Dell Technologies World 2026 | Rob Bruckner, Dell Technologies

Clips
More from Dell Technologies World 2026

Rob Bruckner

President, CSG Commercial

Dell Technologies

play_circle_outline Data sovereignty and privacy benefits of running models locally on devices

play_circle_outline Hardware Design for Agentic AI: Memory, Bandwidth, and Onboard Compute in Hybrid Local/On‑Prem/Cloud Architectures

play_circle_outline Dell Precision lineup configured for a wide range of inferencing capabilities and GPUs

play_circle_outline Retooling Organizations: Integrating Agentic Automation with Modern Development Workflows, Balancing Compute, Memory, Storage, Bandwidth, AI Compute, and Software Layers

Info

Rob Bruckner, Dell Technologies

Rob Bruckner

President, CSG Commercial Dell Technologies

search

Gemma Allen

>> Welcome back to theCUBE. I'm Gemma Allen, joined by my co-host, John Furrier. We are here live on the ground in Las Vegas at Dell Technology. We're at 2026 where all the action is happening. We're going to have a conversation right now about the evolving role of the PC as the ultimate AI endpoint. Joining us is Rob Bruckner, President of Client Commercial Client Solutions at Dell Technologies. Welcome, Rob.

Rob Bruckner

>> Thank you. Good to be here.

John Furrier

>> Yeah, good ...

Gemma Allen

>> So you are, I think, half a year into this role at Dell. You have certainly picked the right moment to join, Rob. A lot happening. Maybe let's start with, this is your first Dell Technologies as a member of staff. I'm sure you've been here as a partner. What's exciting you? Where is your head at, Rob, in this crazy chaotic moment we're in?

Rob Bruckner

>> Yeah. A lot of days it's spinning as well. There's the excitement of AI and AI really entering the PC. You have a lot of additional challenges. We have the supply chain challenges that are around us. And that can, daily, be dynamic. So having the operational excellence of a procurement department just to get things done is an essential part of this. We have lots of innovation happening that takes AI to the next level in a PC. And so all the things that we're doing haven't been invented yet all the time. We're kind of putting together new things that are emerging, bringing together solutions so that our customers can use them in everyday work, and learn as we go and make it even better and better and better instead of waiting around for everybody else to have figured it out and now it just works. So we're going to get to that place eventually, but right now it's about putting that in action.

John Furrier

>> I think the big, I don't say secret or silent or private win, but to me the big win for Dell Technologies is the client side OpenClaw story here because it really shows the future of the relationship that the human's going to have to their interface. We hear about headless systems, but at the end of the day, I'm going to have a device. I'll either talk to it on my body or it's going to be in my hand or a PC. This OpenClaw revolution speaks to this new generation of true personal computing where it's not a developer, it could be just a developer and like a business person running agents.

Rob Bruckner

>> That's right.

John Furrier

>> So the AI builder persona isn't like your classic developer. It could be anybody if they have the right tool. So to me, I think that was a really big win for Dell, but also it's going to probably make more sales because you got to run AI on these OpenClaws and NemoClaws, and everything's going to be happening here.

Rob Bruckner

>> Exactly right. Exactly right. And OpenClaw was to ... I mean, Jensen said at GTC as well, it was the agentic starting. That was like the big moment where agentic AI was really reaching out. People started to realize the power of agentic for basically orchestration of workflows that rethink exactly how you use the computer every day. Now, what we added to Dell was if you worked at OpenClaw, there's two things about OpenClaw that are still emerging. One is to install, configure, and bring OpenClaw, you'll raise your claw basically. Does take some amount of help for a lot of people to get active in that. So we want to make it easy for people to bring that forward in a PC and we want to make sure that it has a wrapper around it that brings the security manageability as we try to bring that forward even into commercial space. And that OpenClaw experience is a composable agentic experience. You can have the ability to compose your workflow versus it being given to you in something that's kind of canned, which limits your ability to really invent.

John Furrier

>> And the operating model is a big theme here. Gemma, you worked at Microsoft. You know Windows.

Gemma Allen

>> I do.

John Furrier

>> Yeah. Windows is on my machine right here. Dell bundles it in. I don't have to install it, get a license. It's already in there. But now with this new era, I can see kind of similarities by having everything on the machine. How do you think about that? Because it's not the same as Windows.

Gemma Allen

>> I think we also though Windows has been on such a journey too. When I was at Microsoft in the Windows VG back in like 2010, 2012, there was this whole debate, is the PC dead, right? We're now at a point where the PC is suddenly like the end point of opportunity for everything. And I think how folks reimagine what a PC is also needs somewhat of a cultural reset. How do you think about that even from the perspective of conversations that you have with clients and customers in the field? Because this means something different than it did 10 years ago, right?

Rob Bruckner

>> In a big way. And the first step is talking about AI PCs in a different way. I think the industry kind of went out pretty quickly to try to get out there with a neural processing unit based architecture originally. It did make the PC a little more personal because it was able to see how you can interface with the device in a little bit better way. It just didn't have the power to bring forward a really heavier workload that brings more value to the PC. So first thing we're doing is overcoming the starting place of what is an AI PC into, Is it a AI PC still? What's the terminology mean now? I know this is kind of a trivial comment to say it's like an AI PC, but it's something that we have to really face with customers right now. Redefine now what the value brings to the PC, and the value starts to emerge in two main areas as you're able to run things more local. First is token expense. So as we all know, agents are very chatty. And as you're trying to put the agents into some level of freedom but also control, you're starting to see a lot of loops that are emerging that are generating lots of tokens. The inference needed to bring values does not require a frontier level inference. It can actually be done with a reasonable size model. So you can start to have a more control of your token budgets or be a little bit more dynamic about when you want to use a frontier token. Second is sovereignty. What do I want to do about my own privacy and my own data, and how does that mean to me and making sure I can contain things that I want to keep private and secure in a sovereign way versus things that I'm fine being more open with? So the world really is a hybrid world for how you do all of those things.

Gemma Allen

>> Let's stay on cost for a second. Jon Siegel spoke yesterday and he mentioned a scenario whereby a developer burned through $3,400 in token costs in one day. I mean, that raised a lot of questions about the developer for one point. But talk about how this new opportunity for Dell and what you guys are bringing to the market solves for that. Solves for how total cost ownership all of those challenges right now around, like we hear about token lead rewards at enterprise. I don't know if you've heard this term. Again, it's crazy. It seems insane, but it is happening. So how does Dell think about that and how does it approach that in your kind of customer solution model, when you're talking to clients in the field?

Rob Bruckner

>> Yeah. It's a fascinating ... Well, this is a fascinating time because this type of topic, we'd be talking about it a year ago.

Gemma Allen

>> We have.

Rob Bruckner

>> So think about what we're seeing right now in our own usage with our own software development community is the actual workflow for optimizing and redoing your code or securing your code, that workflow's being created. So the developer that might use lots of tokens right away, it's because they're optimizing their workflow and they actually need a playground. It's essential for them. Otherwise, they're going to hold their innovation instincts because they're worried about expanding on their token budget. That's why some of the local AI brings that value to where you can still invent your workflow and then as you need to go put it into something a little bit more on the prem or even frontier, you now have a good clean lockdown. If you don't let people innovate when the technology is changing so quickly, then you're going to have a problem where they're afraid to use their tokens. That's the last thing you want your developers to do, right?

John Furrier

>> And the tokens are great because you can do them on your PC. I want to get your vision on this because what's happening is clear. The device is connected. Agent's going to be working on our behalf. The cloud is shown away. I can store stuff on the cloud and then tap it from my machine to go get it, retrieve it.

Rob Bruckner

>> Right.

John Furrier

>> But if I have an agentic force working with me, I could have it all done here. I think Michael Dell said and Jeff Clarke said, "This is the company if you're an entrepreneur." Your PC or box on stage, your server he showed, you could literally run everything from that if it's got connection to something. It could be cloud or another data center. So you start to see this kind of, I won't use the word client server because that's an old term, but you got kind of this power machine that's going to have a lot more backend connected to it. What's your thoughts around, that's my weird view on this, but like that's the way I would see that because if I have agents, I want to have instant uploads coming to me. I got to connect to maybe another system, another data center, service, a new cloud.

Rob Bruckner

>> It's an evolving area because it's evolving at the speed at which model sizes and intelligence can arrive locally on a machine. So this machine, your XPS 14 is example, it has good compute in it, but for real AI agentic workloads, you really need a good like 32, 64 gigabytes of memory. This allows you to put a bigger model in there. You got to have memory band to get the token rates up. It's not just human tokens, it's agent to agent tokens, which can be much 2 to 3X faster than what you do for human token, for example. And you need to have the AI compute built in. So the type of machine you need, people are buying up into the stack, realizing that what they're buying is for the future. Even if they may not be fully running workloads that are local on agentic, they know they're coming, they know they need to be purchasing and getting ready for that. Still, I think you're going to see a very hybrid world because you're going to see different paces of innovation available to you as you look from the frontier to time it takes from frontier to move on prem in a commercial environment, or you might take a parallel to commercial environment. Maybe in the consumer environment, you get VSPs and other type of businesses that are arriving to give you some of those intermediate services out from the frontier and then to local. You're going to see all three domains are going to play equally important but at different levels of capability and innovation available.

John Furrier

>> So do you think about this differently now, your view of the future? Local, just privacy, you mentioned some of those things about sovereignty. Is this now looked at differently or is it still the same thinking around ...

Rob Bruckner

>> I think it's similar, but I think what's different is there's some additional complexity about when you're running your agent, how much intelligence does it need to take the action that you've allowed it to do. Or an agent that's monitoring another agent through the flow, that intelligence parameter then dictates the size of inference you need to go do. And if it's a very complex inference that you have very sophisticated flow, you're going to have to get even more AI capability off the device. But for a lot of just keeping the type of things you and I might do as a consumer, it has the intelligence with 30, 70 billion parameter models to do much of that workflow. Why not do it there? And actually, when you look at possible congestion that's going to happen with all the tokens that are going to be distributed across the networks, doing more things local where you can actually do them makes a lot of sense in putting everything out off the device.

John Furrier

>> First describe what the news was specifically and then how you see them configuring it. Because again, I'm thinking about me. I'm like, "I want to do more. I want more action. I had a PC, I bought a printer. I got a bunch of storage." I can see the, I won't say scroll, but I can see the expansion. So what is the desk side agentic solution? How does that play out?

Rob Bruckner

>> So one of the insights we had as we were watching the agentic workflows start at the frontier actually and then start to move on prem, which is coming for all the enterprise folks, was that these are also things as we've been talking about, you can run locally. And then so we looked at our device stack and what do we have that's available that has the necessary hardware, meaning support for memory size, memory bandwidth for token rates, and AI processing capability keep pace. And all these things have to come in synchronization. It doesn't help you to have a big memory size with no memory bandwidth. All these things have to be balanced. So as we started to design and provision these machines, precision line seems like the right spot for us. This is a place where it's already a big compute. It already has GPUs everywhere. So we took both the mobile and fixed and started to map out now all the capabilities for what these things can do with agentic workflows. All the way from precision in the mobile line with some RTX GPUs into the fixed line where you can add multiple into the fixed slots, the GB10 and the GB300. And these have, if you look at the scale of compute, it's a pretty nice coverage from what you could do to start out in more of the precision, say seven line, for example, to the GP300. This dynamic range has never been seen before in a precision lineup and that allows you to, as I mentioned, it allows you to select where you want to be for the inferencing capability that your agentic flow can have. If you have a more sophisticated flow, if you have a more sophisticated need for intelligence when you're making decisions for the agent, you've got to move up the stack for that intelligence that you're buying. So that's the way we've been mapping this out.

John Furrier

>> You have a super personal computer, basically.

Rob Bruckner

>> Yeah, absolutely.

John Furrier

>> It's got the NVIDIA GB in there, 300.

Rob Bruckner

>> Right.

John Furrier

>> I mean, Gemma, I brought this up because Jeff Clarke said there's an inversion happening. I say flip the script. He says inversion. Super user, he has that 95%. It's funny, Dell made their business by making PCs and then servers during the Windtail days if you remember those days. So what's happening now is it's actually inverted. They're making the super servers, AKA AI factories, but I can see this coming down the other way because-

Rob Bruckner

>> Yeah, interesting observation. Yeah.

John Furrier

>> It's an inversion because that's if you look at the neoclouds, 100% GPU utilization, unheard of on the enterprise. People who cobble together their own AI factories see massive degradation in performance because the storage memory's not working. So they've got a very expensive GPU, but it's at 30, 50% utilization.

Rob Bruckner

>> Right.

John Furrier

>> Okay. So I think this template is pre-build AI factory. So are you thinking the same way? Because I'm thinking, okay, pre-build AI factory, which you're doing for the servers or the systems. What I saw on stage was essentially a pre-built AI factory node.

Rob Bruckner

>> You got it. That's exactly right. So I love your insight of how things have actually moved the opposite direction than what we had in the historical direction. And because of that, there's a lot of things that have been working quite well with Arthur Lewis and his infrastructure group, which includes just as you mentioned. You can't just take the different aspects and just put them in a rack and say go. There's a lot of parts about the solution itself, making sure the solution is out of the box working for people.

John Furrier

>> The Homebrew Computer Club is defunct because of a reason.

Rob Bruckner

>> Yes.

John Furrier

>> You manufacture them.

Rob Bruckner

>> That's exactly right. So we basically took the same concept to go, okay, these concepts also will work for us. You're in an era where not everything's pre-built. When you're on the edge of the innovation that's happening, you've got to now bring the technology to people so they can use it and also be involved with them to keep it updated and secure throughout the lifecycle of the product. So we're using a lot of what ISG had pioneered in our data center world now into that desktop agentic as well. And that includes, for example, the idea of forward deployed engineers, taking engineers out to key customers and helping them understand their business, what they want to resolve and come together as a fleet together across client and data center to meet the needs at a fleet level and see how these can work in a hybrid workflow, all with the deploy engineers working together for that hybrid workflow.

Gemma Allen

>> Well, Rob, it certainly seems as though you have joined this company at a very interesting, exciting and somewhat chaotic time at the industry broadly. It's been a busy few months so far, I'm sure. It feels like an awful lot has happened. What's ahead? What does the next 6 to 12 months look like for you? What are the priorities for the year out?

Rob Bruckner

>> Yeah. I mean, I would split that in kind of two arenas because there's still the run the business aspect of this while all the exciting technologies going on. So we're currently, within our client group, we're retooling the organization for more modern development. So modernization is a big thing for us at Dell. So every aspect of how we do our job, the workflows we have, how we automate the workflows and then bring agentic into the workflow. So retooling our workforce, as you heard on the stage today, is probably the top item for us because we. Believe if we can get there first and best, it'll be a differentiating asset for us to how to deliver those products in the market and provide that additional support to our customers for them as well. So that's the biggest thing for us. And it's really exciting because I'll give you some examples just in things like knowledge work, which is a lot of the work that we have in a commercial environment, it's a lot about getting insight out of large data sets.

Gemma Allen

>> Sure.

Rob Bruckner

>> Typically, you look at creating organizations, you say, let's get so many people to go do things. You multiply the people because you can get more input . And then you have to go organize those people to consolidate into insight. I can load those all into a context window of an AI LLM and generate insight for this, and I can continuously see the insights from that type of knowledge every day if I want to go see if my plan's still right or something has emerged for something new. The billion speed at which you can learn, have insight, and then direct that through a fast execution flow through the back end, with all the aspects of Dell procurement operations operating at that speed, this will be the biggest thing we're doing in the next 12 months as an organization to bring that asset forward. On the product side, the big area of the products obviously is going to be continuing to balance out the compute, the memory, the storage, the bandwidth, the AI compute, and bring us software layers in it to work easy for people to use and easy for people to service and update as new technologies emerge. These are the big hotspots for us right now.

John Furrier

>> All right.

Gemma Allen

>> And I'm going to finish on one thing you said there, the speed at which people can learn and the simplification opportunity that lies ahead because I think at the moment and right now, those are two very important messages. So Rob Bruckner, thank you so much for joining us on theCUBE.

Rob Bruckner

>> Thank you. It was a pleasure.

John Furrier

>> All right, thanks.

Gemma Allen

>> I'm Gemma Allen, joined by my co-host, John Furrier. We are here on the ground in Las Vegas covering Dell Technology. We're at 2026. Some great guests coming up, don't go anywhere.

Rob Bruckner, Dell Technologies

search

Gemma Allen

>> Welcome back to theCUBE. I'm Gemma Allen, joined by my co-host, John Furrier. We are here live on the ground in Las Vegas at Dell Technology. We're at 2026 where all the action is happening. We're going to have a conversation right now about the evolving role of the PC as the ultimate AI endpoint. Joining us is Rob Bruckner, President of Client Commercial Client Solutions at Dell Technologies. Welcome, Rob.

Rob Bruckner

>> Thank you. Good to be here.

John Furrier

>> Yeah, good ...

Gemma Allen

>> So you are, I think, half a year into this role at Dell. You have certainly picked the right moment to join, Rob. A lot happening. Maybe let's start with, this is your first Dell Technologies as a member of staff. I'm sure you've been here as a partner. What's exciting you? Where is your head at, Rob, in this crazy chaotic moment we're in?

Rob Bruckner

>> Yeah. A lot of days it's spinning as well. There's the excitement of AI and AI really entering the PC. You have a lot of additional challenges. We have the supply chain challenges that are around us. And that can, daily, be dynamic. So having the operational excellence of a procurement department just to get things done is an essential part of this. We have lots of innovation happening that takes AI to the next level in a PC. And so all the things that we're doing haven't been invented yet all the time. We're kind of putting together new things that are emerging, bringing together solutions so that our customers can use them in everyday work, and learn as we go and make it even better and better and better instead of waiting around for everybody else to have figured it out and now it just works. So we're going to get to that place eventually, but right now it's about putting that in action.

John Furrier

>> I think the big, I don't say secret or silent or private win, but to me the big win for Dell Technologies is the client side OpenClaw story here because it really shows the future of the relationship that the human's going to have to their interface. We hear about headless systems, but at the end of the day, I'm going to have a device. I'll either talk to it on my body or it's going to be in my hand or a PC. This OpenClaw revolution speaks to this new generation of true personal computing where it's not a developer, it could be just a developer and like a business person running agents.

Rob Bruckner

>> That's right.

John Furrier

>> So the AI builder persona isn't like your classic developer. It could be anybody if they have the right tool. So to me, I think that was a really big win for Dell, but also it's going to probably make more sales because you got to run AI on these OpenClaws and NemoClaws, and everything's going to be happening here.

Rob Bruckner

>> Exactly right. Exactly right. And OpenClaw was to ... I mean, Jensen said at GTC as well, it was the agentic starting. That was like the big moment where agentic AI was really reaching out. People started to realize the power of agentic for basically orchestration of workflows that rethink exactly how you use the computer every day. Now, what we added to Dell was if you worked at OpenClaw, there's two things about OpenClaw that are still emerging. One is to install, configure, and bring OpenClaw, you'll raise your claw basically. Does take some amount of help for a lot of people to get active in that. So we want to make it easy for people to bring that forward in a PC and we want to make sure that it has a wrapper around it that brings the security manageability as we try to bring that forward even into commercial space. And that OpenClaw experience is a composable agentic experience. You can have the ability to compose your workflow versus it being given to you in something that's kind of canned, which limits your ability to really invent.

John Furrier

>> And the operating model is a big theme here. Gemma, you worked at Microsoft. You know Windows.

Gemma Allen

>> I do.

John Furrier

>> Yeah. Windows is on my machine right here. Dell bundles it in. I don't have to install it, get a license. It's already in there. But now with this new era, I can see kind of similarities by having everything on the machine. How do you think about that? Because it's not the same as Windows.

Gemma Allen

>> I think we also though Windows has been on such a journey too. When I was at Microsoft in the Windows VG back in like 2010, 2012, there was this whole debate, is the PC dead, right? We're now at a point where the PC is suddenly like the end point of opportunity for everything. And I think how folks reimagine what a PC is also needs somewhat of a cultural reset. How do you think about that even from the perspective of conversations that you have with clients and customers in the field? Because this means something different than it did 10 years ago, right?

Rob Bruckner

>> In a big way. And the first step is talking about AI PCs in a different way. I think the industry kind of went out pretty quickly to try to get out there with a neural processing unit based architecture originally. It did make the PC a little more personal because it was able to see how you can interface with the device in a little bit better way. It just didn't have the power to bring forward a really heavier workload that brings more value to the PC. So first thing we're doing is overcoming the starting place of what is an AI PC into, Is it a AI PC still? What's the terminology mean now? I know this is kind of a trivial comment to say it's like an AI PC, but it's something that we have to really face with customers right now. Redefine now what the value brings to the PC, and the value starts to emerge in two main areas as you're able to run things more local. First is token expense. So as we all know, agents are very chatty. And as you're trying to put the agents into some level of freedom but also control, you're starting to see a lot of loops that are emerging that are generating lots of tokens. The inference needed to bring values does not require a frontier level inference. It can actually be done with a reasonable size model. So you can start to have a more control of your token budgets or be a little bit more dynamic about when you want to use a frontier token. Second is sovereignty. What do I want to do about my own privacy and my own data, and how does that mean to me and making sure I can contain things that I want to keep private and secure in a sovereign way versus things that I'm fine being more open with? So the world really is a hybrid world for how you do all of those things.

Gemma Allen

>> Let's stay on cost for a second. Jon Siegel spoke yesterday and he mentioned a scenario whereby a developer burned through $3,400 in token costs in one day. I mean, that raised a lot of questions about the developer for one point. But talk about how this new opportunity for Dell and what you guys are bringing to the market solves for that. Solves for how total cost ownership all of those challenges right now around, like we hear about token lead rewards at enterprise. I don't know if you've heard this term. Again, it's crazy. It seems insane, but it is happening. So how does Dell think about that and how does it approach that in your kind of customer solution model, when you're talking to clients in the field?

Rob Bruckner

>> Yeah. It's a fascinating ... Well, this is a fascinating time because this type of topic, we'd be talking about it a year ago.

Gemma Allen

>> We have.

Rob Bruckner

>> So think about what we're seeing right now in our own usage with our own software development community is the actual workflow for optimizing and redoing your code or securing your code, that workflow's being created. So the developer that might use lots of tokens right away, it's because they're optimizing their workflow and they actually need a playground. It's essential for them. Otherwise, they're going to hold their innovation instincts because they're worried about expanding on their token budget. That's why some of the local AI brings that value to where you can still invent your workflow and then as you need to go put it into something a little bit more on the prem or even frontier, you now have a good clean lockdown. If you don't let people innovate when the technology is changing so quickly, then you're going to have a problem where they're afraid to use their tokens. That's the last thing you want your developers to do, right?

John Furrier

>> And the tokens are great because you can do them on your PC. I want to get your vision on this because what's happening is clear. The device is connected. Agent's going to be working on our behalf. The cloud is shown away. I can store stuff on the cloud and then tap it from my machine to go get it, retrieve it.

Rob Bruckner

>> Right.

John Furrier

>> But if I have an agentic force working with me, I could have it all done here. I think Michael Dell said and Jeff Clarke said, "This is the company if you're an entrepreneur." Your PC or box on stage, your server he showed, you could literally run everything from that if it's got connection to something. It could be cloud or another data center. So you start to see this kind of, I won't use the word client server because that's an old term, but you got kind of this power machine that's going to have a lot more backend connected to it. What's your thoughts around, that's my weird view on this, but like that's the way I would see that because if I have agents, I want to have instant uploads coming to me. I got to connect to maybe another system, another data center, service, a new cloud.

Rob Bruckner

>> It's an evolving area because it's evolving at the speed at which model sizes and intelligence can arrive locally on a machine. So this machine, your XPS 14 is example, it has good compute in it, but for real AI agentic workloads, you really need a good like 32, 64 gigabytes of memory. This allows you to put a bigger model in there. You got to have memory band to get the token rates up. It's not just human tokens, it's agent to agent tokens, which can be much 2 to 3X faster than what you do for human token, for example. And you need to have the AI compute built in. So the type of machine you need, people are buying up into the stack, realizing that what they're buying is for the future. Even if they may not be fully running workloads that are local on agentic, they know they're coming, they know they need to be purchasing and getting ready for that. Still, I think you're going to see a very hybrid world because you're going to see different paces of innovation available to you as you look from the frontier to time it takes from frontier to move on prem in a commercial environment, or you might take a parallel to commercial environment. Maybe in the consumer environment, you get VSPs and other type of businesses that are arriving to give you some of those intermediate services out from the frontier and then to local. You're going to see all three domains are going to play equally important but at different levels of capability and innovation available.

John Furrier

>> So do you think about this differently now, your view of the future? Local, just privacy, you mentioned some of those things about sovereignty. Is this now looked at differently or is it still the same thinking around ...

Rob Bruckner

>> I think it's similar, but I think what's different is there's some additional complexity about when you're running your agent, how much intelligence does it need to take the action that you've allowed it to do. Or an agent that's monitoring another agent through the flow, that intelligence parameter then dictates the size of inference you need to go do. And if it's a very complex inference that you have very sophisticated flow, you're going to have to get even more AI capability off the device. But for a lot of just keeping the type of things you and I might do as a consumer, it has the intelligence with 30, 70 billion parameter models to do much of that workflow. Why not do it there? And actually, when you look at possible congestion that's going to happen with all the tokens that are going to be distributed across the networks, doing more things local where you can actually do them makes a lot of sense in putting everything out off the device.

John Furrier

>> First describe what the news was specifically and then how you see them configuring it. Because again, I'm thinking about me. I'm like, "I want to do more. I want more action. I had a PC, I bought a printer. I got a bunch of storage." I can see the, I won't say scroll, but I can see the expansion. So what is the desk side agentic solution? How does that play out?

Rob Bruckner

>> So one of the insights we had as we were watching the agentic workflows start at the frontier actually and then start to move on prem, which is coming for all the enterprise folks, was that these are also things as we've been talking about, you can run locally. And then so we looked at our device stack and what do we have that's available that has the necessary hardware, meaning support for memory size, memory bandwidth for token rates, and AI processing capability keep pace. And all these things have to come in synchronization. It doesn't help you to have a big memory size with no memory bandwidth. All these things have to be balanced. So as we started to design and provision these machines, precision line seems like the right spot for us. This is a place where it's already a big compute. It already has GPUs everywhere. So we took both the mobile and fixed and started to map out now all the capabilities for what these things can do with agentic workflows. All the way from precision in the mobile line with some RTX GPUs into the fixed line where you can add multiple into the fixed slots, the GB10 and the GB300. And these have, if you look at the scale of compute, it's a pretty nice coverage from what you could do to start out in more of the precision, say seven line, for example, to the GP300. This dynamic range has never been seen before in a precision lineup and that allows you to, as I mentioned, it allows you to select where you want to be for the inferencing capability that your agentic flow can have. If you have a more sophisticated flow, if you have a more sophisticated need for intelligence when you're making decisions for the agent, you've got to move up the stack for that intelligence that you're buying. So that's the way we've been mapping this out.

John Furrier

>> You have a super personal computer, basically.

Rob Bruckner

>> Yeah, absolutely.

John Furrier

>> It's got the NVIDIA GB in there, 300.

Rob Bruckner

>> Right.

John Furrier

>> I mean, Gemma, I brought this up because Jeff Clarke said there's an inversion happening. I say flip the script. He says inversion. Super user, he has that 95%. It's funny, Dell made their business by making PCs and then servers during the Windtail days if you remember those days. So what's happening now is it's actually inverted. They're making the super servers, AKA AI factories, but I can see this coming down the other way because-

Rob Bruckner

>> Yeah, interesting observation. Yeah.

John Furrier

>> It's an inversion because that's if you look at the neoclouds, 100% GPU utilization, unheard of on the enterprise. People who cobble together their own AI factories see massive degradation in performance because the storage memory's not working. So they've got a very expensive GPU, but it's at 30, 50% utilization.

Rob Bruckner

>> Right.

John Furrier

>> Okay. So I think this template is pre-build AI factory. So are you thinking the same way? Because I'm thinking, okay, pre-build AI factory, which you're doing for the servers or the systems. What I saw on stage was essentially a pre-built AI factory node.

Rob Bruckner

>> You got it. That's exactly right. So I love your insight of how things have actually moved the opposite direction than what we had in the historical direction. And because of that, there's a lot of things that have been working quite well with Arthur Lewis and his infrastructure group, which includes just as you mentioned. You can't just take the different aspects and just put them in a rack and say go. There's a lot of parts about the solution itself, making sure the solution is out of the box working for people.

John Furrier

>> The Homebrew Computer Club is defunct because of a reason.

Rob Bruckner

>> Yes.

John Furrier

>> You manufacture them.

Rob Bruckner

>> That's exactly right. So we basically took the same concept to go, okay, these concepts also will work for us. You're in an era where not everything's pre-built. When you're on the edge of the innovation that's happening, you've got to now bring the technology to people so they can use it and also be involved with them to keep it updated and secure throughout the lifecycle of the product. So we're using a lot of what ISG had pioneered in our data center world now into that desktop agentic as well. And that includes, for example, the idea of forward deployed engineers, taking engineers out to key customers and helping them understand their business, what they want to resolve and come together as a fleet together across client and data center to meet the needs at a fleet level and see how these can work in a hybrid workflow, all with the deploy engineers working together for that hybrid workflow.

Gemma Allen

>> Well, Rob, it certainly seems as though you have joined this company at a very interesting, exciting and somewhat chaotic time at the industry broadly. It's been a busy few months so far, I'm sure. It feels like an awful lot has happened. What's ahead? What does the next 6 to 12 months look like for you? What are the priorities for the year out?

Rob Bruckner

>> Yeah. I mean, I would split that in kind of two arenas because there's still the run the business aspect of this while all the exciting technologies going on. So we're currently, within our client group, we're retooling the organization for more modern development. So modernization is a big thing for us at Dell. So every aspect of how we do our job, the workflows we have, how we automate the workflows and then bring agentic into the workflow. So retooling our workforce, as you heard on the stage today, is probably the top item for us because we. Believe if we can get there first and best, it'll be a differentiating asset for us to how to deliver those products in the market and provide that additional support to our customers for them as well. So that's the biggest thing for us. And it's really exciting because I'll give you some examples just in things like knowledge work, which is a lot of the work that we have in a commercial environment, it's a lot about getting insight out of large data sets.

Gemma Allen

>> Sure.

Rob Bruckner

>> Typically, you look at creating organizations, you say, let's get so many people to go do things. You multiply the people because you can get more input . And then you have to go organize those people to consolidate into insight. I can load those all into a context window of an AI LLM and generate insight for this, and I can continuously see the insights from that type of knowledge every day if I want to go see if my plan's still right or something has emerged for something new. The billion speed at which you can learn, have insight, and then direct that through a fast execution flow through the back end, with all the aspects of Dell procurement operations operating at that speed, this will be the biggest thing we're doing in the next 12 months as an organization to bring that asset forward. On the product side, the big area of the products obviously is going to be continuing to balance out the compute, the memory, the storage, the bandwidth, the AI compute, and bring us software layers in it to work easy for people to use and easy for people to service and update as new technologies emerge. These are the big hotspots for us right now.

John Furrier

>> All right.

Gemma Allen

>> And I'm going to finish on one thing you said there, the speed at which people can learn and the simplification opportunity that lies ahead because I think at the moment and right now, those are two very important messages. So Rob Bruckner, thank you so much for joining us on theCUBE.

Rob Bruckner

>> Thank you. It was a pleasure.

John Furrier

>> All right, thanks.

Gemma Allen

>> I'm Gemma Allen, joined by my co-host, John Furrier. We are here on the ground in Las Vegas covering Dell Technology. We're at 2026. Some great guests coming up, don't go anywhere.