Dell Technologies World 2024 | Brian Payne & Varun Chhabra | Dell Technologies

Clips
News
More from Dell Technologies World 2024

Brian Payne

Vice President, Cloud Product Management

Dell Technologies

Varun Chhabra

SVP, ISG Product Marketing

Dell Technologies

Dell’s strategy for revolutionizing data centers with AI integration

The transformation of data centers into artificial intelligence data centers is reshaping the industry, highlighting the importance of innovation, efficiency and sustainability.As businesses strive to keep up with technological advancements, the urgency to adopt AI and modernize their infrastructure has become more pronounced, according to Brian Payne (pictured, right), vice president of cloud product management at Dell Technologies Inc.Dell’s Varun Chhabra and Brian Payne talk with theCUBE about AI data centers.“The amount of focus on innovation around AI has just been mind-blowing and staggering,” he said. “The call to action has been clear, and it’s been exciting

play_circle_outline Strategies for Retrofitting Data Centers: Enhancing Efficiency, Sustainability, and Supporting AI Technology

play_circle_outline AI's role in energy efficiency and automation

play_circle_outline Examples of successful AI applications in farming and large-scale deployments

play_circle_outline Transformative Strategies: Sustainable Data Center Plan and AI Factories Revolutionize Business Infrastructure and Customer Engagement

play_circle_outline Dell's early ESG reporting and ambitious goals for sustainability

Info

Brian Payne & Varun Chhabra | Dell Technologies

Brian Payne

Vice President, Cloud Product Management Dell Technologies

Varun Chhabra

SVP, ISG Product Marketing Dell Technologies

search

>> Good afternoon, Dell fans, and welcome back to fabulous Las Vegas, Nevada. It is day three of Dell Tech World. I'm feeling super smart, thanks to all the wonderful guests we've had on the show this week. My name's Savannah Peterson, joined by my fabulous co-host, John Furrier. John, how are you feeling this morning? Feeling inspired?

>> I feel great. I'm totally inspired by the hardware revolution. It's a renaissance in systems engineering. It's a renaissance in application transformation. And the clustered systems, the data center is back, but it's not the old data center, it's the new data center. It's the AI data center. So the AI factory, which you know I love. Fan of the Dell's positioning.

>> Yeah, it's right behind us.

>> We're going to talk about it here. Yeah, we're going to talk about it this segment. How to build a modern, sustainable data center.

>> And do it thinking about efficiency and sustainability, which is super awesome. In this moment, let me welcome two super-alums to theCUBE, Brian Payne and Varun Chopra. Thank you so much for being here.

>> Thanks for having us.

>> We had such a fun chat yesterday.

>> Absolutely.

>> I love that you're back on for day two.

>> Yeah. I love it. Back for a second dose.

>> Yeah. You brought the energy yesterday so I know you'll bring... Brian, you're an OG. 12 years of showing up on theCUBE.

Brian Payne

>> That's right. That's right. I love it.

>> That is amazing. You haven't gotten sick of us yet. I'm super grateful.

Brian Payne

>> Not a bit. Not a bit.

>> Our AI is going to speak Brian Payne soon.

Brian Payne

>> There you go.

>> We've got all your linguistics ingested.

>> Thank you for training all of our models with your insights.

Brian Payne

>> I don't even have to show up next time. It'll just be a hologram. It's a large language model.

>> I love it. Brian, since I didn't have you on yesterday, how's the show been for you so far?

Brian Payne

>> It's been great. I'll tell you what, the amount of focus on innovation around AI has just been mind-blowing and staggering. And our customers have been super excited about it. And of course, to some degree, I think feeling a little bit paranoid, like we've got to get moving. The call to action has been clear, and it's been exciting.

>> Yeah.

>> I think that's actually a really good point. You get to talk to a lot of different customers. What is that energy? We've heard the words overwhelmed. There's kind of a FOMO. I don't want to get left behind. What are you hearing, Varun?

>> It's definitely very top of mind in a way that it hasn't been... it hasn't been building towards it, but I think with AI, it's just taken a step change in function in terms of the urgency of the conversation. You saw on stage, Arthur Lewis, he announced some of the racks that we're going to be making available to customers. 70 kilowatts, 100 kilowatts, 130 kilowatts. These are very, very energy-dense and energy-hungry devices, so that's definitely top of mind for customers as well. And what we're talking to them about is, you've got to think from the bottoms up. Right?

>> Yeah.

>> You've got to think about how you're... if you're building a new data center, which we think a lot of people will, like Jensen talked about, building AI factories. Whether you're retrofitting your existing data center, you've got to think through the entire stack. What's the position that your hardware vendor is taking on? How these are built, how these are designed. What's the software layer in terms of being able to manage these? And what are the services and professional services that you're getting to be able to actually think about this from the ground up? And I think that's a place where we feel like we have capabilities in all of those three areas.

>> Guys, I want to get your perspective because we've had many conversations on theCUBE. And as we've been getting closer to this new world we're going to be moving into quickly, the world's changed. And I want to get your thoughts. Building a motherboard, building a server, putting a rack and stack in a data center has changed. Now the data center of the future is going to look a lot different. You can't just put a bunch of GPUs on the rack. You're going to have power problems.

>> Right.

>> So constraints change but the game is still the same.

>> Yeah.

>> You're building a system, right?

>> Yep.

>> So the operating system is the data center.

>> Yep.

Brian Payne

>> Right.

>> And I think GTC. Jensen pointed that out. This is what we think it looks like. It's got leaves, it's got spines, it's got clustered systems.

>> Supercomputer, if you will.

>> Supercomputer, and it's now a new system. So server, data center all kind of blending in my mind. Okay. That rant's over. But what does it mean from a constraint standpoint? Because sustainability is about energy.

>> Yeah.

>> How do you manage that? How do you design that? Take us through your vision of how that plays in, because it's not as easy as just the old way, rack and stack. What are the constraints? And how do you design it around them or manage them?

Brian Payne

>> Yeah. Just building on what Varun said, we have a sustainable data center strategy. And our customers are engaging us, saying, "What do we do? How can you help us?" And so, to your point, we've got to think about, how do we cope with 8X growth in power over the next six years, all driven by gen AI infrastructure? Well, the good news is that we have a legacy of designing systems for pushing the envelope. If you think back to the advent of the hyperscale public cloud, we were very much involved in building out those data centers, which were pushing the envelope at that point in time. So we've got a legacy of expertise around, how do we design at a rack level? How do we make sure that the design within the system is as efficient as possible so that we can think about optimizing the power to cool the entire rack, the entire data center? Liquid cooling innovation that we're bringing in. That's absolutely necessary because, at the end of the day, what's good for sustainability is also good for business because it's going to help people, A, cope with the challenges of taking advantage of this infrastructure and then running it in an efficient way that is cost-effective.

>> We don't always have that intersection.

Brian Payne

>> No.

>> And it's refreshing.

Brian Payne

>> Yeah.

>> And I think the economics of power constraints, the realities of power constraints, the reality of economic considerations, they're going to drive a natural selection process here for customers. And it's not just the power efficiency that Brian was talking about. Other angles that people ask us questions around from sustainability is, what's the use of recyclable materials? What are you doing for circular economy? Energy efficiency is a big economic consideration but also, as you all know, ESG requirements, regulatory compliance around the circular economy, what percentage of recyclable materials you're using. Before any of this AI stuff even kicked off, Dell a leader in that space. Right?

>> Yeah.

>> So we feel like, across the dimension of energy efficiency and the circular economy, we have a great story across the board from not just the design of the individual components and the servers and the storage, but putting it all together in a rack-scale solution for customers.

>> Yeah. You guys have been a leader in that from day one, even from... Boxes originally started that way, and then changed all the material management kicks in.

>> Yes.

>> A lot of people might not know how much you guys do in Dell, especially around software. This has been tooling out there from desktop days and servers.

>> Yeah, yeah, yeah.

>> How much is software involved? Because software plays a key role in managing the system. Obviously power with ethernet. We love how Ethernet, the routing is going.

>> Absolutely. Absolutely.

>> But outside of that, what software? How much software is built in here? What's going on in Dell? What's behind the curtain?

Brian Payne

>> Yeah, sure. It's critical. We can design our products with sustainable materials, make them energy efficient, but if you don't make the operation... give customers the tools to optimize the operation, you're failing. And so we have a stacked approach. So down at the system level we have telemetry built into our system, where we're constantly monitoring how the system's performing, how the workload is optimized, and then controlling cooling within the server. We have a tool called OpenManage Enterprise. It gives customers the ability to manage their fleet of servers, set power constraints and limits on, how do they get the most operational efficiency, identify zombie servers, things like that? And then we pull it up, and we just have released our Apex AIOps observability capabilities, where now we can apply AI to help customers set policies and automate the operation of their data center so that they can manage the most efficient and most effective way.

>> So like the security conversation, where there's AI for security and security for AI, you guys have AI for Dell and Dell for AI. So share the difference between how you're using AI internally or see that playing out to make you more efficient, and then how the customer has it on their end.

>> I think AI for energy efficiency is a fascinating frontier. If you look at what we have announced, one of the things we announced was Apex AIOps. We talked about it yesterday here. And one of the capabilities that we see ourselves, we are investing more and more in that is, as you start to use AI to actually drive the insights that are being fed to the IT operators, energy efficiency is a space that is ripe for AI-driven insights. Brian talked about it at the component level. At the server level, monitoring what's happening with power management, power consumption, zombie servers. John, you were mentioning, as we start to think about this at a system level design and you start to think about a rack scale across the data center, the individual component capabilities are important but they're not as important as the thinking about it as a system. Right?

>> Yeah.

>> Where do you reduce power consumption? Where do you increase power consumption? Where do you think you're going to need more capabilities from a cooling perspective, et cetera? That has to be done at a system level. And AI is going to help reduce that complexity for our customers and give them those insights in a chatbot LLM interface so that they can take action on it relatively quickly.

>> It's holistic health of their data system and-

>> As a system, right. As a whole system. Not just as a server or a storage. Yeah.

Brian Payne

>> And we can even take it proactively where we can actually forecast, what is the product carbon footprint of your data center expected to be? By looking at those historical trends.

>> Yes, yes, yes.

Brian Payne

>> Which are important tools that people are trying to tackle, sustainability?

>> And also you have the ability too to use agents, are coming online, too. How do you guys see the world of agents? Because chatbots will turn into agents, where agents are actually doing the work. Where do you guys see that intersection? Because it's been a big conversation. I don't think there's really a right answer, but it depends on the workflow. I'm just curious. Where is the innovation going to be automated if I have agents working on my behalf of my customer, or is that not-

>> Yeah, it's going to be interesting. I don't know that any of us knows where that's going to go. We're very much at the early stage of that. But if you take the example I was just sharing around AIOps, today what we offer is information for IT operators and app operators to be able to take action. To your point, John, maybe what the next frontier is, instead of the human having to take action, you can automate that so an agent that's servicing the insights is talking to an agent that's actually taking the action. Yeah. I certainly think we're going to start to see that happening more and more, discrete agents talking to each other to actually drive automation.

>> I think it's going to be interesting.

>> Yeah, it's-

>> I know. I think you're right. You have some examples of success in this recently.

>> Yeah.

>> Why don't you talk to me about something that happened on the farm?

>> Oh, sure, I can do Nature Fresh . We have a really strong partnership with an organization called Nature Fresh Farms. They're here on the booth.

>> Oh, cool.

>> And they really are leaders in applying technology for farming and food produce. And we've had a long-standing partnership, many, many years with them in terms of helping them on so many factors that we've been talking about. One is how to be more energy efficient. How to actually get insights to drive more produce per acre, which then helps create more food. And then also help them reduce their... for example, make their environmental footprint more sustainable. We've helped them reduce their water usage by over 90% over the time that we've been working with them. So we love to talk about servers and how much power they're consuming. There is a whole other world out there in terms of water usage, et cetera, that we can help our customers with. I think our partnership with Nature Fresh Farms has been an amazing example of that. And then on the other end of the spectrum, Brian, maybe you can talk about large-scale deployments as well.

Brian Payne

>> Yeah. So as you think about our industry, Dell, one of the big things is bringing together a technology ecosystem. And you think about the challenges with generative AI and the power that's to power large farms of GPU as a service. There's important things to tackle in terms of the energy grid. So we've got one of our new partners here, a customer, IRIN, who is actually developing our data centers that are hosting massive amounts of XE9680s clustered together, all powered by renewable energy, either water-driven or air-driven. They're air-cooled designs-

>> Amazing.

Brian Payne

>> Yeah.

>> That's awesome.

Brian Payne

>> Because they want to optimize water usage, they've chosen the air-cooled designs. And what they've found is, with our products, they can operate their temperature of their data center at a higher level than they could with other designs out there which, again, lowered the overall operating cost of their environment. So as we think about customers who are looking and struggling with, hey, how am I going to get started up with providing generative AI capabilities within my company? Building out the infrastructure's a challenge. That's where this partner ecosystem comes into play. We can partner them up with someone who's doing it in an incredibly sustainable way and get them on the way to solving problems.

>> We talked about this yesterday. Arthur talked about it on stage. The new liquid cool system that's going to house 72 GPUs in one rack that Jensen was so excited about when we were talking to Michael.

>> Yeah. That was a nerdy moment.

>> It is liquid-cooled. Direct to chip liquid cooling. We think it's going to be 2.5 times more energy efficient than what we had in previous generations. So yes, these are large stamps in terms of power consumption, but the energy efficiency per operation is better than ever before. So the amount of processing they're able to drive per watt, we're looking to move that further and further away and be more efficient there.

>> Yeah. And these clusters are getting bigger and bigger, consuming more power. The cooling's critical. The on-prem dynamic is also interesting. I want to get your thoughts on this because you saw this a lot in the HPC workloads. HPC workloads, as we covered I think with you guys at your Innovation Center last year, they were old use cases. All the build the wing at Boeing and all kinds of biology, high-end stuff.

Brian Payne

>> Yeah.

>> Not common use cases. So that's coming down. The democratization of HPC is happening. You're starting to see workloads being scoped, known workloads. A bank application, for instance, might not be an HPC workload. But you inject gen AI into it, that's now an end-to-end workload from the device to the back end. So you can scope that.

>> Before we started, we were talking about SuperCompute, the amazing event that's traditionally been an HPC event. And we're seeing basically the lines between HPC and AI being blurred right now.

>> Yeah.

>> SuperCompute and other such events are now as much AI events as they are HPC events.

>> Yeah.

>> Totally agree. That is interesting.

>> And it'll be very interesting to see how some of these traditional HPC deployments evolve based on the innovation that's happening in what we would call AI workloads that may be discrete from HPC. We definitely think there's going to be a lot of interplay. And of course a lot of the work that's happening with AI is built on the shoulders of the innovation that's been happening in HPC for such a long time.

>> And healthcare. We just had a healthcare, medicine from Northwestern. Their specific, direct quote here on theCUBE: "Our data's on-prem. We want the AI factory here."

Brian Payne

>> Yeah.

>> Yeah.

>> And they've already made advances with your Innovation Lab, HPC Innovation Lab.

Brian Payne

>> Yeah.

>> And again, that points to what I'm seeing. And would love to get your reaction. As an end-to-end workload that was traditionally managed by, IT, has now new requirements for gen AI, which is complicated because it's generative, right?

>> Yeah.

>> It's a different category, as Jensen points out. That now can be scoped. I can now deploy GPUs in a trusted environment and say, "That's scoped." And it's no over-provisioning. It's basically a design.

>> Yeah.

>> That's a use case that I don't need the cloud for. I can say I could host that at a colo facility, I could locate it in the hospital itself.

>> Yeah. Go ahead.

Brian Payne

>> Oh, sorry. I was just going to say, as we think about the democratization of the technology and needing to get it closer to the data, you quickly run into a challenge of, how do I deploy that? How do I retrofit my data center to support that? That's one area where customers are looking to us for help. And we've developed services to help them with, hey, what is a sustainable data center plan? How do I get a multi-year plan to meet the needs of my business, respond to data gravity, and do it in a sustainable way? And so engaging customers on both, on the generative AI front, of, hey, how do I develop applications to solve those problems? And then how do I get the infrastructure that can help them?

>> And again, it's not to make it sound like it's true because it's not. Having a data modernization service like you guys have, a modern data center, sustainable data center service. It's great. Sounds sexy top line. Great. But there's a lot of blocking and tackling you've got going on.

Brian Payne

>> Yeah.

>> You've got asset management, waste recovery.

Brian Payne

>> Yes.

>> So what are those details? That's the heavy-duty stuff.

>> Take a step back. We talked about the AI factory. Like you were talking about, AI factory, it follows the data.

>> Yeah.

>> We believe AI factories will be everywhere. They'll be in data centers, they'll be at the edge, they'll be on users' desktops. Right?

>> Yeah.

>> Services is a huge part of that conversation. And we're being asked, because of all the experience we've had with AI even before gen AI was a thing, thousands of customers. We think we've got a lot of unique IP that we can help customers with from a process perspective. So asset recovery services. If you think about, hey, how do you responsibly recycle your e-waste? Make sure that that helps you with the recyclable, the circular economy. Data center and modernization services. We started this conversation talking about, you've got to think from the ground up in terms of how you've got to retrofit your data centers or build them from scratch. We have a lot of experience we can help customers with, and they're looking to us for that. And then even things like advisory services for building sustainable data centers. Right?

>> Yeah.

>> All of those capabilities we think are an important part of the help we can provide customers and help us be trusted advisor to them on all of these initiatives.

>> This is one of those cases where you become what you're known for. You've done it for yourself.

>> Yeah.

>> It's not something that's, "Hey, let's start a service and start servicing people."

>> Yeah.

>> Well, you've got the capabilities. Why not bring it out to the customers? That's

>> Customers need that help from us.

>> Yeah. Yeah, that's a great win.

Brian Payne

>> In fact, our own Dell Digital team used that service to plan our generative AI capabilities.

>> I was going to ask that. And has the demand for sustainable data centers gone up since we've seen this AI revolution like crazy?

>> Yeah. I guess-

>> I can imagine it wasn't necessarily as popular of a conversation.

>> I think it has been building for a while, right?

Brian Payne

>> Yeah.

>> Everything that's been happening in the industry, even before AI, ESG initiatives, sustainable data centers, et cetera were really driving a lot of those conversations. We certainly think there is a lot more urgency to it because now it's a roadblock to, if you don't solve it, it's a bottleneck to getting your AI efforts. So you're right, yeah.

>> Much more front of mind in the sense of, like, "Ooh, we've got to do this right now or else we're in a bad spot."

>> And it's easier to measure the ROI. It's easier to measure the impact of the business if you don't get those things figured out now.

>> Well, I think AI is an obvious. Jumps off the page from a cost perspective and from a, "Sustainable." It blows all the metrics out of the water. Wait a minute, all this horsepower. That's compute. Power's being used.

>> Yeah.

>> And OpenAI, they're case in point. They've been a big poster child. It costs a lot to run these generative AI workloads from terms of a power.

>> Yeah. Absolutely.

>> So it jumps off the page, "Hey we're looking, doing great." Boom.

>> Yeah.

>> So it's like you've got-

>> Yeah, exactly.

>> You've got to address it.

>> Yeah.

>> Yeah. And like we were talking about earlier with this faster development cycle, machines being able to be made faster, two to three years, traditional cycle now nine months. People craving AI computers. You guys announcing them, which is very exciting. There's going to be a lot more e-waste in theory, depending. And I love that you guys, you have your Concept Luna project, which is really cool. I don't know if you know about this, John, but when I was doing my research for you guys, I thought this was so interesting. It reduces the disassembly time, which is not something I thought about in terms of sustainability, from an hour down to minutes.

>> That's not obvious.

>> No, which is super unique when thinking about this. Coming back to this holistic ecosystem health, what a thoughtful thing to be thinking about. And there are 57 million tons of e-waste around the world every year.

>> Yeah.

>> That's every year

>> E-waste, who thinks about that? Well, you guys do. I don't.

>> They do. That's what I mean. This is about sustainability.

>> Yeah, it really is. And to your point, it's only going to get more magnified.

Brian Payne

>> Yeah.

>> Right.

Brian Payne

>> I think in order to address these really systemic challenges, we take an end-to-end approach as a company. And so it's literally-

>> Very clear.

Brian Payne

>> Ingrained into how we think about engaging with the supply chain. How do we think about the design of our products? To your point about disassembly time. And then on the back end, how do we do recycling, repurposing? And it's just built into the way we operate.

>> Yeah.

Brian Payne

>> And I think that enables us to be successful. But we still have big challenges ahead to get to our stated goals for 2030.

>> I know. It's ambitious goals you've laid out. I was very impressed. But Dell's been putting out an ESG report since 1998, which is pretty early, especially for large company, so I am not surprised. All right, so we let Varun do something special yesterday and I want to let you have your shot, Brian. Since we're talking about a sustainable world, it's all about not just what we do in our day jobs but also making the planet better for our friends and our families and our lives. Is there anyone you'd like to say hello to?

Brian Payne

>> I think I should say hello to my wife Julie and to my two kids, who we're trying to make a more sustainable world for, Charlotte and Claire.

>> Awesome.

>> I love that.

>> Brian's doing a good job. He's working.

>> So what's the conversation like with the fam around the dinner table about AI? Talk to them about it?

Brian Payne

>> Yeah. Our kids are different ages, so we'll have different perspectives. I have a 19-year-old that's a freshman in college, and she's just chomping at the bit to learn everything she can. Figure out how to code and how to apply it. My 17-year-old is also looking into it, figuring out how it's going to do her homework.

>> Yes. I know.

Brian Payne

>> She's lucky.

>> It's a cheeky assistant. That's a cheeky-

Brian Payne

>> The reasoning is really good in the math equations, too.

>> Yeah. Yeah. AI is our little cheeky assistant. Brian, Varun, this has been a really great session.

Brian Payne

>> Thank you, Savannah.

>> This, we went longer than expected. We're actually over time because you're all so interesting. So thank you for bringing this very important conversation. It's been fantastic both times, too. We're really getting to hang out this week. I love it.

>> Likewise.

>> And John, thank you for your fabulous insights you candor, as always.

>> Thank you. Thank you.

>> And thank all of you for tuning in to our fantastic three days of live coverage here in Las Vegas, Nevada at Dell Tech World. My name's Savannah Peterson. You're watching theCUBE, the leading source for enterprise tech news.

Brian Payne & Varun Chhabra | Dell Technologies

search

>> Good afternoon, Dell fans, and welcome back to fabulous Las Vegas, Nevada. It is day three of Dell Tech World. I'm feeling super smart, thanks to all the wonderful guests we've had on the show this week. My name's Savannah Peterson, joined by my fabulous co-host, John Furrier. John, how are you feeling this morning? Feeling inspired?

>> I feel great. I'm totally inspired by the hardware revolution. It's a renaissance in systems engineering. It's a renaissance in application transformation. And the clustered systems, the data center is back, but it's not the old data center, it's the new data center. It's the AI data center. So the AI factory, which you know I love. Fan of the Dell's positioning.

>> Yeah, it's right behind us.

>> We're going to talk about it here. Yeah, we're going to talk about it this segment. How to build a modern, sustainable data center.

>> And do it thinking about efficiency and sustainability, which is super awesome. In this moment, let me welcome two super-alums to theCUBE, Brian Payne and Varun Chopra. Thank you so much for being here.

>> Thanks for having us.

>> We had such a fun chat yesterday.

>> Absolutely.

>> I love that you're back on for day two.

>> Yeah. I love it. Back for a second dose.

>> Yeah. You brought the energy yesterday so I know you'll bring... Brian, you're an OG. 12 years of showing up on theCUBE.

Brian Payne

>> That's right. That's right. I love it.

>> That is amazing. You haven't gotten sick of us yet. I'm super grateful.

Brian Payne

>> Not a bit. Not a bit.

>> Our AI is going to speak Brian Payne soon.

Brian Payne

>> There you go.

>> We've got all your linguistics ingested.

>> Thank you for training all of our models with your insights.

Brian Payne

>> I don't even have to show up next time. It'll just be a hologram. It's a large language model.

>> I love it. Brian, since I didn't have you on yesterday, how's the show been for you so far?

Brian Payne

>> It's been great. I'll tell you what, the amount of focus on innovation around AI has just been mind-blowing and staggering. And our customers have been super excited about it. And of course, to some degree, I think feeling a little bit paranoid, like we've got to get moving. The call to action has been clear, and it's been exciting.

>> Yeah.

>> I think that's actually a really good point. You get to talk to a lot of different customers. What is that energy? We've heard the words overwhelmed. There's kind of a FOMO. I don't want to get left behind. What are you hearing, Varun?

>> It's definitely very top of mind in a way that it hasn't been... it hasn't been building towards it, but I think with AI, it's just taken a step change in function in terms of the urgency of the conversation. You saw on stage, Arthur Lewis, he announced some of the racks that we're going to be making available to customers. 70 kilowatts, 100 kilowatts, 130 kilowatts. These are very, very energy-dense and energy-hungry devices, so that's definitely top of mind for customers as well. And what we're talking to them about is, you've got to think from the bottoms up. Right?

>> Yeah.

>> You've got to think about how you're... if you're building a new data center, which we think a lot of people will, like Jensen talked about, building AI factories. Whether you're retrofitting your existing data center, you've got to think through the entire stack. What's the position that your hardware vendor is taking on? How these are built, how these are designed. What's the software layer in terms of being able to manage these? And what are the services and professional services that you're getting to be able to actually think about this from the ground up? And I think that's a place where we feel like we have capabilities in all of those three areas.

>> Guys, I want to get your perspective because we've had many conversations on theCUBE. And as we've been getting closer to this new world we're going to be moving into quickly, the world's changed. And I want to get your thoughts. Building a motherboard, building a server, putting a rack and stack in a data center has changed. Now the data center of the future is going to look a lot different. You can't just put a bunch of GPUs on the rack. You're going to have power problems.

>> Right.

>> So constraints change but the game is still the same.

>> Yeah.

>> You're building a system, right?

>> Yep.

>> So the operating system is the data center.

>> Yep.

Brian Payne

>> Right.

>> And I think GTC. Jensen pointed that out. This is what we think it looks like. It's got leaves, it's got spines, it's got clustered systems.

>> Supercomputer, if you will.

>> Supercomputer, and it's now a new system. So server, data center all kind of blending in my mind. Okay. That rant's over. But what does it mean from a constraint standpoint? Because sustainability is about energy.

>> Yeah.

>> How do you manage that? How do you design that? Take us through your vision of how that plays in, because it's not as easy as just the old way, rack and stack. What are the constraints? And how do you design it around them or manage them?

Brian Payne

>> Yeah. Just building on what Varun said, we have a sustainable data center strategy. And our customers are engaging us, saying, "What do we do? How can you help us?" And so, to your point, we've got to think about, how do we cope with 8X growth in power over the next six years, all driven by gen AI infrastructure? Well, the good news is that we have a legacy of designing systems for pushing the envelope. If you think back to the advent of the hyperscale public cloud, we were very much involved in building out those data centers, which were pushing the envelope at that point in time. So we've got a legacy of expertise around, how do we design at a rack level? How do we make sure that the design within the system is as efficient as possible so that we can think about optimizing the power to cool the entire rack, the entire data center? Liquid cooling innovation that we're bringing in. That's absolutely necessary because, at the end of the day, what's good for sustainability is also good for business because it's going to help people, A, cope with the challenges of taking advantage of this infrastructure and then running it in an efficient way that is cost-effective.

>> We don't always have that intersection.

Brian Payne

>> No.

>> And it's refreshing.

Brian Payne

>> Yeah.

>> And I think the economics of power constraints, the realities of power constraints, the reality of economic considerations, they're going to drive a natural selection process here for customers. And it's not just the power efficiency that Brian was talking about. Other angles that people ask us questions around from sustainability is, what's the use of recyclable materials? What are you doing for circular economy? Energy efficiency is a big economic consideration but also, as you all know, ESG requirements, regulatory compliance around the circular economy, what percentage of recyclable materials you're using. Before any of this AI stuff even kicked off, Dell a leader in that space. Right?

>> Yeah.

>> So we feel like, across the dimension of energy efficiency and the circular economy, we have a great story across the board from not just the design of the individual components and the servers and the storage, but putting it all together in a rack-scale solution for customers.

>> Yeah. You guys have been a leader in that from day one, even from... Boxes originally started that way, and then changed all the material management kicks in.

>> Yes.

>> A lot of people might not know how much you guys do in Dell, especially around software. This has been tooling out there from desktop days and servers.

>> Yeah, yeah, yeah.

>> How much is software involved? Because software plays a key role in managing the system. Obviously power with ethernet. We love how Ethernet, the routing is going.

>> Absolutely. Absolutely.

>> But outside of that, what software? How much software is built in here? What's going on in Dell? What's behind the curtain?

Brian Payne

>> Yeah, sure. It's critical. We can design our products with sustainable materials, make them energy efficient, but if you don't make the operation... give customers the tools to optimize the operation, you're failing. And so we have a stacked approach. So down at the system level we have telemetry built into our system, where we're constantly monitoring how the system's performing, how the workload is optimized, and then controlling cooling within the server. We have a tool called OpenManage Enterprise. It gives customers the ability to manage their fleet of servers, set power constraints and limits on, how do they get the most operational efficiency, identify zombie servers, things like that? And then we pull it up, and we just have released our Apex AIOps observability capabilities, where now we can apply AI to help customers set policies and automate the operation of their data center so that they can manage the most efficient and most effective way.

>> So like the security conversation, where there's AI for security and security for AI, you guys have AI for Dell and Dell for AI. So share the difference between how you're using AI internally or see that playing out to make you more efficient, and then how the customer has it on their end.

>> I think AI for energy efficiency is a fascinating frontier. If you look at what we have announced, one of the things we announced was Apex AIOps. We talked about it yesterday here. And one of the capabilities that we see ourselves, we are investing more and more in that is, as you start to use AI to actually drive the insights that are being fed to the IT operators, energy efficiency is a space that is ripe for AI-driven insights. Brian talked about it at the component level. At the server level, monitoring what's happening with power management, power consumption, zombie servers. John, you were mentioning, as we start to think about this at a system level design and you start to think about a rack scale across the data center, the individual component capabilities are important but they're not as important as the thinking about it as a system. Right?

>> Yeah.

>> Where do you reduce power consumption? Where do you increase power consumption? Where do you think you're going to need more capabilities from a cooling perspective, et cetera? That has to be done at a system level. And AI is going to help reduce that complexity for our customers and give them those insights in a chatbot LLM interface so that they can take action on it relatively quickly.

>> It's holistic health of their data system and-

>> As a system, right. As a whole system. Not just as a server or a storage. Yeah.

Brian Payne

>> And we can even take it proactively where we can actually forecast, what is the product carbon footprint of your data center expected to be? By looking at those historical trends.

>> Yes, yes, yes.

Brian Payne

>> Which are important tools that people are trying to tackle, sustainability?

>> And also you have the ability too to use agents, are coming online, too. How do you guys see the world of agents? Because chatbots will turn into agents, where agents are actually doing the work. Where do you guys see that intersection? Because it's been a big conversation. I don't think there's really a right answer, but it depends on the workflow. I'm just curious. Where is the innovation going to be automated if I have agents working on my behalf of my customer, or is that not-

>> Yeah, it's going to be interesting. I don't know that any of us knows where that's going to go. We're very much at the early stage of that. But if you take the example I was just sharing around AIOps, today what we offer is information for IT operators and app operators to be able to take action. To your point, John, maybe what the next frontier is, instead of the human having to take action, you can automate that so an agent that's servicing the insights is talking to an agent that's actually taking the action. Yeah. I certainly think we're going to start to see that happening more and more, discrete agents talking to each other to actually drive automation.

>> I think it's going to be interesting.

>> Yeah, it's-

>> I know. I think you're right. You have some examples of success in this recently.

>> Yeah.

>> Why don't you talk to me about something that happened on the farm?

>> Oh, sure, I can do Nature Fresh . We have a really strong partnership with an organization called Nature Fresh Farms. They're here on the booth.

>> Oh, cool.

>> And they really are leaders in applying technology for farming and food produce. And we've had a long-standing partnership, many, many years with them in terms of helping them on so many factors that we've been talking about. One is how to be more energy efficient. How to actually get insights to drive more produce per acre, which then helps create more food. And then also help them reduce their... for example, make their environmental footprint more sustainable. We've helped them reduce their water usage by over 90% over the time that we've been working with them. So we love to talk about servers and how much power they're consuming. There is a whole other world out there in terms of water usage, et cetera, that we can help our customers with. I think our partnership with Nature Fresh Farms has been an amazing example of that. And then on the other end of the spectrum, Brian, maybe you can talk about large-scale deployments as well.

Brian Payne

>> Yeah. So as you think about our industry, Dell, one of the big things is bringing together a technology ecosystem. And you think about the challenges with generative AI and the power that's to power large farms of GPU as a service. There's important things to tackle in terms of the energy grid. So we've got one of our new partners here, a customer, IRIN, who is actually developing our data centers that are hosting massive amounts of XE9680s clustered together, all powered by renewable energy, either water-driven or air-driven. They're air-cooled designs-

>> Amazing.

Brian Payne

>> Yeah.

>> That's awesome.

Brian Payne

>> Because they want to optimize water usage, they've chosen the air-cooled designs. And what they've found is, with our products, they can operate their temperature of their data center at a higher level than they could with other designs out there which, again, lowered the overall operating cost of their environment. So as we think about customers who are looking and struggling with, hey, how am I going to get started up with providing generative AI capabilities within my company? Building out the infrastructure's a challenge. That's where this partner ecosystem comes into play. We can partner them up with someone who's doing it in an incredibly sustainable way and get them on the way to solving problems.

>> We talked about this yesterday. Arthur talked about it on stage. The new liquid cool system that's going to house 72 GPUs in one rack that Jensen was so excited about when we were talking to Michael.

>> Yeah. That was a nerdy moment.

>> It is liquid-cooled. Direct to chip liquid cooling. We think it's going to be 2.5 times more energy efficient than what we had in previous generations. So yes, these are large stamps in terms of power consumption, but the energy efficiency per operation is better than ever before. So the amount of processing they're able to drive per watt, we're looking to move that further and further away and be more efficient there.

>> Yeah. And these clusters are getting bigger and bigger, consuming more power. The cooling's critical. The on-prem dynamic is also interesting. I want to get your thoughts on this because you saw this a lot in the HPC workloads. HPC workloads, as we covered I think with you guys at your Innovation Center last year, they were old use cases. All the build the wing at Boeing and all kinds of biology, high-end stuff.

Brian Payne

>> Yeah.

>> Not common use cases. So that's coming down. The democratization of HPC is happening. You're starting to see workloads being scoped, known workloads. A bank application, for instance, might not be an HPC workload. But you inject gen AI into it, that's now an end-to-end workload from the device to the back end. So you can scope that.

>> Before we started, we were talking about SuperCompute, the amazing event that's traditionally been an HPC event. And we're seeing basically the lines between HPC and AI being blurred right now.

>> Yeah.

>> SuperCompute and other such events are now as much AI events as they are HPC events.

>> Yeah.

>> Totally agree. That is interesting.

>> And it'll be very interesting to see how some of these traditional HPC deployments evolve based on the innovation that's happening in what we would call AI workloads that may be discrete from HPC. We definitely think there's going to be a lot of interplay. And of course a lot of the work that's happening with AI is built on the shoulders of the innovation that's been happening in HPC for such a long time.

>> And healthcare. We just had a healthcare, medicine from Northwestern. Their specific, direct quote here on theCUBE: "Our data's on-prem. We want the AI factory here."

Brian Payne

>> Yeah.

>> Yeah.

>> And they've already made advances with your Innovation Lab, HPC Innovation Lab.

Brian Payne

>> Yeah.

>> And again, that points to what I'm seeing. And would love to get your reaction. As an end-to-end workload that was traditionally managed by, IT, has now new requirements for gen AI, which is complicated because it's generative, right?

>> Yeah.

>> It's a different category, as Jensen points out. That now can be scoped. I can now deploy GPUs in a trusted environment and say, "That's scoped." And it's no over-provisioning. It's basically a design.

>> Yeah.

>> That's a use case that I don't need the cloud for. I can say I could host that at a colo facility, I could locate it in the hospital itself.

>> Yeah. Go ahead.

Brian Payne

>> Oh, sorry. I was just going to say, as we think about the democratization of the technology and needing to get it closer to the data, you quickly run into a challenge of, how do I deploy that? How do I retrofit my data center to support that? That's one area where customers are looking to us for help. And we've developed services to help them with, hey, what is a sustainable data center plan? How do I get a multi-year plan to meet the needs of my business, respond to data gravity, and do it in a sustainable way? And so engaging customers on both, on the generative AI front, of, hey, how do I develop applications to solve those problems? And then how do I get the infrastructure that can help them?

>> And again, it's not to make it sound like it's true because it's not. Having a data modernization service like you guys have, a modern data center, sustainable data center service. It's great. Sounds sexy top line. Great. But there's a lot of blocking and tackling you've got going on.

Brian Payne

>> Yeah.

>> You've got asset management, waste recovery.

Brian Payne

>> Yes.

>> So what are those details? That's the heavy-duty stuff.

>> Take a step back. We talked about the AI factory. Like you were talking about, AI factory, it follows the data.

>> Yeah.

>> We believe AI factories will be everywhere. They'll be in data centers, they'll be at the edge, they'll be on users' desktops. Right?

>> Yeah.

>> Services is a huge part of that conversation. And we're being asked, because of all the experience we've had with AI even before gen AI was a thing, thousands of customers. We think we've got a lot of unique IP that we can help customers with from a process perspective. So asset recovery services. If you think about, hey, how do you responsibly recycle your e-waste? Make sure that that helps you with the recyclable, the circular economy. Data center and modernization services. We started this conversation talking about, you've got to think from the ground up in terms of how you've got to retrofit your data centers or build them from scratch. We have a lot of experience we can help customers with, and they're looking to us for that. And then even things like advisory services for building sustainable data centers. Right?

>> Yeah.

>> All of those capabilities we think are an important part of the help we can provide customers and help us be trusted advisor to them on all of these initiatives.

>> This is one of those cases where you become what you're known for. You've done it for yourself.

>> Yeah.

>> It's not something that's, "Hey, let's start a service and start servicing people."

>> Yeah.

>> Well, you've got the capabilities. Why not bring it out to the customers? That's

>> Customers need that help from us.

>> Yeah. Yeah, that's a great win.

Brian Payne

>> In fact, our own Dell Digital team used that service to plan our generative AI capabilities.

>> I was going to ask that. And has the demand for sustainable data centers gone up since we've seen this AI revolution like crazy?

>> Yeah. I guess-

>> I can imagine it wasn't necessarily as popular of a conversation.

>> I think it has been building for a while, right?

Brian Payne

>> Yeah.

>> Everything that's been happening in the industry, even before AI, ESG initiatives, sustainable data centers, et cetera were really driving a lot of those conversations. We certainly think there is a lot more urgency to it because now it's a roadblock to, if you don't solve it, it's a bottleneck to getting your AI efforts. So you're right, yeah.

>> Much more front of mind in the sense of, like, "Ooh, we've got to do this right now or else we're in a bad spot."

>> And it's easier to measure the ROI. It's easier to measure the impact of the business if you don't get those things figured out now.

>> Well, I think AI is an obvious. Jumps off the page from a cost perspective and from a, "Sustainable." It blows all the metrics out of the water. Wait a minute, all this horsepower. That's compute. Power's being used.

>> Yeah.

>> And OpenAI, they're case in point. They've been a big poster child. It costs a lot to run these generative AI workloads from terms of a power.

>> Yeah. Absolutely.

>> So it jumps off the page, "Hey we're looking, doing great." Boom.

>> Yeah.

>> So it's like you've got-

>> Yeah, exactly.

>> You've got to address it.

>> Yeah.

>> Yeah. And like we were talking about earlier with this faster development cycle, machines being able to be made faster, two to three years, traditional cycle now nine months. People craving AI computers. You guys announcing them, which is very exciting. There's going to be a lot more e-waste in theory, depending. And I love that you guys, you have your Concept Luna project, which is really cool. I don't know if you know about this, John, but when I was doing my research for you guys, I thought this was so interesting. It reduces the disassembly time, which is not something I thought about in terms of sustainability, from an hour down to minutes.

>> That's not obvious.

>> No, which is super unique when thinking about this. Coming back to this holistic ecosystem health, what a thoughtful thing to be thinking about. And there are 57 million tons of e-waste around the world every year.

>> Yeah.

>> That's every year

>> E-waste, who thinks about that? Well, you guys do. I don't.

>> They do. That's what I mean. This is about sustainability.

>> Yeah, it really is. And to your point, it's only going to get more magnified.

Brian Payne

>> Yeah.

>> Right.

Brian Payne

>> I think in order to address these really systemic challenges, we take an end-to-end approach as a company. And so it's literally-

>> Very clear.

Brian Payne

>> Ingrained into how we think about engaging with the supply chain. How do we think about the design of our products? To your point about disassembly time. And then on the back end, how do we do recycling, repurposing? And it's just built into the way we operate.

>> Yeah.

Brian Payne

>> And I think that enables us to be successful. But we still have big challenges ahead to get to our stated goals for 2030.

>> I know. It's ambitious goals you've laid out. I was very impressed. But Dell's been putting out an ESG report since 1998, which is pretty early, especially for large company, so I am not surprised. All right, so we let Varun do something special yesterday and I want to let you have your shot, Brian. Since we're talking about a sustainable world, it's all about not just what we do in our day jobs but also making the planet better for our friends and our families and our lives. Is there anyone you'd like to say hello to?

Brian Payne

>> I think I should say hello to my wife Julie and to my two kids, who we're trying to make a more sustainable world for, Charlotte and Claire.

>> Awesome.

>> I love that.

>> Brian's doing a good job. He's working.

>> So what's the conversation like with the fam around the dinner table about AI? Talk to them about it?

Brian Payne

>> Yeah. Our kids are different ages, so we'll have different perspectives. I have a 19-year-old that's a freshman in college, and she's just chomping at the bit to learn everything she can. Figure out how to code and how to apply it. My 17-year-old is also looking into it, figuring out how it's going to do her homework.

>> Yes. I know.

Brian Payne

>> She's lucky.

>> It's a cheeky assistant. That's a cheeky-

Brian Payne

>> The reasoning is really good in the math equations, too.

>> Yeah. Yeah. AI is our little cheeky assistant. Brian, Varun, this has been a really great session.

Brian Payne

>> Thank you, Savannah.

>> This, we went longer than expected. We're actually over time because you're all so interesting. So thank you for bringing this very important conversation. It's been fantastic both times, too. We're really getting to hang out this week. I love it.

>> Likewise.

>> And John, thank you for your fabulous insights you candor, as always.

>> Thank you. Thank you.

>> And thank all of you for tuning in to our fantastic three days of live coverage here in Las Vegas, Nevada at Dell Tech World. My name's Savannah Peterson. You're watching theCUBE, the leading source for enterprise tech news.