Unlocking AI - What’s Needed to Build a Modern Data Center Infrastructure - Six Five On The Road at Dell Tech World 2025

At Dell Tech World 2025, hosts Patrick Moorhead and Daniel Newman speak with Dell Technologies’ President of Infrastructure Solutions Group, Arthur Lewis, about how AI is fundamentally reshaping data center design and accelerating AI adoption across industries. Their discussion sets the stage for understanding how data centers are evolving to become the engines of AI-driven innovation.

Key takeaways include:

🔹The AI-Driven Data Center Revolution: The conversation explores the transformative impact of AI on traditional data centers, highlighting the shift towards consolidation and re-configuration for space and energy efficiency.

🔹Architecting for AI: Lewis details the essential considerations for customers re-architecting their data centers, including the benefits of a disaggregated architecture, combining flexibility and simplicity to support diverse AI workloads.

🔹Unlocking AI's Potential: Modernizing data centers will facilitate the seamless adoption of AI, offering significant cost savings, and improving decision-making, agility, and competitiveness for businesses.

🔹Building a Robust AI Ecosystem: Lewis outlines Dell Technologies' efforts in fostering an innovative AI ecosystem, emphasizing the importance of end-to-end solutions and strategic partnerships to empower customers across every use case and industry.

‍

Learn more at Dell Technologies.

‍

Watch the full video at Six Five Media, and be sure to subscribe to our YouTube channel, so you never miss an episode.

‍

Or listen to the audio here:

‍

Disclaimer: Six Five On The Road is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded, and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors, and we ask that you do not treat us as such.

‍

Patrick Moorhead: We are back. The Six Five is On The Road here in Las Vegas for Dell Tech World 2025. Daniel, it's been a great show. Surprisingly, we're talking about AI from client PCs, industrial edge to the data center.

‍

Daniel Newman: Yeah, we knew this was going to be the case, Pat. These technologies proliferate quickly, but not that quickly. But one of the things that's been great about this year is sort of seeing it all being brought to life right. Early on, there's a lot of building out the infrastructure. We're seeing these 747s full of racks being delivered and they're getting plugged in and we're turning it on and we're tuning these things up. But this week, starting with Michael's keynote, we're hearing a lot more about the customer and how they're applying it and getting value and ultimately how we as consumers are getting real measurable value from AI.

‍

Patrick Moorhead: That's right. It's one thing, you know, to stand up a CSP and the type of infrastructure they have, it's a whole different ball game to go into the enterprise data center and then the enterprise edge. I can't imagine a better person to talk about this than Arthur Lewis. Arthur, great to see you.

‍

Arthur Lewis: Great to be here. Thank you.

‍

Patrick Moorhead: Yeah, thank you. You heard a lot of great stuff. A lot of talk about disaggregated infrastructure. It's a good time to be in tech.

‍

Arthur Lewis: Absolutely.

‍

Daniel Newman: Well, behind the scenes we were talking to Arthur a little bit about working out and bench pressing. But I think the new real ironman, the new world's strongest man is going to be. Can you deadlift in one of these 72 node racks?

‍

Patrick Moorhead: That is, yes.

‍

Daniel Newman: These things are pretty heavy.

‍

Arthur Lewis: The ultimate flex right there.

‍

Daniel Newman: The ultimate flex is I can lift the rack single handedly. Arthur, it's great to see you though. As I said kind of in the preamble, AI is red hot. You know, a lot of customer centricity here. But tell us a little bit. You're running, you know, ISG for all of Dell. What are you seeing and hearing?

‍

Arthur Lewis: Yeah, look, I mean this is, you know, as we talked about last year, what, you know, one of the most interesting times in my lifetime. I've been in the industry for 30 years and never have I been so excited because, you know, with, you know, moving from perception to generative to reasoning AI, you now have a technology that is giving customers the ability to make use of their most valuable asset, their data. It all comes down to data. And so, you know, customers are taking a look at understanding, ROI, understanding the use case, understanding their model strategy, but really it comes down to understanding their data strategy and how they have to re-architect their data center because it's clear that the data center of the last three decades ain't the data center of the next three decades. And so customers are wrestling with how to get started small as I think about re-architecting my data center for the future.

‍

Patrick Moorhead: Yeah. I mean you can't just put a Grace Blackwell, a GB 300 and stick it in a data center that was architected five, ten, whatever years ago. What should they be thinking about when re-architecting it? Because they don't have the ability to just go clean slate and build a new data center. In fact, they don't want to just go out and build new data centers. How can they leverage what they have and protect those investments and make other investments if they need to?

‍

Arthur Lewis: Yeah. So what we like to talk to enterprise customers is about starting small. But think about sort of the long game. Because at the end of the day, again, it kind of comes down to, you know, what is your data strategy? Because obviously data feeds artificial intelligence and artificial intelligence requires a lot of premium grade data. Right. In order to feed it. So we like to talk about what you think about standing up private clouds, which are going to be incredibly important for the future. How do you stand up an AI platform for your underlying unstructured structured data as well your ingest data? I mean, think about it. Effectively cleaning, prepping, ingesting, querying and ultimately securing data across diverse data sets becomes an incredibly complicated task. Right. So starting with disaggregated infrastructure, what do you think about your data platform? All really tied back to your data strategy. Right. So that's kind of how we think about it.

Patrick Moorhead: Can I just thank you for bringing up data strategy like you and I had a conversation two and a half years ago. Both of us acknowledged, you know, as a challenge. But now you're doing a lot about that.

‍

Daniel Newman: Bravo.

‍

Arthur Lewis: Well, yeah. And there's a lot yet to be done. We've done a lot and there's a lot to be done. Because again, if you really think about it, the world's value sits in all of this data. And today 80% of that data sits in cold backup.

‍

Patrick Moorhead: That's right.

‍

Arthur Lewis: Think about a world where you have silos broken down, everything connected, the majority of the world's storage sitting in hot and warm tiers, constantly in circulation, feeding these AI engines, feeding these agents. I mean, you're going to have thousands of agents running through the data center, performing millions of tasks on a daily basis. This requires a connected system. Silos of the past absolutely have to be dismantled and customers need to be thinking about that.

‍

Daniel Newman: It's kind of interesting, Arthur. A couple of weeks ago at Think Arvind Krishna, CEO of IBM, said 99% of enterprise data  had not touched AI yet. So I mean the 80% you're mentioning and also it sounds like the vast majority of the rest of data in enterprises. It's such a big opportunity for you and what you're doing. But you know, to Pat's point about, you know, thank you for talking about data, you know, this is not, it's interesting because this isn't really a new problem. I mean, we had this, you know, big data era, the first part of the platform era, the analytics era, the advanced analytics era, the machine learning era by the way, which is kind of a lot of what AI has become. And for whatever reason, this whole phase, like enterprises never really got their data right. Are you seeing something different now or is AI maybe an enabler of that is the fact that unstructured data can be more immediately closer to compute and utilized in its form. Like, are we getting there?

‍

Arthur Lewis: Yeah, I know, absolutely. Like I said, you know, you move from, you know, deep learning to machine learning to sort of perception to now you get these long thinking reasoning models and you know, the outcomes that you get from applying AI to your data are exponentially different than they were five years ago. Right. And you look at, across any industry, across any type of use case, right? Artificial intelligence is going to raise human cognition because humans are not going to have to focus on the mundane tasks, you know, that they think about doing on a daily basis. They're going to be able to up level their thinking, to focus on higher order problems. We're going to make everybody smarter, it's going to make everybody better. And you know, what we say is, you know, for enterprises that say, well, I'm going to be a fast follower and let someone else figure it out, they're kind of dead in the water, right? The revolution is not a spectator sport. You got to get on the bus, you got to start small, you got to practice, you got to learn, you got to build your skill sets, you got to enable key partners and, and you got to start to figure it out.

‍

Daniel Newman: Are you sort of indicating though that there was an inflection? I mean, I'm genuinely curious. Like the era of the mid teens where we were all piping big data and talking about how our company runs better because we're using data better, right? Sounds like most of it never. That didn't really materialize. But you're kind of saying the AI inflection, if I'm hearing you right, is that the value is so meaningful material now that companies really can't any longer sort of kick the can down the road. They have to modernize their data center, be ready for AI adoption or they're going to, if I heard you right, basically they're going to fall behind really quick.

‍

Arthur Lewis: It's an eye opening exercise because you know, when you sit down and you talk to enterprise customers, you know, every question leads to an answer, leads to five questions, leads to five answers, leads to 10 more questions. And what customers are realizing is not only do I have to modernize my data center, I need to modernize my internal operations, right? Because I need to ensure that all of my data is available and discoverable to be able to analyze it, process and actually apply artificial intelligence to it. Right. So I think there is an inflection point where now you see the tangible value of using your data with artificial intelligence and the outcomes and the productivity gains to be had that people can't sit on the sidelines anymore.

‍

Patrick Moorhead: So let's get into some nuts and bolts, right? I mean I  profoundly believe in everything that we've discussed so far. What I'm trying to figure out though is what is it when I modernize it, how does that actually make me go quicker? Or is it really just saying you can't do modern day AI without rearchitecting? Or can I actually do this quicker if I re. I'm wondering if you could even do this without re-architecting because it would just be, it would be slow, it would be expensive and it would be very highly complex and rigid. Other than that, it's great.

‍

Arthur Lewis: Yeah, no you can.

‍

Daniel Newman: Simple enough. Yeah.

‍

Arthur Lewis: I mean you can absolutely do it without re-architecting it. The question is, to what extent can you actually do it, right? And like what we counsel customers is hey, you gotta kind of pick five or 10 different use cases, start small, build it out to really understand what a new architecture looks like for you. Because you know, from my perspective, and you know, this may come off, you know, sounding bad, but you know, for many years I think data centers were viewed as a cost center, right? And when you think about a cost center as a business leader, you think about how do I optimize it for cost in the future? There's no question that a data center becomes a value center. It becomes an intelligent enterprise. And when you think about a value center, you start thinking about how to invest and augment and build out more capabilities. So I think customers kind of, you're not going to kind of do like a clean sheet exercise, blueprint everything and do a wholesale, pull everything out and move everything in. You have to build a plan, you have to have a strategy. You got to start small, you got to start thinking about the use case, but you got to start thinking about the long game. Because, you know, last year, you know, we were talking about generative AI, now we're talking about these reasoning models. We didn't talk about tree of thought logic a year ago. Right. So what's going to come next year?

‍

Patrick Moorhead: That's right.

‍

Arthur Lewis: That's the equivalent of a reasoning model. Because AI is the worst today that it's ever going to be.

‍

Patrick Moorhead: Yeah. I mean even something simple, not simple, but server consolidation, I mean you get 20 to 1 consolidation and you have this energy. Right. Power that you could devote into putting into GPUs or accelerators or anything like that. No, it is daunting. And I'm sure you could ask this question which is, wait a second, I talked to the CSPs and they're telling me I can only do this in the public cloud. Michael was very provocative and I do agree with him when he said on stage that it is going to be low cost. You do have more control. How much of a reality are we in here that we can do all this stuff on premises?

‍

Arthur Lewis: Well, let's start with the fundamental fact that 85% of the world's data sits on prem today. And you know, so let's start with that. So, you know, we believe strongly and we've seen and have a lot of proof points that AI is absolutely going to follow the data.

‍

Patrick Moorhead: Yeah.

‍

Arthur Lewis: And when you think about where the data is going to be created in the future, it's going to be created on prem. And so customers are taking a look at cost, security and performance as kind of like the three measures here. You know, definitely cost, probably 60 to 100% more expensive to to run in the cloud. You're going to have higher levels of latency, your performance, you're not going to be able to hit your SLAs on the tokens that you want to generate based on the specific use case. And security becomes a really big issue. Right. Because you know, people need to fundamentally understand the technology and ensure that there's strong security and governance across everything. Because, you know, again, generative AI, now, these reasoning models, this is kind of like sequencing your DNA. Like, you know, are you really going to want to put this stuff in the cloud? Because you're not just putting data in the cloud, you're now putting insights into the cloud. And that scares a lot of customers. But ultimately they don't have to put it in the cloud because 85% of the world's data sits on prem. That's where it's going to sit.

‍

Patrick Moorhead: So you're not doing this alone. I've seen a lot of partners, partners that I would never have expected for you to partner with, like Google, running Google Gemini on your infrastructure. Tell me more about this new breed of AI ecosystem that you're building.

‍

Arthur Lewis: Yeah, so, you know, the AI factory would not and could not exist without a very strong ecosystem of partners. And we'll be announcing joint innovations with Nvidia, with AMD, with Intel, with Meta, with Mistral, with Hugging Face, with Glean, with Red Hat, with Eny, with Accenture, with Deloitte. And I'm so incredibly proud that we're going to talk about our collaboration to bring Gemini models on Prem for Dell customers. 

‍

Patrick Moorhead: No, it is. I mean, say that again. 

‍

Arthur Lewis: Google Gemini on Prem for Dell customers.

‍

Patrick Moorhead: Right.

‍

Arthur Lewis: Amazing.

‍

Patrick Moorhead: Yeah, it really is.

‍

Daniel Newman: We talked about that at length when Pat and I were doing the post mortem on the keynote itself. We talked about the partners, and that was one area that both of us really seemed to feel very strongly about was the partnership with Google. What that says, how they're sort of enabling, you know, because Google has every reason to not do that realistically. And they do it because I think they see what you see, the size of the opportunity, the size of the market, 85% of that data sitting on prem. And of course, wanting to be realistic about what a partnership looks like and what an enterprise looks like. Arthur, as we only have a moment left here, I would love to kind of know for all the customers, I mean, what an attendance, what a group. It's hard to get around this place. Thankfully, it's a large center. But like, what do you want customers that spend time with you at Dell Technologies World to really walk away with this week?

‍

Arthur Lewis: We want them to walk away with. This is a revolution and a redesigning of data centers that is picking up steam. And our AI factories sit at the center of this revolution. We've taken the learnings from the largest CSPs and applied it all the way down for all these enterprise deployments. We have AI factories across every vertical, every industry, every use case. And, you know, we just want customers to know that they can sit down with us, because with us, they're not just ready for the future, they're defining the future.

‍

Daniel Newman: Well, Arthur, I want to thank you so much for joining us today here at Dell Technologies World Day One. We always love having you on the Six Five. Let's do it again soon.

‍

Arthur Lewis: Thanks, guys. I really appreciate it.

‍

Patrick Moorhead: Thank you.

‍

Daniel Newman: And thank you, everyone, for tuning in to this Six Five On The Road here at Dell Technologies World 2025. We're going to step out for a minute, but we'll be back and we'll talk to you soon.

‍