‘Our importance to the partner community and, frankly, the partner community’s importance to us has never been bigger,’ Lewis said. ‘The opportunities that partners have to develop services on top of the [Dell] AI Factory is an innovator’s dream.’
“We’ve tested it,” he said. “We’ve deployed it. It is a very, very powerful network for AI. When you are deploying these very expensive GPUs if you are stalling the GPUs because of the network, you are wasting a lot of money. You will not idle the GPUs with a SONiC [Software for Open Networking in the Cloud] operating system deployment.”
Dell is supporting SONiC across its PowerSwitch family of networking switches and across the portfolio, including Nvidia’s Spectrum 6 ultra-high-speed Ethernet switch and Broadcom’s Tomahawk 6, which is touted as the industry’s first 102.4-Terabit- per-second Ethernet switch chip.
Lewis said Dell has an “incredibly exciting” front-end and back-end network opportunity for partners given that AI requires a “different set of requirements” for the communications superhighway.
“It is high bandwidth, low latency, but it has to be an uncluttered network, and for the same reason that Linux became the standard operating system for servers 20 years ago I believe that SONiC is going to become that operating system for the network,” he said. “It’s open. It’s multi-vendor and continuously improved by a global community.”
Lewis said history is set to repeat itself with the rise of SONiC in the AI networking era. “Is the reason Linux became the standard operating system for servers a different set of circumstances than exists today?” asked Lewis. “I would argue that it is not. It’s exactly the same thing. So it’s not about envisioning something that hasn’t happened. It’s about replicating something that happened 20 years ago. That gives us a lot of belief about this.”
Lewis said partners now see Dell not only as an AI leader, but as a “key player” in networking. “Our importance to the partner community and, frankly, the partner community’s importance to us has never been bigger.” he said. “The opportunities that partners have to develop services on top of the [Dell] AI Factory is an innovator’s dream. It’s an entrepreneur’s dream. There are so many challenges and problems that customers are going to have to tackle. This is going to be a huge boon to them.”
Lewis, who is widely credited with delivering an AI-innovation- packed ISG portfolio that powered 40 percent sales growth to $60.8 billion in Dell’s last fiscal year, said the company’s “very strong fundamental understanding” of AI and what it means to customers of all sizes have been key to Dell’s success. “We’ve had a clear vision of that for the last five years,” he said.
“There’s probably a couple of different secret sauces, but to me what I’m extremely excited about is we have a very strong vision of what the future looks like,” he said. “That focus drives incredible innovation because we’re not zigging and zagging, trying to get a direction. We are marching in a straight line. We know exactly what has to happen. We know exactly what we need to do, and we have a relentless passion for getting there incredibly fast.”
Lewis, a 20-year Dell veteran, said Dell’s ability to deliver solutions that are “extremely relevant” in the agentic AI era sets the company apart from competitors.
“Those solutions can only be crafted by a company that has the type of portfolio that we have,” he said. “If you focus just on compute or just on network or just on storage or just on data management, you are not able to craft the type of solution that we can craft. We engineer it all as one thing.”
Below is more of CRN’s conversation with Lewis at Dell Technologies World.
What is the key to the success of the Dell Infrastructure Solutions Group portfolio?
Let’s start with the premise that we have a winning portfolio, and then let’s ask the question: ‘Why do we have a winning portfolio?’
It goes back to our very strong fundamental understanding of artificial intelligence and the workflow and what it means to customers of all sizes, what it means to neocloud, what it means to sovereign [clouds], what it means to enterprises, what it means to medium-size businesses, what it means to small businesses. We’ve had a clear vision of that for the last five years.
Do we have a broad portfolio? Actually, when I take a look at the individual products within the portfolio, we’ve actually streamlined the portfolio. But what we’ve done is we’ve gone deeper with the products that we have to build solutions that matter for customers.
Not only did we talk about individual products [here at Dell Technologies World], which is great, but to me more exciting is we talked about the AI Data Platform, Dell Private Cloud, Dell AI Factory, and why all of these things are meaningful for customers in the world of AI.
[Dell Vice Chairman and COO] Jeff [Clarke] started down this path three years ago internally at Dell. That was very prescient of him to do so at the time. He understood that you can use AI to improve the work that you do today and the manner in which you do it, and you will get a certain amount of productivity. But if you are bold, you will remap everything.
If you think about all of our processes, they were all mapped to how humans do work right now. Now you have to remap everything to how humans manage, supervise and direct work that’s being done by the machine. That is what unlocks that 10X, 20X and 30X productivity. So you go from 10 percent, 15 percent or 20 percent to maybe 10X, 15X or 20X level of productivity.
We have taken that to heart in my development organization where we are moving to spec-driven development, moving away from agile [development]. Our ability to deliver features to customers is going to accelerate. I kind of joke that the innovation is happening so fast now we could probably do a Dell Technologies World every six months, and it would be material updates to what we just talked about.
What is the Dell secret sauce in terms of accelerating feature and product development, and how will that impact customers and partners in the agentic AI era?
There’s probably a couple of different secret sauces, but to me what I’m extremely excited about is we have a very strong vision of what the future looks like. That focus drives incredible innovation because we’re not zigging and zagging, trying to get a direction. We are marching in a straight line. We know exactly what has to happen. We know exactly what we need to do, and we have a relentless passion for getting there incredibly fast. The opportunity is significant to talk to customers and help them solve problems.
If you take a look at the data center over the last 20 years, there has been an accumulation of workloads that are very specific with infrastructure and data. Customers may have 15, 20 or 25 of these different workloads. That may have worked over the past 20 or 25 years, but now with this agentic technology all of these systems have to be connected with data silos collapsing.
I don’t believe that there will be any more dark data, any more cold data. All of this data will be in constant motion, feeding AI agents that proliferate across a dataset. So for customers that’s incredibly daunting as they look at how do they turn their brown field into essentially an agentic AI factory.
Our ability to help them navigate this incredible transformation is just incredibly exciting for us. We have a supreme passion around it. We feel like we can help. That focus and passion is what I believe is really driving the team.
What is the Dell agentic AI vision as you look into the future?
At the macro level it’s agentic systems that are all interconnected with common data sources, where the data is in constant motion and constant circulation. It’s not just about unstructured data. It’s about structured data, semi-structured data, multimodal data.
All of this data has to be aggregated, fed into a vector graph, fed into a knowledge base, turned into something that the AI can actually interact with, and then the agents are going to consume that data like never before.
Now you’re going to start talking about things like synthetic data generation because you want to continue to drive token efficiency and optimal results. The opportunity to innovate in this space is massive.
I kind of joke with the team that from a customer’s perspective it can be paralyzing that you have a technology that can be deployed in so many ways because you can get stuck on choosing where do you start and how do you do it. There’s any number of different ways. You could go into a five-year planning process where you try to go through every permutation of every type of use case and every type of architecture.
Our job is to go in and have a very confident conversation with customers, get them started and go. We’ve learned this in spec- driven development. Repetition matters here. You’ve got to get your reps in. It’s kind of like an athlete. You’ve got to practice.
You can just watch the film, but until you get out on the field and you do the reps you’re not really training your body or training your muscle. You have got to get in there and do the reps. You’ve got to get going. You can’t just watch the film. You have got to get on the practice field and do your reps.
What are the products Dell announced here at Dell Technologies World that are going to propel customers forward into this agentic AI era where agents are working together?
I think you can map the announcements to Jeff’s five imperatives during his keynote address. Those imperatives were so clear and actionable.
No. 1: You’ve got to build a strong data foundation. How do you go do that? You do that with our AI data platform.
No. 2: You’ve got to build distributed AI, both for training and inference, but also for the edge, the core and the cloud. You have Dell AI Factory and the Dell Private Cloud for that.
No. 3: You’ve got to build secure autonomous systems. You do that with PowerProtect Data Manager and PowerProtect Data Domain, which, by the way, is also a part of the [Dell DB] Boost ecosystem, so it works across many different types of backup software.
No 4: You’ve got to build a full-stack solution. We’ve done that with [Dell] AI Factory [with Nvidia].
No 5: You’ve got to build a token generation engine that’s ready for the world of agentic technology.
I love the individual products. I absolutely love them. But to me the solutions message is what’s different about Dell now.
We have solutions that are extremely relevant in the world of agentic technology. Those solutions can only be crafted by a company that has the type of portfolio that we have.
If you focus just on compute or just on network or just on storage or just on data management, you are not able to craft the type of solution that we can craft. We engineer it all as one thing.
So if you’re going to go buy a PC, you’re not going to go buy a license for Windows from Microsoft and a keyboard from someone else. You’re going to buy a finished product. The same thing is happening here. You are going to buy a solution. That’s incredibly exciting to me.
What kind of gains are partners and customers going to get from buying the complete solution starting with storage?
Let’s talk about data first. Even though there are many forecasts about the explosion of data, I have not seen one that properly contextualizes the amount of data that’s going to be generated. Every single transaction that an agent performs has to be recorded, transcribed and cataloged in some form or fashion. That’s going to drive massive data requirements. In addition, that data is going to have to feed the AI. That’s going to require a lot of speed.
So, whether it’s sort of your traditional storage or this new tier of memory for disaggregated inference, we kind of cover it all [at Dell]. We have the underlying file, block and object [storage]. We have Lightning, which is a parallel file system but also doubles as your context engine and storage access engine for disaggregated inference.
When you say ‘storage,’ it sort of implies a place where you warehouse data. That is no longer the case. Now storage is where data lives and where it’s managed. It’s the fuel that is feeding your AI.
So when you think about PowerScale, ObjectScale, Lightning or PowerFlex or some of the underlying platforms like PowerStore, PowerProtect or Data Domain, we have the universe where the data is going to live and feed the AI at speed. Only a company that has got the portfolio of our size can actually do the kinds of things that we can do. We were built for this moment.
How big a differentiator is the Lightning file system?
It is massive. As a stand-alone parallel file system for those tier- zero workloads, it sits in a category by itself at 150 Gigabytes per second of [read] throughput per rack, but now that it’s doubling as your disaggregated inference to expand the context window as you split the prefill and decode stage of inference, there’s massive value in there.
These are emerging technologies. Customers don’t understand them very well. What is incredibly exciting for me is we are moving way up the stack in the conversation.
Traditionally as an OEM the customer kind of knew what they wanted; they would jot it down in an RFP [request for proposal]. They would send it over. We would have a conversation. We would respond, etc.
Now we are way up the stack with customers asking us, ‘Dell, help me understand the use case that you think I should go after;’ Hey Dell, help me understand the software stack and model selection; Hey Dell, help me understand my data strategy.’ And then once they kind of get through those three questions they ask, ‘Hey Dell, help me understand the underlying architecture.’ Then you get to the infrastructure conversation.
So we’ve moved way up as this trusted adviser to customers, which is a very inspiring place to be to have that level of trust with so many of the world’s most important, largest customers, as well as small and medium businesses. Everybody is coming to us and asking us, ‘Hey, how should we think about this? What should we do?’
How does the utterly parabolic demand and the 20 to 30 times gains in productivity coming from agentic AI change the dynamic on how you develop products for the future?
It doesn’t change our design philosophy at all. We have a pretty strong vision. We know what we want. We know you know how to get there.
What you’re seeing is a fundamental reset of the market. So you have gone from this magical world where you have a tool that you can ask a question and it gives you a pretty good answer to a world where you could ask the tool the same question and it can reason in an auto-regressive fashion to give you an even better answer.
Now that same tool that was capable of reasoning is now capable of executing functions. From the start to where we are now is just three years—36 months. The advancements have just been insane.
Now for the agents to perform these tasks they need something to be able to do the task. That something is a computer. The good thing is we provide those computers. Those computers are called servers, but those computers are also called GB10s. They are also called GB300s. You have sort of that spectrum from [Dell DGX] Spark to [agentic] station to NVL 72 to use the Nvidia terminology. The scale of the growth doesn’t affect our design philosophy because we know exactly what customers need over the next several years.
How big an opportunity is there for Dell to make gains in the networking market with the complete solution you are bringing to the market? Why choose Dell for networking in the AI era?
This one is incredibly exciting to me. I said this on stage: Artificial intelligence provides a different set of requirements for the communications superhighway.
It is high bandwidth, low latency, but it has to be an uncluttered network, and for the same reason that Linux became the standard operating system for servers 20 years ago I believe that SONiC is going to become that operating system for the network.
It’s open. It’s multi-vendor and continuously improved by a global community. We’ve enabled SONiC across the entirety of our portfolio. We [have] expanded the portfolio to include Spectrum 6 and Tomahawk 6, Ethernet, as well as Nvidia’s Quantum X800 Infiniband solution. With the exception of Infiniband we’ve enabled SONiC across the entirety of the portfolio, and where we’re deploying it, it is working exceptionally well. So I think this is a front-end and back-end network opportunity.
How important is SONiC to this agentic AI story, and why is it such a rich environment for a new era of networking?
You don’t have to do a bunch of mental gymnastics or envision something that hasn’t happened. All you have to do is say, ‘Is the reason Linux became the standard operating system for servers a different set of circumstances than exists today?’ I would argue that it is not. It’s exactly the same thing. So it’s not about envisioning something that hasn’t happened. It’s about replicating something that happened 20 years ago. That gives us a lot of belief about this.
Plus, we’ve seen it. We’ve tested it. We’ve deployed it. It is a very, very powerful network for AI. When you are deploying these very expensive GPUs, if you are stalling the GPUs because of the network you are wasting a lot of money. You will not idle the GPUs with a SONiC operating system deployment.
So how big an advantage is that for Dell in networking in the agentic AI era?
It’s a huge advantage. Maybe it’s better to be lucky than good, but sometimes hard work and luck kind of come together. SONiC is exactly the operating system that AI requires, and to me what happened with Linux 20 years ago is exactly the same thing that’s happening now.
We’re not envisioning something that hasn’t happened yet. We’re basically saying, ‘Hey, that thing that happened before, we believe it’s going to happen again for these reasons.’
What’s the message to partners about the networking opportunity with Dell and SONiC?
Well, they see us as a key player in AI but now as a key player in networking. Our importance to the partner community and, frankly, the partner community’s importance to us has never been bigger. The opportunities that partners have to develop services on top of the AI factory is an innovator’s dream. It’s an entrepreneur’s dream. There are so many challenges and problems that customers are going to have to tackle. This is going to be a huge boon to them.
How does a complete Dell AI platform with compute, storage and networking provide an advantage versus competitors?
I think it’s very important to understand that all the announcements and innovation that happened this week [weren’t] decided the week before. These are decisions that we made two and three years ago. Why did we make those decisions? We made those decisions because we understand the technology. We meet with customers and we understand their problems.
It is very important to me that when we innovate, we understand the problem that we’re trying to solve. It’s not just the next turn of the crank of a feature that people think was important yesterday therefore it must be important today. So if it did X yesterday it should do Y tomorrow.
The question we always ask is, ‘What problem are we trying to solve for the customer?’ We’re not just taking a laundry list of features that customers ask us for. We’re asking the question, ‘What problem are you trying to solve?’ That’s a pretty interesting question because then the customer kind of comes back and goes, ‘Oh, that’s a good question. I wanted this feature, but I didn’t really take the time to understand to ask why I wanted this feature. What problem am I trying to solve?’
That’s what allows you to get to true innovation. It’s a series of decisions that you make over years that are very, very hard to replicate. If somebody wanted to come in and copy the portfolio, it would take them time. By the time they were done, we would be past them.
You have to have a deep understanding of the technology and a very deep understanding of the customer set in order to be able to solve problems in a meaningful way that allows you to drive the innovation that resonates with customers. Dell Technologies World is three or four days. It was three or four years in the making.
What are you most proud of when you look at the breadth and depth of the ISG product portfolio?
I would take a step back and say I’m incredibly proud of the relevance of the company when it comes to artificial intelligence. Without regard to whether it’s the PC, the monitor, the server, the network, the storage, it’s really hard to have a conversation around artificial intelligence without talking about Dell Technologies. I’m incredibly proud about that.
I’m incredibly proud that we are seen as an innovation leader in what is a defining technology for all of humanity. It’s an incredibly powerful place to be in. We’ve worked very, very hard to get here. And then underneath the covers, I’m incredibly proud of the innovation of the team, the passion and the grit, the striving for innovation, the recognition that they’re getting. The output is the products. So it all kind of nicely flows together.
With the market moving so fast, what is the vision for the next three years?
I would say we laid out a pretty strong vision on the five imperatives that Jeff talked about. I would say we are going to innovate a lot on those five imperatives, but what’s important here is that the technology is going to change in ways that maybe we can’t envision.
So we’ve got to make sure that we are very flexible in how we innovate to make sure that we are capturing all of the great technology that’s out there.
The infrastructure is now the long pole in the tent. That has not been the case in a very long time, which is why there’s this race to build out infrastructure to keep up with the capabilities of the software.
I love our hand. We are in the pole position in the defining technology of, in my opinion, all of humanity. We are a trusted adviser to all kinds of companies: neoclouds, sovereign entities, large enterprises, small enterprises, small business, medium business. It’s a really good place to be in, and it feels really good.
Is there any way to quantify how big an advantage Dell has because of the way it does development?
That is a little difficult to quantify, but I would just kind of go back to when you take a look at our portfolio the ability to engineer the compute, the network, the storage, the data management, the software, the ecosystem, the services as one thing, that’s incredibly powerful. Our understanding fundamentally of how the technology is going to be used in a way that maybe the customer doesn’t understand it is an incredible advantage. These are built on top of the company advantages that we have like the largest sales force, the largest partner ecosystem, the best global operations team, the best services team, the greatest portfolio in the world. All of these advantages start to stack up, and they become meaningful.
