too, so people can stay tuned, John and attend one of our sessions for more information. But aside from that, to talk about the increased performance, one of the things that Matt launched and talked about last week was our new U7i instances. Those are our high memory instances that in my view have an astonishing 32 terabytes of memory that's attached to them there that's available to customers. So I remember a time, John in my career here that a gigabyte was a lot of memory and so to offer this kind of memory, and it's not just memory, it's additional performance that we're offering as well here with these instances. So these are based off of our custom Sapphire Rapid skew from Intel, and this is the latest generation of those instances. And the highest amount of memory that we have, and in large point part to what we have in our total ecosystem here is we have done, in our memory optimized our general purpose and our compute optimized instances. We also offer this Sapphire rapid skew to folks as well. So if you don't need 32 terabytes of memory and you need something a little smaller, we have everything out there from a few gigabytes all the way up to terabytes as well to suit what customers need at that high performance level. And beyond that, I want to take this opportunity to also mention that in terms of performance, we stay on that cutting edge John of performance. And so while we have Sapphire Rapids on the Intel side, we have launched Genoa on the AMD side, so the latest processor there. And as everyone, you know and I think we have had conversations before about, we've also launched our own silicon purpose-built CPU silicon out there as well to provide that latest performance and folks can actually even check out the R8G instances that we have that are in private preview right now. So you're interested in an ARM-based system and you're ready for some additional performance. That's a great opportunity as well.

>> Yeah, I mean I always love to see the innovation from AWS, especially on the performance side from silicon into the chips that's really going to power the new infrastructure. I got to ask you the drivers behind this because we're seeing these large memory instances from HPE customers. They're hungry for more compute, they want more action, they need more horsepower. And so what's driving these large memory instance, the demand side of it? What are you seeing? What are the game-changing applications? Is it migration to more performance? What are the key drivers?

Art Baudo

>> Yeah, so I want to cover two things. One is referencing back to just a couple weeks ago at SAP Sapphire where our own CEO Matt Garman was there talking about the importance of our relationship that is well working with SAP Sapphire and I believe a lot of customers at HPE Discover are also SAP customers. And SAP databases require a tremendous amount of power and corresponding memory. So those U7i instances that I mentioned are using some of those SAP, are using those SAP workloads and they're designed for those HANA workloads here in the cloud. So that's one of the key areas that I have seen a great deal of growth for customers out there, particularly those that are potentially in the audience. And I'm sure there's this strange thing John out there going on about something called artificial intelligence. I don't know if you've heard that term in the last six months.

>> Yeah, I've heard about generative AI. Is that a new thing?

Art Baudo

>> Yeah, I know exactly. And so there are a lot of customers when people are working on training models or developing their own model or language model here associated with generative AI, they need increased performance and in many cases increased memory. And so we have been offering a lot of different options to customers to be able to get started with that performance as well. Then that ranges from those that are using, we do have some customers, John, that are using CPU to do AI performance. So that ranges from using one of the Intel AMD or Graviton based products all the way to customers that are doing more advanced stuff. And we offer some of the most advanced accelerators in the world with the NVIDIA GPUs and also our own Tranium and Inferentia silicon. As you mentioned before, our strong history here in silicon development and silicon creation, that kind of spans the CPU to those AI, generative AI spaces and also is underpinned by what we call our Nitro system as

>> Well. Yeah, you guys always deliver the goods when it comes to performance and the Amazon U7i looks good for that memory piece because one of the things we've been reporting and want to get your reaction to this is that these game-changing apps run a lot of unstructured data, some semi-structured data and AI needs to eat that up and it's their appetite for compute and systems clusters GPUs, CPUs. Memory is a big part of this new kind of what you call clustered systems. I call it clustered systems, but the demand to run these workloads is not your classic throw some compute at it. It's compute memory and other. So you start to see the confluence of almost the configuration setup up to provision that or to have instances to support those customers and those use cases. Are you seeing the same thing? I mean, what's driving the appetite for this new kind of configuration? Is it the AI stuff or is it more just they need more horsepower?

Art Baudo

>> So I think that in general, there has been a trend in a need for more horsepower period, and I think that that predates any conversation about artificial intelligence. We have been seeing that continued increase. And on the SAP side, we see that with the memory with the increased size of databases and customers of ours who would use, who are interested in that would use those if you will, U7i or our high memory instance families. And we have a lot of offerings in those spaces. And then we have the customers who are using and doing work that are training large language models using everything from Claude 3 to Llama and other things such as that. And those folks are using a lot of the GPU and Tranium and inferential type resources where they're banding together or clustering together in some cases now 10,000 GPUs John, which is to me astronomical, I mean just to do some of their immediate tasks. So the landscape is shifting such that you need more performance, more memory, and more compute. And the cloud environment is such that we are tailoring that to the different offerings that we have and to the different customer needs. You want this for SAP? That's where our U7i instances come in. And if you want something using those GPUs, we have all these accelerators including the H100 and the like and P5 and stuff available for customers to take a look at.

>> All right, that's a great point. I mean basically you're calling out that you guys can give optionality on performance and value based upon whatever they're looking for. Where does that play well for an enterprise customer? Can you give an example? I mean, if you've got the options to kind of play with the knobs, if you will, I want this for that. Where does that happening, in migrations? Is it hybrid cloud? What's the value of having that optionality for performance and value? Where does that fit for enterprise customers that are moving from on-prem to the cloud?

Art Baudo

>> Yeah, so I can give a couple of examples. One, I want to start with one thing that we talked about before, how SAP is actually now starting to run some of their own cloud work and testing work here using Graviton. They provided us some quotes for that previously. But one of the beauties, John, as you know of the cloud has been the ability for you to expand and contract. The cloud is the ultimate place where you don't have to purchase that extra infrastructure that you need for a period of time for a specific task. And if you're doing a training, a large language model, for example, you may require an enormous amount of a CPU or GPU effort in order to do that, or likely CPU and GPU together and memory, but you may do that over a period of time and then not need that for a while and then want to do some retraining and update. The cloud offers customers that opportunity to use the resources when they need it and pay for it only when they need it, as opposed to having to acquire those expensive resources. And at this point in time, those resources have become increasingly expensive and our customers are seeing the cloud as an excellent way to migrate their workloads over and we continue to see people move over to the cloud. Also, the cloud offers a lot of security as well. I mean all of the data that's stored, if for example, in an S3 bucket, our storage vehicle, is completely encrypted. So there is no danger of that content being visible to anyone else, and it's protected and it's accessible as well, which I think is key to our customers accessibility and global accessibility and ease of doing that.

>> SAP is a great example of classic enterprise app now running in the cloud. How do customers move their stuff to the cloud? I mean obviously the trend is there. You got the horsepower, you got the capabilities to tune it and run it effectively and securely. Talk about the ease of use in the migration. You guys been on this journey for a while. It's happening a lot. What's the efficiency plan? How do people do this right?

Art Baudo

>> Yeah. So I would tell customers that almost the majority of time it really is just what we would call lift and shift, John, where you can start a workload that you're using on an x86 based system or an ARM based system or even for that matter, we also offer Mac as well in the cloud. And if you're doing any development on any one of those platforms, you can start by moving those workloads almost immediately and directly to the cloud. And in some cases, maybe you didn't have, you're using x86 and you want to move or transform to ARM. We also offer software and Amis and other things to be able to help you convert and move that over if you have the interest for those that do. And likewise, if you're on the x86 side, it's easy to just take those workloads and move them into the cloud. And to my point about starting earlier about the offering of Sapphire Rapids in Genoa by moving to the cloud, John, you get some of the latest and greatest instances on the x86 side in addition to the performance that we deliver on our Graviton and ARM based systems as well. So we try to make it as simple as possible for customers to move over.

>> The big conversation at HPE is obviously the impact of AI. Amazon, you've been deploying AI for customers. Take us through those deployments. What are some of the needs? What's evolving? What do you have a clear line of sight on when it comes to deploying AI in AWS and what's around the corner?

Art Baudo

>> Well, we've been deploying AI at AWS and leveraging AI for a lot of different functions at Amazon at large for many, many years now already. And I think people have seen that in some of the amazon.com side we're working with things like Alexa, obviously clear implication of doing and listening to natural speech processing to be able to translate that and the number of times you go someplace John and someone asks, "Alexa set a timer," or whatever. That's a form also of artificial intelligence, but also the things that we're doing and the tool sets that we're providing to customers from our SageMaker to Bedrock where we're trying to simplify and integrate that to make it easiest for customers to get started in the cloud using AI. So we're offering tools and we're offering our experience to customers as well. We're trying to do things such as, I was just demoing to customers last week, utilization of code that we generated for them to be able to do things and replicate SQL queries into CloudFront and the like so that we can empower your developers and empower people, engineers, you can use that to simplify people's jobs, that generation of code, that AI that we offer as well. And also for the customers that are going to build and generate their own large language model in the cloud, we also offer the system to do that as well. And we're excited for customers that have migrated those implementations over. And I think folks have heard some of the work that we're doing with companies such as Nvidia launching the Grace Blackwell stuff in the coming years here happening. And that's a sign of the partnerships that we have across the board for artificial intelligence and the work that we continuously do.

>> And again, the title of this talk is Harnessing AWS to build game changing applications. As the product marketing manager at EC2, you're in the middle of the action. The compute is critical. EC2 has been a big part of the power source that drives the infrastructure as a service for AWS. It's involved in all the aspects of what's around it. You mentioned some of the memory instances, which is great. What does the game changing apps look like? How are you looking at that market right now and what's coming online, because we're seeing the same pattern we saw in Cloud 1.0 or first generation cloud of SaaS apps being deployed in the cloud? Now you have more complex systems coming on board because they got to power this new gen AI architecture, which is heavy data, a lot of processing power, but it is not yesterday's processing. It's not yesterday's data processing. It's got a new look to it. It's got neural networks, you've got all kinds of new things. What are some of these game changing applications you seeing coming in from the enterprise?

Art Baudo

>> We're seeing a lot of different game changing applications and some of those applications are things such as we work with companies such as AstraZeneca that is in the cloud using the cloud for drug discovery for example. We've worked with other companies such as artificial intelligence company, Leonardo.ai, that has been processing images, excuse me, in the cloud and that is image recognition that they're doing. So the game-changing applications, I think John vary from industry to industry and company to company. But I think that the salient point is that everyone is using these game-changing applications are helping to simplify and also provide more value for a lot of customers. They're opening doors to let more people, as I mentioned on the coding example as well, allow more people to get involved in the process to reduce, to keep the engineers that you have and your staff focused on what is most important to you and what is delivering that innovation. So we're seeing innovation across the board with a lot of things. I mentioned SAP before, Tufts has done work here and some of our other customers that are out there that have done a lot of discussions and movement into the cloud, there's a lot of that happening as we go forward here.

>> Well, I love how these apps are changing the world. A lot of hard problems being solved and more horsepower drives more AI. Again, the world's changing and the more we get more compute we have the more we can process. So I think we're seeing those solutions that we look back that were either high-performance computing or not gettable, attainable to secure are now in sight. This is the key to the creative Cambrian explosion we're living in.

Art Baudo

>> Yeah, and to your point about the amount of data, we've been collecting data for things for a long period of time, John, about a lot of activities and I work with a lot of customers. Let's take the healthcare example. You've been collecting a lot of data about genome for a period of time, but the challenge has been processing that data and what to do with it and AI and the additional performance and memory that we're able to offer in the cloud helps you be able to create those revolutions. So it's being able to actually process the data and then be able to do something with that information and make intelligent choices. So storing lots of information in an Excel spreadsheet, it makes it really challenging for your staff to be able to decide what to do and decide how to make decisions going forward here. But one of the things that AI in concert with having this additional memory and performance in the cloud allows you to do is help improve the process to do that and make it faster and be able to provide game-changing responses to customers out there.

>> It's interesting Art, you mentioned that I've been talking to a lot of customers around how they're looking at the on-premise future versus cloud. And one CIO said to me, "A lot of people are overthinking their on-premise AI strategy because a lot of it's just simple data processing. You can just throw it in the cloud with EMR and just you're done," right? So the cloud, it can do a lot of that data processing and a lot of this stuff that people are trying to architect on-premise are a little bit overbuilt. And so you start to see the data processing discussion come back front and center where there's clear lines of sight, where it's really advantageous to keep it in the cloud. And that's getting more complex too in the sense of you guys are building security and optionality in there. So very nuanced point, but sometimes people don't even need to do the on-prem for the gen AI, I just do the data processing in the cloud.

Art Baudo

>> So I am a clear cloud advocate-

>> Of course you're biased.

Art Baudo

>> Of course I work for a cloud provider, AWS. But I would also say that cloud can simplify the effort, it can simplify a lot of innovations and it can simplify when an industry or a customer is looking at a startup, do you want to invest in some of that equipment? It's costly sometimes and you may not know that result. The other thing is, is that with regard to the cloud, there's a lot of security that's embedded within. We guarantee here at AWS that nobody that works for Amazon has any access to any customer data that's placed into the cloud. All data at rest is encrypted and we run end-to-end encryption within the cloud as well here. And we've designed that from the bottom up here. And we've even had our architecture reviewed by third parties, John to do that and talk about that last year. We do that and we're committed to transparency for customers so that they understand what that security story is. So we are working on that side to make sure that the cloud is the most secure possible thing. And I think the other area is the tremendous innovations that the cloud offers customers, both from managed services that we can deliver all the way to just the silicon and the hardware that's here and available. You can get anything from a bare metal instance here to a serverless component.

>> And the memory instance that are available is a great example of always having the best choice of horsepower.

Art Baudo

>> That's right.

>> Great to have you on this conversation. We're on HP Discover Coverage. Thanks. Good to see you again. And hey, let's see those game changing applications, keep it coming online. Thanks for doing it.

Art Baudo

>> Absolutely. And it was great to be here. Thank you for inviting me.

>> Okay. I'm John Furrier here at the Palo Alto Studios for HPE Discover 2024. Thanks for watching.

Harnessing AWS to build game-changing applications

search

>> Welcome back to HPE, discover this conversation. I'm John Furrier, host of theCUBE here in the Palo Alto studio. We're going to have a conversation with AWS harnessing the AWS Cloud to build game changing applications. Art Baudo is here. Welcome back to theCUBE. Good to see you.

Art Baudo

>> It is nice to see you too, John. It's good to be here.

>> So principal product marketing manager at EC2, you're on the front end of all these new applications that are coming online. We're seeing a lot of on-premise two cloud migrations and the talk of every conversation we've been in theCUBE for the past six months to a year has been, "I need more horsepower, where we got more scale on the workloads. Next gen clouds happening, generative AI is happening on the consumer side." You guys are providing a lot of that horsepower. HPE is seeing a lot of action here. Matt Garman was at SAP Sapphire talking about the Amazon products supporting a lot of memory. Take us through what you guys are talking about here with HPE?

Art Baudo

>> Yeah, so I'm really happy to be here to talk to the folks at HPE Discover, and we have a lot of other events happening here as well too, so people can stay tuned, John and attend one of our sessions for more information. But aside from that, to talk about the increased performance, one of the things that Matt launched and talked about last week was our new U7i instances. Those are our high memory instances that in my view have an astonishing 32 terabytes of memory that's attached to them there that's available to customers. So I remember a time, John in my career here that a gigabyte was a lot of memory and so to offer this kind of memory, and it's not just memory, it's additional performance that we're offering as well here with these instances. So these are based off of our custom Sapphire Rapid skew from Intel, and this is the latest generation of those instances. And the highest amount of memory that we have, and in large point part to what we have in our total ecosystem here is we have done, in our memory optimized our general purpose and our compute optimized instances. We also offer this Sapphire rapid skew to folks as well. So if you don't need 32 terabytes of memory and you need something a little smaller, we have everything out there from a few gigabytes all the way up to terabytes as well to suit what customers need at that high performance level. And beyond that, I want to take this opportunity to also mention that in terms of performance, we stay on that cutting edge John of performance. And so while we have Sapphire Rapids on the Intel side, we have launched Genoa on the AMD side, so the latest processor there. And as everyone, you know and I think we have had conversations before about, we've also launched our own silicon purpose-built CPU silicon out there as well to provide that latest performance and folks can actually even check out the R8G instances that we have that are in private preview right now. So you're interested in an ARM-based system and you're ready for some additional performance. That's a great opportunity as well.

>> Yeah, I mean I always love to see the innovation from AWS, especially on the performance side from silicon into the chips that's really going to power the new infrastructure. I got to ask you the drivers behind this because we're seeing these large memory instances from HPE customers. They're hungry for more compute, they want more action, they need more horsepower. And so what's driving these large memory instance, the demand side of it? What are you seeing? What are the game-changing applications? Is it migration to more performance? What are the key drivers?

Art Baudo

>> Yeah, so I want to cover two things. One is referencing back to just a couple weeks ago at SAP Sapphire where our own CEO Matt Garman was there talking about the importance of our relationship that is well working with SAP Sapphire and I believe a lot of customers at HPE Discover are also SAP customers. And SAP databases require a tremendous amount of power and corresponding memory. So those U7i instances that I mentioned are using some of those SAP, are using those SAP workloads and they're designed for those HANA workloads here in the cloud. So that's one of the key areas that I have seen a great deal of growth for customers out there, particularly those that are potentially in the audience. And I'm sure there's this strange thing John out there going on about something called artificial intelligence. I don't know if you've heard that term in the last six months.

>> Yeah, I've heard about generative AI. Is that a new thing?

Art Baudo

>> Yeah, I know exactly. And so there are a lot of customers when people are working on training models or developing their own model or language model here associated with generative AI, they need increased performance and in many cases increased memory. And so we have been offering a lot of different options to customers to be able to get started with that performance as well. Then that ranges from those that are using, we do have some customers, John, that are using CPU to do AI performance. So that ranges from using one of the Intel AMD or Graviton based products all the way to customers that are doing more advanced stuff. And we offer some of the most advanced accelerators in the world with the NVIDIA GPUs and also our own Tranium and Inferentia silicon. As you mentioned before, our strong history here in silicon development and silicon creation, that kind of spans the CPU to those AI, generative AI spaces and also is underpinned by what we call our Nitro system as

>> Well. Yeah, you guys always deliver the goods when it comes to performance and the Amazon U7i looks good for that memory piece because one of the things we've been reporting and want to get your reaction to this is that these game-changing apps run a lot of unstructured data, some semi-structured data and AI needs to eat that up and it's their appetite for compute and systems clusters GPUs, CPUs. Memory is a big part of this new kind of what you call clustered systems. I call it clustered systems, but the demand to run these workloads is not your classic throw some compute at it. It's compute memory and other. So you start to see the confluence of almost the configuration setup up to provision that or to have instances to support those customers and those use cases. Are you seeing the same thing? I mean, what's driving the appetite for this new kind of configuration? Is it the AI stuff or is it more just they need more horsepower?

Art Baudo

>> So I think that in general, there has been a trend in a need for more horsepower period, and I think that that predates any conversation about artificial intelligence. We have been seeing that continued increase. And on the SAP side, we see that with the memory with the increased size of databases and customers of ours who would use, who are interested in that would use those if you will, U7i or our high memory instance families. And we have a lot of offerings in those spaces. And then we have the customers who are using and doing work that are training large language models using everything from Claude 3 to Llama and other things such as that. And those folks are using a lot of the GPU and Tranium and inferential type resources where they're banding together or clustering together in some cases now 10,000 GPUs John, which is to me astronomical, I mean just to do some of their immediate tasks. So the landscape is shifting such that you need more performance, more memory, and more compute. And the cloud environment is such that we are tailoring that to the different offerings that we have and to the different customer needs. You want this for SAP? That's where our U7i instances come in. And if you want something using those GPUs, we have all these accelerators including the H100 and the like and P5 and stuff available for customers to take a look at.

>> All right, that's a great point. I mean basically you're calling out that you guys can give optionality on performance and value based upon whatever they're looking for. Where does that play well for an enterprise customer? Can you give an example? I mean, if you've got the options to kind of play with the knobs, if you will, I want this for that. Where does that happening, in migrations? Is it hybrid cloud? What's the value of having that optionality for performance and value? Where does that fit for enterprise customers that are moving from on-prem to the cloud?

Art Baudo

>> Yeah, so I can give a couple of examples. One, I want to start with one thing that we talked about before, how SAP is actually now starting to run some of their own cloud work and testing work here using Graviton. They provided us some quotes for that previously. But one of the beauties, John, as you know of the cloud has been the ability for you to expand and contract. The cloud is the ultimate place where you don't have to purchase that extra infrastructure that you need for a period of time for a specific task. And if you're doing a training, a large language model, for example, you may require an enormous amount of a CPU or GPU effort in order to do that, or likely CPU and GPU together and memory, but you may do that over a period of time and then not need that for a while and then want to do some retraining and update. The cloud offers customers that opportunity to use the resources when they need it and pay for it only when they need it, as opposed to having to acquire those expensive resources. And at this point in time, those resources have become increasingly expensive and our customers are seeing the cloud as an excellent way to migrate their workloads over and we continue to see people move over to the cloud. Also, the cloud offers a lot of security as well. I mean all of the data that's stored, if for example, in an S3 bucket, our storage vehicle, is completely encrypted. So there is no danger of that content being visible to anyone else, and it's protected and it's accessible as well, which I think is key to our customers accessibility and global accessibility and ease of doing that.

>> SAP is a great example of classic enterprise app now running in the cloud. How do customers move their stuff to the cloud? I mean obviously the trend is there. You got the horsepower, you got the capabilities to tune it and run it effectively and securely. Talk about the ease of use in the migration. You guys been on this journey for a while. It's happening a lot. What's the efficiency plan? How do people do this right?

Art Baudo

>> Yeah. So I would tell customers that almost the majority of time it really is just what we would call lift and shift, John, where you can start a workload that you're using on an x86 based system or an ARM based system or even for that matter, we also offer Mac as well in the cloud. And if you're doing any development on any one of those platforms, you can start by moving those workloads almost immediately and directly to the cloud. And in some cases, maybe you didn't have, you're using x86 and you want to move or transform to ARM. We also offer software and Amis and other things to be able to help you convert and move that over if you have the interest for those that do. And likewise, if you're on the x86 side, it's easy to just take those workloads and move them into the cloud. And to my point about starting earlier about the offering of Sapphire Rapids in Genoa by moving to the cloud, John, you get some of the latest and greatest instances on the x86 side in addition to the performance that we deliver on our Graviton and ARM based systems as well. So we try to make it as simple as possible for customers to move over.

>> The big conversation at HPE is obviously the impact of AI. Amazon, you've been deploying AI for customers. Take us through those deployments. What are some of the needs? What's evolving? What do you have a clear line of sight on when it comes to deploying AI in AWS and what's around the corner?

Art Baudo

>> Well, we've been deploying AI at AWS and leveraging AI for a lot of different functions at Amazon at large for many, many years now already. And I think people have seen that in some of the amazon.com side we're working with things like Alexa, obviously clear implication of doing and listening to natural speech processing to be able to translate that and the number of times you go someplace John and someone asks, "Alexa set a timer," or whatever. That's a form also of artificial intelligence, but also the things that we're doing and the tool sets that we're providing to customers from our SageMaker to Bedrock where we're trying to simplify and integrate that to make it easiest for customers to get started in the cloud using AI. So we're offering tools and we're offering our experience to customers as well. We're trying to do things such as, I was just demoing to customers last week, utilization of code that we generated for them to be able to do things and replicate SQL queries into CloudFront and the like so that we can empower your developers and empower people, engineers, you can use that to simplify people's jobs, that generation of code, that AI that we offer as well. And also for the customers that are going to build and generate their own large language model in the cloud, we also offer the system to do that as well. And we're excited for customers that have migrated those implementations over. And I think folks have heard some of the work that we're doing with companies such as Nvidia launching the Grace Blackwell stuff in the coming years here happening. And that's a sign of the partnerships that we have across the board for artificial intelligence and the work that we continuously do.

>> And again, the title of this talk is Harnessing AWS to build game changing applications. As the product marketing manager at EC2, you're in the middle of the action. The compute is critical. EC2 has been a big part of the power source that drives the infrastructure as a service for AWS. It's involved in all the aspects of what's around it. You mentioned some of the memory instances, which is great. What does the game changing apps look like? How are you looking at that market right now and what's coming online, because we're seeing the same pattern we saw in Cloud 1.0 or first generation cloud of SaaS apps being deployed in the cloud? Now you have more complex systems coming on board because they got to power this new gen AI architecture, which is heavy data, a lot of processing power, but it is not yesterday's processing. It's not yesterday's data processing. It's got a new look to it. It's got neural networks, you've got all kinds of new things. What are some of these game changing applications you seeing coming in from the enterprise?

Art Baudo

>> We're seeing a lot of different game changing applications and some of those applications are things such as we work with companies such as AstraZeneca that is in the cloud using the cloud for drug discovery for example. We've worked with other companies such as artificial intelligence company, Leonardo.ai, that has been processing images, excuse me, in the cloud and that is image recognition that they're doing. So the game-changing applications, I think John vary from industry to industry and company to company. But I think that the salient point is that everyone is using these game-changing applications are helping to simplify and also provide more value for a lot of customers. They're opening doors to let more people, as I mentioned on the coding example as well, allow more people to get involved in the process to reduce, to keep the engineers that you have and your staff focused on what is most important to you and what is delivering that innovation. So we're seeing innovation across the board with a lot of things. I mentioned SAP before, Tufts has done work here and some of our other customers that are out there that have done a lot of discussions and movement into the cloud, there's a lot of that happening as we go forward here.

>> Well, I love how these apps are changing the world. A lot of hard problems being solved and more horsepower drives more AI. Again, the world's changing and the more we get more compute we have the more we can process. So I think we're seeing those solutions that we look back that were either high-performance computing or not gettable, attainable to secure are now in sight. This is the key to the creative Cambrian explosion we're living in.

Art Baudo

>> Yeah, and to your point about the amount of data, we've been collecting data for things for a long period of time, John, about a lot of activities and I work with a lot of customers. Let's take the healthcare example. You've been collecting a lot of data about genome for a period of time, but the challenge has been processing that data and what to do with it and AI and the additional performance and memory that we're able to offer in the cloud helps you be able to create those revolutions. So it's being able to actually process the data and then be able to do something with that information and make intelligent choices. So storing lots of information in an Excel spreadsheet, it makes it really challenging for your staff to be able to decide what to do and decide how to make decisions going forward here. But one of the things that AI in concert with having this additional memory and performance in the cloud allows you to do is help improve the process to do that and make it faster and be able to provide game-changing responses to customers out there.

>> It's interesting Art, you mentioned that I've been talking to a lot of customers around how they're looking at the on-premise future versus cloud. And one CIO said to me, "A lot of people are overthinking their on-premise AI strategy because a lot of it's just simple data processing. You can just throw it in the cloud with EMR and just you're done," right? So the cloud, it can do a lot of that data processing and a lot of this stuff that people are trying to architect on-premise are a little bit overbuilt. And so you start to see the data processing discussion come back front and center where there's clear lines of sight, where it's really advantageous to keep it in the cloud. And that's getting more complex too in the sense of you guys are building security and optionality in there. So very nuanced point, but sometimes people don't even need to do the on-prem for the gen AI, I just do the data processing in the cloud.

Art Baudo

>> So I am a clear cloud advocate-

>> Of course you're biased.

Art Baudo

>> Of course I work for a cloud provider, AWS. But I would also say that cloud can simplify the effort, it can simplify a lot of innovations and it can simplify when an industry or a customer is looking at a startup, do you want to invest in some of that equipment? It's costly sometimes and you may not know that result. The other thing is, is that with regard to the cloud, there's a lot of security that's embedded within. We guarantee here at AWS that nobody that works for Amazon has any access to any customer data that's placed into the cloud. All data at rest is encrypted and we run end-to-end encryption within the cloud as well here. And we've designed that from the bottom up here. And we've even had our architecture reviewed by third parties, John to do that and talk about that last year. We do that and we're committed to transparency for customers so that they understand what that security story is. So we are working on that side to make sure that the cloud is the most secure possible thing. And I think the other area is the tremendous innovations that the cloud offers customers, both from managed services that we can deliver all the way to just the silicon and the hardware that's here and available. You can get anything from a bare metal instance here to a serverless component.

>> And the memory instance that are available is a great example of always having the best choice of horsepower.

Art Baudo

>> That's right.

>> Great to have you on this conversation. We're on HP Discover Coverage. Thanks. Good to see you again. And hey, let's see those game changing applications, keep it coming online. Thanks for doing it.

Art Baudo

>> Absolutely. And it was great to be here. Thank you for inviting me.

>> Okay. I'm John Furrier here at the Palo Alto Studios for HPE Discover 2024. Thanks for watching.