Advancing AI 2026 | Trish Damkroger, HPE

Clips
More from Advancing AI 2026

Trish Damkroger

SVP & GM, HPC & AI Business Unit

HPE

play_circle_outline Convergence of HPC, AI, modeling, simulation, and data analytics

play_circle_outline AMD Advancing AI Day: Lisa Su Unveils EPYC Gen6, MI430X as HPE Partnership Drives Exascale at National Labs

play_circle_outline From Cray Heritage to GX5000 Liquid Cooling: Tackling Power Management and Programming for Mixed Workloads

play_circle_outline Liquid cooling deployment complexity and infrastructure risks (leaks, plumbing)

Info

Trish Damkroger, HPE

Trish Damkroger

SVP & GM, HPC & AI Business Unit HPE

search

(INTRO)

Dave Vellante

>> Hi everybody, welcome back to our live coverage. This is theCUBE's coverage of AMD's Advancing AI Day 2 here, we're at Moscone West at the top of the escalator. Stop by and see us, my name is Dave Vellante. Trish Damkroger is here, Senior Vice President and General Manager for HPC and AI infrastructure solutions at Hewlett Packard Enterprise. Good to see you, thanks for coming on.

Trish Damkroger

>> Good to see you.

Dave Vellante

>> Well, it's been a while since you've been in theCUBE. The last time we were talking, it was remote during COVID. We just put those years behind us.

Trish Damkroger

>> Exactly, I don't remember the pandemic.

Dave Vellante

>> It never happened. But I feel like for somebody who's been in HPC, this moment is built for you. it's just really amazing. I've been waiting for it. I remember I was at IDC and the folks who ran HPC were part of my group.

Trish Damkroger

>> Okay.

Dave Vellante

>> and they kept telling me, it's going to happen, it's going to happen, AI, AI, and finally it's here. wow.

Trish Damkroger

>> It has, yes.

Dave Vellante

>> Probably much bigger than we ever thought. So how do you describe it? How do you think about the intersection between HPC and AI? It's a marriage.

Trish Damkroger

>> As you said, I've been part of HPC for a long time, so I've seen it ebb and flow, but I think two things. For HPC, there's three workloads. there always has been, modeling and simulation, AI, and data analytics. So it's been a workload for HPC for years and years. But it's cool because now those things that we call super computers are now the exact same infrastructure that you needed for modeling and simulation that you need for AI. And I'm finally seeing a lot of this convergence, as you say, between where you have AI steering your HPC workflows, or you have data models versus physics -based models so you can go faster. And I was just at the Genesis Mission Summit in DC yesterday, and they were talking about how agents are now running, so it's not even steering, you have agents running your simulations. So, yeah. Wow. Super cool time to be in this world.

Dave Vellante

>> That's remarkable. We were talking offline. You were at Livermore. I was just recently at Oak Ridge checking out what they're doing with AMD HPE. Because I saw where it's being built. It was empty at the time. I'm sure it's starting to fill up. It's just so much activity going on. And, of course, a lot of excitement here. AMD is doing a great job. They need a bigger boat, I've been saying. So next year, I'm sure we're going to be across the street. What's the relationship like with AMD? It goes, at least the first thing out of her mouth today was HPC, right? It's in their DNA.

Trish Damkroger

>> She has been a huge supporter and fan of high -performance computing, and also the National Labs. So we have an amazing relationship with AMD. You're probably at Oak Ridge. We have the Lux system, which is the first AI system for the Genesis Mission that I just talked about. The big AI program and quantum program that's coming out of the Department of Energy. But we also have the next generation of their system. So they were the first system to exascale Frontier, which is an AMD and HPE -based system. And we are under contract to deliver their Discovery system, which will be the next big system. And it is also an AMD -based system. In fact, Lisa announced both the EPYC Gen 6 and then also their AMD Instinct MI430X. So very exciting about the relationship. Also number one, well, it was number one on the TOP500. China placed a system there in June.

Dave Vellante

>> Yes, saw that.

Trish Damkroger

>> But the number two system is also an AMD system at El Capitan at Lawrence Livermore.

Dave Vellante

>> Right, right.

Trish Damkroger

>> Where I came from.

Dave Vellante

>> Very cool. At Frontier, I have a picture of Lisa and Antonio's signature on Frontier, which is amazing.

Trish Damkroger

>> it was funny. I think Oak Ridge was asking, how do we, oh no, somebody was asking, how do you get those signature panels? I'm like, nobody gives up the signature panels for these supercomputers. What are you talking about?

Dave Vellante

>> Not a chance, right?

Trish Damkroger

>> No, no.

Dave Vellante

>> It's a point of pride. Yeah, and my takeaway in talking to the scientists at Oak Ridge was really this is an and story. There's CPUs. There's GPUs. There will be quantum as the tooling gets built out. And so it's not – we like to think about zero -sum games in this industry, but it's not. It's just this expanding – okay, I won't use the word bubble. Balloon. There you go.

Trish Damkroger

>> Yeah, bubble's not a good word.

Dave Vellante

>> It's an exciting time. So what are some of the other projects you're working on or envisioning and some of the challenges that you're focused on solving?

Trish Damkroger

>> So I think one of the biggest challenges is as the silicon gets bigger and requires more and more power, one of the things that HPE with our Cray heritage is known for is our liquid cooling technology. And so we continue to drive innovation in that space because that's going to be more and more important, whether it's being able to cool systems that have these higher TDP parts, but also warm water. So in Europe and some of the other countries, they are demanding that we bring in warm water. So it's bad enough we have to cool these hot silicon parts, now they're telling us to do it with warm water, no less.

Dave Vellante

>> 45 degrees, right?

Trish Damkroger

>> Yes, exactly. So those are some of the innovations that will be in our GX5000, which is going to be part of that DX, or the Discovery system. I think also just power management, the whole programming environment, how to be able to have that workflow versus just single workloads. Like, you may do a modeling simulation, but you may want AI. You're talking about quantum, how are we going to add quantum to this mix? So there's a lot of opportunity for innovation still in this space.

Dave Vellante

>> You brought up liquid cooling. I want to ask you some questions about that. I've been to many, many HPE Discovers, both here and overseas. Frankfurt, Barcelona, many events. But it was amazing more recently to see all the liquid cooling activity going on. You're seeing artisan plumbers. I remember the first Cray system I ever saw. The hose infrastructure was kind of scary.

Trish Damkroger

>> Right.

Dave Vellante

>> And now it's like beautiful.

Trish Damkroger

>> It is, isn't it?

Dave Vellante

>> And where there's huge focus on connection, integrity, and you have these artisan plumbers because you don't want green goo floating all over the floor. And people don't understand. There's so many complex pieces of these systems. And I've heard stories about, the infrastructure gets shipped and they leave it outside because it's, oh, it's hoses. We can leave it outside by the picnic table. No, these are highly sensitive systems. But it just goes to show you how complex these systems are and how much effort has to go into every little piece of the value chain, because if that goes wrong, you're talking about millions, hundreds of millions of dollars of infrastructure because of a leak. Can't have that.

Trish Damkroger

>> Right, right. Yes, and your system's down, which doesn't do anybody any good, right?

Dave Vellante

>> Right.Sovereignty's a hot topic. We've been writing several pieces lately. We have one of our new analysts, Amit, is doing a lot of work there. And the big takeaway is sovereignty's not just about data residency, right? It's much more than that. People want their own sort of technology stack. There's legal, there's financial, of course there's territorial, obviously operational as well. And so, there's a spectrum of sovereignness. What are you seeing in terms of sovereign AI? What's HPE's perspective on that?

Trish Damkroger

>> Yeah, no, I think Sovereign is a big topic. It was a big topic at our Discover that we had in Las Vegas and then the next week I was at ISC.

Dave Vellante

>> Right, ISC, yeah.

Trish Damkroger

>> And so I think for us, the way we define or look at Sovereign is it's people want end-to-end ownership, not only of their infrastructure but really their data. It's about the transparency of the data going into what the models are doing so that you can trust the data coming out or your actual output of those models. And I don't think it's necessarily, we've been doing Sovereign for 50 years because that's pretty much what national labs want all the time, right? But it's interesting how the pharmaceutical companies or a lot of the regulated industries are also asking for Sovereign. So it's a much broader topic than just our national labs or our nation states that we would think of before. So it's just, you got to think about control, controlling your data, compliance, because there's so many regulatory things, and then security. we need to make sure that we can protect the information that's there.

Dave Vellante

>> it's interesting you talk about the national labs in there, it's always been sovereign. kind of air gapped, if you will.

Dave Vellante

>> Yes. In order to get on the internet.

Trish Damkroger

>> 100%.

Dave Vellante

>> It's not trivial. We have a studio at the New York Stock Exchange, and even you see there, the focus on security, changing passwords every week, even simple things like that, and having separate networks that are fenced off. So, what are you finding as some of the customer challenges and learnings that they can take from the labs? Because there are trade-offs, right? If you go all in on sovereignty, even things like phone home and patching, you might be giving that up. So organizations have to trade off. What are those discussions like? What kind of advice would you give to folks?

Trish Damkroger

>> And I do think it's really dependent upon what's important to you. So we work with the customers to develop the software stack that makes sense for them. But like you say, it's going to be easier to patch if you don't have to take, if you're not completely isolated in the systems. And I think that the most important advice we give is you just really understand the trade-offs and that trade-off space. Because there's compliance that you don't always have options about, but there definitely is ease of use that you have more options with.

Dave Vellante

>> the trust pendulum is kind of swinging against, the frontier model vendors, the hyperscalers, the US government sometimes, and so you can understand that. But at the same time, you talk about trade-offs, there are some advantages with leading tech companies. And so I wonder how that's going to play out. sometimes you get immediate backlash, headlines, and then things maybe calm down a little bit and you find a balance. It's kind of like everything's going to the cloud. We're now starting to see more of a balance, distributed infrastructure. Any perspectives on that?

Trish Damkroger

>> No, and I always see, you always see these pendulum swings, as you say, and everything was going to go to the cloud and then the cost of it, the security of it, data, the storage price, et cetera. So then it started coming back on-prem And I think the thing with Sovereign and just that security, it's again going to be a pendulum. Right now, with what's going on, people are really scared about AI attacking their data, attacking their information. And yeah, so right now, maybe people are going to be a little bit more hesitant to be as open. But I think the pendulum will swing?

Dave Vellante

>> Yeah, yeah. You mentioned ISC recently. and Supercomputing coming up, I guess it's in November, in Chicago this year. It's such a great show. When did Supercomputing start?

Trish Damkroger

>> Okay, I believe they're, no. More, way more. They're probably on the 35th year, something like that. Yeah, you're right.

Dave Vellante

>> I think you're right, we're celebrating the 35th year anniversary. I think so. So okay, and then it was such a niche show, and now it's like an NVIDIA GTC.

Trish Damkroger

>> It's huge.

Dave Vellante

>> It's awesome. I know. Everybody's there, and the innovation is just tremendous. you see, like you say, liquid cooling, you see storage, you see compute, you see racks. it's software, autonomous vehicles, really, really fun event. What are you looking forward to this year for Supercomputing?

Trish Damkroger

>> as I mentioned earlier, I am a huge advocate of SC, being chair and being on the steering committee and part of it for way too many years. And I can remember talking to the exhibitors, the exhibit chair and kind of the people that put that on. I'm like, oh, everybody's scaling back. this is, right around COVID time, the pandemic. And now they have had to expand because of the amazing growth in the liquid cooling and power distribution and all these things that again, supercomputers are at the core of. So SC has now become a mainstream event. And I love to see the new innovation out there on the floor and the new technology. And then for me, it's all about seeing all my friends. even in this industry, it's a small world and it's great getting together with the people that you don't get to see very often.

Dave Vellante

>> A hundred percent. It's nice to see, and it really is a fun show.

Trish Damkroger

>> Yes.

Dave Vellante

>> The Supercomputing group is just so great. we have a big set there. They make it reasonably affordable for us to do that, but the best part is it's an ecosystem event. It's not like one company owns it, and so we get support from all of our great clients. My team is fact checking me. It's the 37th year.

Trish Damkroger

>> Okay, 37th year, okay. I apologize. So we already went past 35, yes.

Dave Vellante

>> That was, well, let me see, two years ago in Atlanta. Right?

Trish Damkroger

>> Yeah, yeah.

Dave Vellante

>> Which was a great show. I love how they move around cities. And the other point is, we've learned so much. We're now applying, we've learned so much from HPC, we're now applying that in commercial mainstream.

Trish Damkroger

>> 100%.

Dave Vellante

>> And it's so important. enterprises, this is new to them.

Trish Damkroger

>> Right, yeah, liquid cooling, but now even they're saying, okay, we're going to have to do this. And so, how do you retrofit and plumb your data centers that weren't built for that?

Dave Vellante

>> I was talking to one of my colleagues, I talk to a lot of enterprises and data center operators in the enterprise, and they struggle sometimes. Sometimes like, well, do we want to retrofit our data center? We've been air -cooled for years. Got to go back to the mainframe days. Right. And one of my colleagues who's known for forecasting prowess said, I think eventually many enterprises are going to go to liquid cooling because the cost advantages and performance advantages are going to be so great. They don't necessarily want to hear this, but if you look out maybe 10 -plus years from now, the cost per token and the performance is going to be so much more attractive. They almost have to do it.

Trish Damkroger

>> Right.

Dave Vellante

>> You're not going to do it all in the cloud, right?

Trish Damkroger

>> Right,

Dave Vellante

>> that's cost prohibitive. You're not going to do it all air -cooled because it...

Trish Damkroger

>> You'd have one server in a rack if you're going to air... it's going to get to the point where it's going to be ridiculous.

Dave Vellante

>> And the cost of doing it is going to maybe be de minimis relative to the value that you're going to get out of it.

Trish Damkroger

>> Right.

Dave Vellante

>> And this is kind of what we're learning. I think HPC is like a harbinger of things to come. So pay attention to high performance computing and of course HPE and the Cray acquisition, very strategic, you're at the heart of it, leader in that space. I'll give you final thoughts on this whole industry, this event.

Trish Damkroger

>> Well I think this is another example of just the growth that is needed, whether it's just needing CPUs for your organic workloads and even to run your AI, right? to the accelerated compute of the different GPUs that AMD introduced today. And there's just, the opportunity is great. And I love that it's all, the foundations of HPC and supercomputing are at the core of everything that's happening these days.

Dave Vellante

>> Great, it is like the main spring. Trish, thanks so much for coming on.

Trish Damkroger

>> Thank you. Appreciate it.

Dave Vellante

>> And we are here documenting, covering, reporting on AMD Advancing AI. I put forth a premise last weekend that AMD didn't have to take out NVIDIA, it just had to become an indispensable second AI platform, but Lisa Su threw down the gauntlet today I think she has bigger ambitions. Keep it right there, you're watching theCUBE, I'm Dave Vellante, we'll be right back right after this short break.

Trish Damkroger, HPE

search

(INTRO)

Dave Vellante

>> Hi everybody, welcome back to our live coverage. This is theCUBE's coverage of AMD's Advancing AI Day 2 here, we're at Moscone West at the top of the escalator. Stop by and see us, my name is Dave Vellante. Trish Damkroger is here, Senior Vice President and General Manager for HPC and AI infrastructure solutions at Hewlett Packard Enterprise. Good to see you, thanks for coming on.

Trish Damkroger

>> Good to see you.

Dave Vellante

>> Well, it's been a while since you've been in theCUBE. The last time we were talking, it was remote during COVID. We just put those years behind us.

Trish Damkroger

>> Exactly, I don't remember the pandemic.

Dave Vellante

>> It never happened. But I feel like for somebody who's been in HPC, this moment is built for you. it's just really amazing. I've been waiting for it. I remember I was at IDC and the folks who ran HPC were part of my group.

Trish Damkroger

>> Okay.

Dave Vellante

>> and they kept telling me, it's going to happen, it's going to happen, AI, AI, and finally it's here. wow.

Trish Damkroger

>> It has, yes.

Dave Vellante

>> Probably much bigger than we ever thought. So how do you describe it? How do you think about the intersection between HPC and AI? It's a marriage.

Trish Damkroger

>> As you said, I've been part of HPC for a long time, so I've seen it ebb and flow, but I think two things. For HPC, there's three workloads. there always has been, modeling and simulation, AI, and data analytics. So it's been a workload for HPC for years and years. But it's cool because now those things that we call super computers are now the exact same infrastructure that you needed for modeling and simulation that you need for AI. And I'm finally seeing a lot of this convergence, as you say, between where you have AI steering your HPC workflows, or you have data models versus physics -based models so you can go faster. And I was just at the Genesis Mission Summit in DC yesterday, and they were talking about how agents are now running, so it's not even steering, you have agents running your simulations. So, yeah. Wow. Super cool time to be in this world.

Dave Vellante

>> That's remarkable. We were talking offline. You were at Livermore. I was just recently at Oak Ridge checking out what they're doing with AMD HPE. Because I saw where it's being built. It was empty at the time. I'm sure it's starting to fill up. It's just so much activity going on. And, of course, a lot of excitement here. AMD is doing a great job. They need a bigger boat, I've been saying. So next year, I'm sure we're going to be across the street. What's the relationship like with AMD? It goes, at least the first thing out of her mouth today was HPC, right? It's in their DNA.

Trish Damkroger

>> She has been a huge supporter and fan of high -performance computing, and also the National Labs. So we have an amazing relationship with AMD. You're probably at Oak Ridge. We have the Lux system, which is the first AI system for the Genesis Mission that I just talked about. The big AI program and quantum program that's coming out of the Department of Energy. But we also have the next generation of their system. So they were the first system to exascale Frontier, which is an AMD and HPE -based system. And we are under contract to deliver their Discovery system, which will be the next big system. And it is also an AMD -based system. In fact, Lisa announced both the EPYC Gen 6 and then also their AMD Instinct MI430X. So very exciting about the relationship. Also number one, well, it was number one on the TOP500. China placed a system there in June.

Dave Vellante

>> Yes, saw that.

Trish Damkroger

>> But the number two system is also an AMD system at El Capitan at Lawrence Livermore.

Dave Vellante

>> Right, right.

Trish Damkroger

>> Where I came from.

Dave Vellante

>> Very cool. At Frontier, I have a picture of Lisa and Antonio's signature on Frontier, which is amazing.

Trish Damkroger

>> it was funny. I think Oak Ridge was asking, how do we, oh no, somebody was asking, how do you get those signature panels? I'm like, nobody gives up the signature panels for these supercomputers. What are you talking about?

Dave Vellante

>> Not a chance, right?

Trish Damkroger

>> No, no.

Dave Vellante

>> It's a point of pride. Yeah, and my takeaway in talking to the scientists at Oak Ridge was really this is an and story. There's CPUs. There's GPUs. There will be quantum as the tooling gets built out. And so it's not – we like to think about zero -sum games in this industry, but it's not. It's just this expanding – okay, I won't use the word bubble. Balloon. There you go.

Trish Damkroger

>> Yeah, bubble's not a good word.

Dave Vellante

>> It's an exciting time. So what are some of the other projects you're working on or envisioning and some of the challenges that you're focused on solving?

Trish Damkroger

>> So I think one of the biggest challenges is as the silicon gets bigger and requires more and more power, one of the things that HPE with our Cray heritage is known for is our liquid cooling technology. And so we continue to drive innovation in that space because that's going to be more and more important, whether it's being able to cool systems that have these higher TDP parts, but also warm water. So in Europe and some of the other countries, they are demanding that we bring in warm water. So it's bad enough we have to cool these hot silicon parts, now they're telling us to do it with warm water, no less.

Dave Vellante

>> 45 degrees, right?

Trish Damkroger

>> Yes, exactly. So those are some of the innovations that will be in our GX5000, which is going to be part of that DX, or the Discovery system. I think also just power management, the whole programming environment, how to be able to have that workflow versus just single workloads. Like, you may do a modeling simulation, but you may want AI. You're talking about quantum, how are we going to add quantum to this mix? So there's a lot of opportunity for innovation still in this space.

Dave Vellante

>> You brought up liquid cooling. I want to ask you some questions about that. I've been to many, many HPE Discovers, both here and overseas. Frankfurt, Barcelona, many events. But it was amazing more recently to see all the liquid cooling activity going on. You're seeing artisan plumbers. I remember the first Cray system I ever saw. The hose infrastructure was kind of scary.

Trish Damkroger

>> Right.

Dave Vellante

>> And now it's like beautiful.

Trish Damkroger

>> It is, isn't it?

Dave Vellante

>> And where there's huge focus on connection, integrity, and you have these artisan plumbers because you don't want green goo floating all over the floor. And people don't understand. There's so many complex pieces of these systems. And I've heard stories about, the infrastructure gets shipped and they leave it outside because it's, oh, it's hoses. We can leave it outside by the picnic table. No, these are highly sensitive systems. But it just goes to show you how complex these systems are and how much effort has to go into every little piece of the value chain, because if that goes wrong, you're talking about millions, hundreds of millions of dollars of infrastructure because of a leak. Can't have that.

Trish Damkroger

>> Right, right. Yes, and your system's down, which doesn't do anybody any good, right?

Dave Vellante

>> Right.Sovereignty's a hot topic. We've been writing several pieces lately. We have one of our new analysts, Amit, is doing a lot of work there. And the big takeaway is sovereignty's not just about data residency, right? It's much more than that. People want their own sort of technology stack. There's legal, there's financial, of course there's territorial, obviously operational as well. And so, there's a spectrum of sovereignness. What are you seeing in terms of sovereign AI? What's HPE's perspective on that?

Trish Damkroger

>> Yeah, no, I think Sovereign is a big topic. It was a big topic at our Discover that we had in Las Vegas and then the next week I was at ISC.

Dave Vellante

>> Right, ISC, yeah.

Trish Damkroger

>> And so I think for us, the way we define or look at Sovereign is it's people want end-to-end ownership, not only of their infrastructure but really their data. It's about the transparency of the data going into what the models are doing so that you can trust the data coming out or your actual output of those models. And I don't think it's necessarily, we've been doing Sovereign for 50 years because that's pretty much what national labs want all the time, right? But it's interesting how the pharmaceutical companies or a lot of the regulated industries are also asking for Sovereign. So it's a much broader topic than just our national labs or our nation states that we would think of before. So it's just, you got to think about control, controlling your data, compliance, because there's so many regulatory things, and then security. we need to make sure that we can protect the information that's there.

Dave Vellante

>> it's interesting you talk about the national labs in there, it's always been sovereign. kind of air gapped, if you will.

Dave Vellante

>> Yes. In order to get on the internet.

Trish Damkroger

>> 100%.

Dave Vellante

>> It's not trivial. We have a studio at the New York Stock Exchange, and even you see there, the focus on security, changing passwords every week, even simple things like that, and having separate networks that are fenced off. So, what are you finding as some of the customer challenges and learnings that they can take from the labs? Because there are trade-offs, right? If you go all in on sovereignty, even things like phone home and patching, you might be giving that up. So organizations have to trade off. What are those discussions like? What kind of advice would you give to folks?

Trish Damkroger

>> And I do think it's really dependent upon what's important to you. So we work with the customers to develop the software stack that makes sense for them. But like you say, it's going to be easier to patch if you don't have to take, if you're not completely isolated in the systems. And I think that the most important advice we give is you just really understand the trade-offs and that trade-off space. Because there's compliance that you don't always have options about, but there definitely is ease of use that you have more options with.

Dave Vellante

>> the trust pendulum is kind of swinging against, the frontier model vendors, the hyperscalers, the US government sometimes, and so you can understand that. But at the same time, you talk about trade-offs, there are some advantages with leading tech companies. And so I wonder how that's going to play out. sometimes you get immediate backlash, headlines, and then things maybe calm down a little bit and you find a balance. It's kind of like everything's going to the cloud. We're now starting to see more of a balance, distributed infrastructure. Any perspectives on that?

Trish Damkroger

>> No, and I always see, you always see these pendulum swings, as you say, and everything was going to go to the cloud and then the cost of it, the security of it, data, the storage price, et cetera. So then it started coming back on-prem And I think the thing with Sovereign and just that security, it's again going to be a pendulum. Right now, with what's going on, people are really scared about AI attacking their data, attacking their information. And yeah, so right now, maybe people are going to be a little bit more hesitant to be as open. But I think the pendulum will swing?

Dave Vellante

>> Yeah, yeah. You mentioned ISC recently. and Supercomputing coming up, I guess it's in November, in Chicago this year. It's such a great show. When did Supercomputing start?

Trish Damkroger

>> Okay, I believe they're, no. More, way more. They're probably on the 35th year, something like that. Yeah, you're right.

Dave Vellante

>> I think you're right, we're celebrating the 35th year anniversary. I think so. So okay, and then it was such a niche show, and now it's like an NVIDIA GTC.

Trish Damkroger

>> It's huge.

Dave Vellante

>> It's awesome. I know. Everybody's there, and the innovation is just tremendous. you see, like you say, liquid cooling, you see storage, you see compute, you see racks. it's software, autonomous vehicles, really, really fun event. What are you looking forward to this year for Supercomputing?

Trish Damkroger

>> as I mentioned earlier, I am a huge advocate of SC, being chair and being on the steering committee and part of it for way too many years. And I can remember talking to the exhibitors, the exhibit chair and kind of the people that put that on. I'm like, oh, everybody's scaling back. this is, right around COVID time, the pandemic. And now they have had to expand because of the amazing growth in the liquid cooling and power distribution and all these things that again, supercomputers are at the core of. So SC has now become a mainstream event. And I love to see the new innovation out there on the floor and the new technology. And then for me, it's all about seeing all my friends. even in this industry, it's a small world and it's great getting together with the people that you don't get to see very often.

Dave Vellante

>> A hundred percent. It's nice to see, and it really is a fun show.

Trish Damkroger

>> Yes.

Dave Vellante

>> The Supercomputing group is just so great. we have a big set there. They make it reasonably affordable for us to do that, but the best part is it's an ecosystem event. It's not like one company owns it, and so we get support from all of our great clients. My team is fact checking me. It's the 37th year.

Trish Damkroger

>> Okay, 37th year, okay. I apologize. So we already went past 35, yes.

Dave Vellante

>> That was, well, let me see, two years ago in Atlanta. Right?

Trish Damkroger

>> Yeah, yeah.

Dave Vellante

>> Which was a great show. I love how they move around cities. And the other point is, we've learned so much. We're now applying, we've learned so much from HPC, we're now applying that in commercial mainstream.

Trish Damkroger

>> 100%.

Dave Vellante

>> And it's so important. enterprises, this is new to them.

Trish Damkroger

>> Right, yeah, liquid cooling, but now even they're saying, okay, we're going to have to do this. And so, how do you retrofit and plumb your data centers that weren't built for that?

Dave Vellante

>> I was talking to one of my colleagues, I talk to a lot of enterprises and data center operators in the enterprise, and they struggle sometimes. Sometimes like, well, do we want to retrofit our data center? We've been air -cooled for years. Got to go back to the mainframe days. Right. And one of my colleagues who's known for forecasting prowess said, I think eventually many enterprises are going to go to liquid cooling because the cost advantages and performance advantages are going to be so great. They don't necessarily want to hear this, but if you look out maybe 10 -plus years from now, the cost per token and the performance is going to be so much more attractive. They almost have to do it.

Trish Damkroger

>> Right.

Dave Vellante

>> You're not going to do it all in the cloud, right?

Trish Damkroger

>> Right,

Dave Vellante

>> that's cost prohibitive. You're not going to do it all air -cooled because it...

Trish Damkroger

>> You'd have one server in a rack if you're going to air... it's going to get to the point where it's going to be ridiculous.

Dave Vellante

>> And the cost of doing it is going to maybe be de minimis relative to the value that you're going to get out of it.

Trish Damkroger

>> Right.

Dave Vellante

>> And this is kind of what we're learning. I think HPC is like a harbinger of things to come. So pay attention to high performance computing and of course HPE and the Cray acquisition, very strategic, you're at the heart of it, leader in that space. I'll give you final thoughts on this whole industry, this event.

Trish Damkroger

>> Well I think this is another example of just the growth that is needed, whether it's just needing CPUs for your organic workloads and even to run your AI, right? to the accelerated compute of the different GPUs that AMD introduced today. And there's just, the opportunity is great. And I love that it's all, the foundations of HPC and supercomputing are at the core of everything that's happening these days.

Dave Vellante

>> Great, it is like the main spring. Trish, thanks so much for coming on.

Trish Damkroger

>> Thank you. Appreciate it.

Dave Vellante

>> And we are here documenting, covering, reporting on AMD Advancing AI. I put forth a premise last weekend that AMD didn't have to take out NVIDIA, it just had to become an indispensable second AI platform, but Lisa Su threw down the gauntlet today I think she has bigger ambitions. Keep it right there, you're watching theCUBE, I'm Dave Vellante, we'll be right back right after this short break.