Good afternoon, everyone. Thank you again for joining the UBS Global Tech Conference. I am David Vogt. I'm the hardware networking analyst here. We're excited to have with us today from Dell Technologies, Arthur Lewis, President of ISG. So, Arthur, welcome.
Thank you for having me.
Before we get going, Dell has asked me to read their Safe Harbor Statement. Statements in this presentation that relate to future results and events are forward-looking statements and are based on Dell Technologies' current expectations. In some cases, you can identify these statements by such forward-looking words as anticipate, believe, could, estimate, expect, intend, confidence, made plan, potential, should, will, and would, or similar expressions. Actual results and events in future periods may differ materially from those expressed or implied by these forward-looking statements because of a number of risks, uncertainties, and other factors, including those discussed in Dell Technologies' periodic reports filed with the SEC. Dell Technologies assumes no obligation to update its forward-looking statements. Great. All right, Zack, I took your job. Arthur, thank you again for joining. For those that don't know, Arthur runs the ISG division.
And so, maybe we can start with, I think, for some people that may not have met you in the past, kind of what falls under your purview, kind of your experience here, and how should we think about the ISG business under your watch going forward, and then we can dig into some of the details.
Yeah. And again, thank you for having me here. Good afternoon, everybody. Hope you're enjoying the conference. Just by way of history, I joined Dell in 2006. I came by way of acquisition, a company called Alienware. From 2006 to 2019, I held various leadership positions, product leadership positions, go-to-market leadership positions, both in the client as well as the ISG space. In 2019, I assumed leadership of our server business, and then in September of 2023, I assumed leadership of the ISG business. And in that capacity, responsible for overall ISG strategy, our product portfolio, which includes servers, networking, storage, and all the solutions that comprise those products, the engineering of those products, the CTO, as well as the general field sales enablement.
Perfect. All right. AI servers. I think what everyone's here to hear. So in Q3, obviously, you had strong order intake, strong backlog. Revenue, I think, came in line with sort of the expectations that you had going into the quarter. Can we kind of talk about what you saw in the quarter, some of the main drivers that drove that backlog and that order intake in the quarter, and then we can go from there?
Yeah. So very pleased with the progress that we're making in AI-optimized servers. We saw $3.6 billion of revenue growth from an orders perspective, which was up 11% sequentially, shipments of $2.9 billion. So from a last four-quarter basis, we've taken in orders of $11.5 billion and shipped $8.5 billion. We saw a backlog of around $4.5 billion. But importantly, very excited about the robust demand that we see in the marketplace. So when we take a look at the next five-quarter pipeline, exiting Q1, we said it was multiples of our backlog. Exiting Q2, we said it was several multiples of our backlog. And exiting Q3, we continue to say that it's several multiples of our backlog and that it grew 50% sequentially quarter- over- quarter. Right? So continue to see very robust, strong demand.
We have visibility to all of the significant opportunities out there across tier two CSPs, enterprise, and sovereign opportunities, and I think enterprise was also a pretty nice, we saw the next five-quarter pipeline grow 55% sequentially, and we now have 2,000 unique buyers, so we have a seat at the table with the tier two CSPs, with the enterprise, with sovereign, and so feel really good about the progress that we're making and the future in front of us.
So just maybe you mentioned you have over 2,000 discrete partners, potential partners. What is driving the differentiation in the marketplace? Right? So you have a variety of different players, whether it's OEM, ODM. Maybe just talk a little bit about what Dell brings to bear and what makes your solution attractive and has enabled you to win pretty significant orders over the last four quarters, as well as have this several multiples of pipeline on a trailing five-quarter basis.
Yeah. So we've built and honed what I believe is a very strong and special value proposition, and it starts with the breadth of the portfolio. It cuts across server, networking, and storage. So you think about what we're delivering on the compute side of the house. We started with the 9680, and the design point around the 9680 was to be the densest, most energy-efficient product that supported diversity of silicon and diversity of networking. So with the 9680, you could do an NVIDIA GPU, AMD GPU, an Intel GPU. It would support Ethernet. It would support InfiniBand. The predecessor to that product is the 9680L, and that's a direct-to-chip liquid-cooled 4U product. So 33% denser than its predecessor, two and a half times more energy efficient with the same diversity of silicon and diversity of networking.
We've also introduced two new products directed at the enterprise, primarily for fine-tuning and inferencing, the XE7740 and XE7745, which is a 4U chassis, which will enable eight double-wide or 16 single-wide cards for customers that are looking to do more TCO in more of the fine-tuning and inferencing use cases. And not only can we sell those as standalone servers, but we sell them in rack scale up to 72 to 96, up to 144 GPUs across the rack. We've seen the innovation that we've delivered in PowerScale with significant improvements in OneFS in both streaming, read, and write capabilities, and the progress that we're making on Project Lightning, which is our parallel file system. So you take a look at the breadth of the portfolio, the innovation.
We were first to market to deliver NVL 72, even ahead of the hyperscalers, which was an incredible feat by the engineering team, and we're incredibly proud of the work that we've done. You kind of consider the breadth of the portfolio, the pace of innovation, getting to liquid cooling, getting to the first GB200 rack in the marketplace, doing what we're doing with PowerScale, the Dell Data Lakehouse, the parallel file system. You couple that, David, with the ecosystem, and not just the traditional ecosystem of the NVIDIA, Intel, and the Broadcoms of the world, but the unique one-of-a-kind partnerships that we have with companies like Hugging Face and Meta, with Llama, really helping enterprise think through their deployment models.
And you couple those with these services capabilities, and not just the support services that Dell has a rich heritage in, but really the deployment and integration services. This is what allows us to deploy 50,000 GPUs in six weeks. Right? That's another incredible feat and capability that we've built over the last 18 months. And you take those three, and then you think about the AI factory solutions that we build on top of that. We now have 40-plus solutions, again, that cross the portfolio against a myriad of use cases. And then we have our pursuits team. So a lot of these conversations are engineer-to-engineer-led conversations. And so we've built out capabilities within our CTO organization, which would include data center architect, system architect, storage architect, network architect, and a thermal architect to ensure that we're driving the level of customization and meeting the requirements of these customers.
We've been able to stitch this value proposition and hone it in over the last 18 months, and it's resonating extremely well. The opportunity to engineer solutions end-to-end on one team and optimize it for the customer, again, is a very strong value proposition, and we see it in the tier two CSPs, and we see it in the enterprise alike.
So can I ask a follow-up to that? So obviously, the end-to-end solution, I think, clearly allows you to win business. It's resonating. To some degree, though, you're dependent upon GPU availability. Correct? And you're also dependent upon when tier two and enterprise customers want to take deployment. So how do we think about this pipeline metric, your backlog number that you've publicly disclosed versus revenue rack? So from initial conversation to deployment revenue rack, I know there's no standard answer, but just maybe give us a sense for what those conversations look like and how long that generally takes versus historically, right? Servers were a book and ship business. Right? It wasn't that complicated of a sale, I would imagine, relative to what we're seeing today. So maybe help us out and understand kind of the challenges of getting from point A to the final delivery point.
Yeah. I mean, for you guys, I wish there was a simple if-then statement that would help you sort of convert all these different metrics to give you sort of more predictable revenue guidance. But it is a very lumpy, non-linear business. Our focus 100% is on what are the opportunities that are out there, meeting with the customer, understanding their needs, and applying our value proposition to deliver for the customer. The vast majority of customers have a time-to-value metric that they're looking for, and they want to know what is the best available technology and the timeline that I'm looking for, and they're looking for high levels of customization. They're looking for the services capabilities that we bring to bear, and they're looking for a total solutions component to it, so our focus is we have a big pipeline. We have visibility to a lot of opportunities.
We work with each and every single customer to really understand their needs, develop whatever customized solution that they need, and deliver it on their timeline, and as we take a look at our backlog, and we take a look at our pipeline, and what do we see that's coming in the next quarters, a lot of that's sort of embedded in the guidance that we provide. Right? So, Yvonne provided guidance for Q4, and then as we get into the Q4 earnings, we'll talk about what guidance for next year looks like, but again, I don't think the team has had this much fun in a very, very long time. I mean, you got an ISG business that's growing 34%, delivering operating income of 13.3%, $1.5 billion of operating income dollars, growing faster than revenue at 41%, so it's an incredibly exciting time, and we're very early innings.
Our ballgame hasn't even started with AI. Right? The opportunity is significant.
So that's a great lead-in to my next question. So when you think about the pipeline, this pipeline that grew 50% quarter- over- quarter, how do you define what gets thrown into that pipeline? Right? It's an unweighted, not probability-adjusted pipeline. Is it indications of interest from potential customers? What does that sort of communication look like? And then it gets layered into your, I would imagine, your CRM system. And how should we think about that?
Yeah. So we have a pretty good process around what kind of gets into the pipeline because, like I said, we have visibility to a lot of the sovereign opportunities that sit out there, primarily in the Middle East, India, Japan, Australia, U.S. government, followed by U.K., France, Germany. So we sort of take a look at where customers are in this architectural decision-making to really understand that we have enough maturity in an offer to actually put it into the pipeline because that's typically how we do our demand forecasting and start to kind of get supply lined up. But we typically wait until the customer has kind of made a pretty strong decision around the architectural choice that they're making to make sure that we can lock up the supply that we need.
And then, just obviously, there's a lot of noise and chatter in the marketplace over the last couple of months. Leading ODMs had some challenges. There's been this perception that their challenges have been Dell's business to win. Not my words, but maybe the market's words. How do you think about the competitive landscape from the solutions perspective that you offer versus a more traditional ODM model that doesn't have all the bells and whistles effectively and all the bundled solutions and services as we go tier two sovereign enterprise? Clearly, you have the view that what enterprise and sovereign want is this, in my words, white-glove service effectively. Do you think that's what ultimately you touched on it earlier is going to be a key deciding factor?
And so maybe the hyperscalers are just, "Look, they want an ODM solution, but everyone else needs a much more robust solution that helps them go down their own journey.
Yeah. I mean, this is an answer that probably surprises you a little bit, but I don't really spend a whole lot of time thinking about the competition. I mean, this industry is moving so quickly. The pace of innovation is happening so fast. We're head down doing the stuff that we need to do to deliver for customers. ODMs can be formidable competitors. Other companies out there could be formidable competitors. I don't spend too much time worrying about that. We have enough stuff to go do. And we feel like we have a strong value prop, and that's what we're all focused on.
Got it. Liquid cooling. How important is liquid cooling from a customer perspective today in terms of the solution that you bring to the table and kind of how that's getting kind of incorporated into what customers' roadmaps look like going forward?
Yeah. Look, I think it's very important. And clearly, it's going to be more important in the tier two CSP and sovereign space where you have more greenfield opportunities. I think there'll be some enterprises that have the data center readiness to take on liquid cooling. So that's the way I would sort of think about it from an importance perspective. Look, we've been a leader in liquid-cooled solutions for two decades now. We've moved from air-cooled systems to liquid-assisted systems to now direct-to-chip liquid-cooled systems. But it's not just the direct-to-chip liquid cooling because when people are buying liquid-cooled systems, they're typically buying it at rack scale. So we're doing the cooling and vertically integrating the rack. We're looking at our own CDUs. We're looking at our own cold plates, our own manifolds, our own rear-door heat exchangers, our own heat capture cabinets.
And that becomes incredibly important because a lot of the current suppliers maybe don't have the scale or the sub-tier supply chain to support the needs of high-scale customers. And so we want to make sure that we can vertically integrate that, design it, and own the supply chain to ensure that we're delivering to customer requirements at the scale and quality that they deserve.
So where do you stand today with that vertical integration with the condensing units and the cold plates? Is that the solution that you're offering, or are you still relying?
We're not there yet.
You're not there yet.
We'll be there in the next eight to 12 months.
Eight to 12 months. Okay. And then ultimately, does that add to the level of competitive advantage that you think you bring to the table?
100%.
Complete solution for.
100%. Yeah. When we take a look at the breadth of the portfolio, it's not broad, but it's deep. Right? And we want to make sure that not only are we offering the best individual solutions, but we're offering the best of rack scale architecture, the scalable solution, which is rows of racks, and being able to deliver it in a sled or in a modular data center. Right? Being able to go from L10 to L13 depending on the requirements of the customer.
Got it. When we think about sort of what customers are looking for today from an AI perspective, how much of it is dependent upon pricing versus performance? Is price even a tier two sovereign enterprise? Does pricing even enter into the equation right now since it's obviously a relatively high-demand product, relatively short-supplied GPU product? How do customers think, and not the hyperscalers, obviously, but how does that market think about pricing today?
Yeah. I mean, of course, pricing plays into account. I mean, we've talked about the fact that it's great revenue and incremental margin dollars, but dilutive at the rate when you take a look at the overall ISG P&L. The level of differentiation and the value prop does allow us to price at a premium to our competitors and continue to win at a pretty rapid clip. Right? So again, our job is to innovate, deliver value. And if we can deliver value and show TCO to the differentiation we're providing, we will be rewarded.
I know you don't disclose your AI server margins. Has there been much variability in AI server margins over the last four or five quarters or so? Relatively consistent?
Yeah. I wouldn't say it's material. We've seen an uptick sequentially from Q4 to Q1 to Q2 to Q3, but it's not anything that would move the needle, and we do see higher rates of margin on the enterprise side because we're able to sell different configurations, upsell more servers, networking, more storage, and services along with that, but yeah, no, it's been pretty consistent.
If I go back two quarters ago, there were some market fears that the AI server margins were relatively depressed. That's clearly not the case from your comment. That was other parts of the business. I think the North American storage business might have been a little bit softer, and that's typically a high-margin business. Is that kind of what happened a couple of quarters ago?
Yeah. Q1 was a very anomalous quarter. We're up basically 530 basis points from Q1, 230 basis points up over Q4. And as we've guided for Q4, that momentum in operating income growth will continue as we go into Q4.
I know this is maybe a better question for Yvonne, but I'm going to ask you since you're in the business. So is AI server margin less cyclical going forward from a seasonality perspective, quarter to quarter to quarter when I think about ISG margin? So does it dampen the cyclicality of the profitability?
No. I could answer the question two ways. There's no cyclicality to AI server margins themselves. There's cyclicality to the ISG business in terms of the rate of the mix of the AI-optimized servers.
But AI server revenue is going to be a much larger, theoretically, percentage of ISG revenue next year versus this year versus the prior year. So that to some degree dampens, I would imagine, the swing factor in margin.
Yeah. But our job is to make sure that the other parts of the portfolio continue to grow and that we operate safely within our framework of 11%- 14% operating margins.
Which Q3 of this last quarter, so we're squarely there.
We'll be above that.
Got it. Traditional servers. Obviously, a market that's been tough for quite some time. Obviously, there's some degree of debate around how much of it is substitutive from AI, how much of it is just, generally speaking, white-box ODMs have made tremendous inroads into that market. Kind of where do we sit today from your portfolio and how are we thinking about? I know Yvonne kind of gave a soft guide on the call for traditional server. Even into next fiscal year, I think she said AI server grows the fastest, followed by traditional server, and then followed by storage if I'm not mistaken. Is that the right order?
Yes.
Okay. Just wanted to make sure. I don't want to get bad information. So how are we thinking about where you're positioned traditional server and what kind of sort of backdrop do you have there?
Yeah. Look, I think we're very well positioned. We're in our 16th generation of servers. We'll be launching our first 17G server in a couple of weeks and the broader portfolio coming in the first half of next year. As we exited Q3, we now have six consecutive quarters of sequential growth, four consecutive quarters of year-over-year growth. And one of the main reasons why you saw the 230 basis points of sequential improvement in gross margin was due to TRU expansion and pricing discipline that we have on the server business. So we've shown that we've been able to grow and improve margins. Why are customers buying and why are we growing? Clearly, it's a great portfolio, but there's still a significant install base that's 14th generation servers and older. And we've come off of sort of an elongated period of digestion. So it was ripe for refresh.
And we kind of said it coming into the year that this would be a server-led refresh, and it's kind of playing out the way that we thought it would. Customers are looking at, importantly, what does a modern data center look like? And compute is going to become an incredibly important commodity when we think about a world driven by artificial intelligence. So this notion of being able to buy more efficient servers gives customers room to grow and future-proof their data center in terms of building out the computational horsepower that they are going to need to drive workloads of today, but importantly, workloads that are coming in the next 12, 24, and 36 months.
In addition to being more power efficient, is this iteration of the new server portfolio more physically space-saving, but smaller? So does it save footprint in a data center? Does that factor into a potential customer's decision?
No. It absolutely does, so when you're looking at a 16G server and you sort of compare it to 14th generation technology, depending on your choice of configuration, you can consolidate one to three to five, right? So five servers, one server can do the work of five servers in some cases, or three servers on average. But what we see is that it's not like that's driving the server units down. It's customers are using it to beef up their compute capabilities, knowing that there are significant workloads, some of which are going to be repatriated from cloud.
Got it. That should be a relatively good demand driver for the foreseeable future.
It's a tailwind for sure.
For Dell. Got it. And then storage, I think that's the market where I think people have had, I think, the most consternation industry-wide. It just hasn't really grown, hasn't really recovered, I should say, as quickly as maybe people had originally expected. You pointed out that your server business has improved sequentially for a number of quarters now. Storage typically has been sort of a laggard but has generally followed strength in server, traditional CPU-based server. So where do you think we are today? I mean, obviously, your storage business was a little bit better this quarter, but it's been a little bit fits and starts in kind of what do you need to see to get the confidence that it's actually going to grow again? If the storage market grows mid-single digits next year, you can actually grow at a reasonably healthy clip.
Yeah. So coming into the year, we said this was going to be a server-led recovery, and that storage typically lags two to three quarters. Right? So if you kind of figure the storage recovery or the server recovery sort of kicked off in the middle of Q2, that kind of puts you at, hey, at the beginning of next year, we should start to see the storage recovery. The market forecast is for it to grow about 4% next year. And we said on earnings that we would grow at a premium to market and gain share in storage. And why do we believe we're going to continue to gain share? We're very focused on customer use cases and how we think about modern data centers. And they kind of come in three flavors, David. One is I want to set up a private hybrid cloud.
The second is around building out a data platform, and the third is around building out a security platform. If you kind of pivot that to the portfolio, it plays very strongly, but before I get there, there's also an architectural component that customers are looking at in terms of what meets their storage requirements, and what they're looking for is container-based disaggregated architecture, and the reason being, they're looking for the highest level of performance, highest level of utilization of both the storage and the server node, as well as flexibility in what is increasingly becoming a multi-hypervisor environment, right? So if you kind of pivot the main trends and architectures customers are looking for, and you take a look at the portfolio, in the cloud space, we have PowerMax for the ultimate mission-critical workloads and PowerStore as a premier mid-range product.
We just launched PowerStore Prime with the industry's leading five-to-one guarantee. So when you're looking for ultimate in performance, security, and TCO with a five-to-one data reduction rate guarantee, it really hits the mark. And we saw the proof points in Q3 with double-digit growth in PowerStore. And then we had the industry-leading software-defined for structured databases with PowerFlex that also grew double digits. So have a very, very strong story for customers looking to build out private hybrid clouds. Then the customers are looking to build out data platforms. And this one is somewhat related to AI. Right? We start with the base of PowerScale. And again, we saw the F710 and F910 grow double digits in Q3. And we saw the all-flash business in the unstructured space grow double digits in Q3, which are strong proof points.
But we have PowerScale, which is a decade leader in MQ and distributed file systems. We will be incorporating performance object into it. And then importantly, above the storage layer, building a Dell Data Lakehouse for metadata management and then delivering in the first half of next year Project Lightning, which is our parallel file system. We believe that is going to be a differentiated and a very unique and differentiated data platform for customers. And then on the security side, we have the leading target appliance in PowerProtect Data Domain and other MQ leader. We're number one in that space by a wide margin with 60+% share. And we're now three years into PowerProtect Data Manager, which is our modern data mover to support all of the modern software backups to go along with the target appliance. All of this is Dell IP.
All of it is Dell IP. Right? And so that's where our focus is. And that's why we are pretty confident that we're going to get back to gaining share next year.
So, small part of the business. No one ever talks about your networking business. Anything to think about there going forward in terms of disruption in the industry? How much of a focus is the networking piece of the business? I know it's relatively small within ISG. How should investors think about the commitment to that business?
Yeah. Look, I mean, the networking business is pretty interesting and is very important, especially in the world of AI. Networking topologies are going to change in a meaningful way. In many ways, in every way, actually, the network is the bottleneck to AI. Networking in an AI workload is 300x what it is in a traditional compute workload. That's both in the scale-up and scale-out fabrics. So our opportunity to deliver Dell switching with SONiC operating systems, supporting InfiniBand and Ethernet with Dell optics and cabling is a very strong value proposition. We grew that business in Q3. We plan to grow it in Q4, and we look to grow it next year.
Got it. We'd like to give management teams an opportunity to kind of talk about what's misunderstood, underappreciated, what's the market missing, what are investors missing. What do you think is the most underappreciated aspect of the business that you run within Dell?
Look, I mean, I think investors generally hold us in very high esteem and hold us to a very high mark, which I think is great, but look, I think people don't always appreciate the breadth and depth of our portfolio. When you think about the compute portfolio, the AI portfolio that didn't exist 18 months ago, you think about all the products that I just talked about, the 9680, the 9680L, the 7745, the 7740, the IR5000 rack scale, the IR7000, the NVL72, the NVL4 that's coming soon, the XE7725. None of these products existed 18 months ago. The work that we've done on networking, the advancements that we've done on storage, the solutions that we've built around it, the level of breadth and depth across the portfolio is like none other in the industry, and the pace of innovation, I think sometimes goes underappreciated.
It's like, "Well, isn't this what you guys are supposed to do?" Absolutely is what we're supposed to do, but man, the teams are just doing it at such a torrid pace. It's just incredibly impressive to see. Look, we're a humble team of 14,000 R&D folks that are just looking to deliver on behalf of our customers.
Got it. My final question is, when we think about fiscal, I guess your fiscal 2026, calendar 2025, outside of AI server momentum and GPU considerations, what are you most excited about next year? You talked about the litany of opportunity within storage. You talked about the litany within server. Server was the this year was a server-led year. What are you most excited about for next year outside of AI within the portfolio under your purview?
Getting back to gaining share in storage. Full stop.
Full stop. Got it. Actually, I have one other question from the crowd. So the other question is, so if AI continues at pace, you take share in storage, margin should be better next year than this year. Storage is a better margin product. Correct?
Yes.
Just to clarify.
Yep.
Got it. So that is an important kind of cog in the model for ISG is.
100%.
Taking back share.
Yes.
And then maybe just one thing I didn't understand is when I think about where you've lost share, where does that portfolio sit today within Dell? Meaning where you've lost share in some legacy disk drive businesses, which I don't think are pretty critical going forward. Where else has there been some share loss that you think you can stem going forward?
Yeah. Well, I mean, again, what gives us a lot of confidence in focusing on the Dell IP storage is one of the headwinds coming into this year has been sort of the HCI portfolio. But when we take a look at customer trends and what they're looking for and where we can add value, we think we can overcome those headwinds and get back to share gain next year.
Got it. Okay. Great. I think we're out of time. Arthur, thank you very much.
Thanks for having me, Dave.
Thanks to everyone.