Earnings Call Transcript - DELL Q3 2024

Operator, Operator

Good afternoon and welcome to the Fiscal Year 2024 Third Quarter Financial Results Conference Call for Dell Technologies Inc. I would like to inform all participants that this call is being recorded at the request of Dell Technologies. This broadcast is the copyrighted property of Dell Technologies Inc. Any rebroadcast of this information in whole or in part without prior written permission from Dell Technologies is prohibited. After the prepared remarks, we will have a question-and-answer session. I would now like to turn the call over to Rob Williams, Head of Investor Relations. Mr. Williams, you may begin.

Rob Williams, Head of Investor Relations

Thanks, everyone, for joining us. With me today are Jeff Clarke, Yvonne McGill and Tyler Johnson. Our earnings materials are available on our IR website, and I encourage you to review these materials and the presentation, which includes additional content to complement our discussion this afternoon. Guidance will be covered on today's call. During this call, unless otherwise indicated, all references to financial measures refer to non-GAAP financial measures, including non-GAAP gross margin, operating expenses, operating income, net income and diluted earnings per share. A reconciliation of these measures to their most directly comparable GAAP measures can be found in our web deck and our press release. Growth percentages refer to year-over-year change unless otherwise specified. Statements made during this call that relate to future results and events are forward-looking statements based on current expectations. Actual results and events could differ materially from those projected due to a number of risks and uncertainties, which are discussed in our web deck and our SEC filings. We assume no obligation to update our forward-looking statements. Now I'll turn it over to Jeff.

Jeff Clarke, CEO

Thanks, Rob. We delivered Q3 revenue of $22.3 billion, with solid profitability and strong cash flow. Operating income was $2 billion. Diluted EPS was $1.88, and cash flow from operations was $2.2 billion. In ISG, the demand environment for traditional servers improved over the course of the quarter, and demand for AI servers continues to be strong across a wider range of customers. Demand for storage was down as expected. ISG revenue was flat quarter-on-quarter, with sequential growth in servers and networking revenue, driven by AI-optimized servers as we begin to convert more PowerEdge XE9680 backlog into revenue. For the quarter, we shipped over $0.5 billion of AI-optimized servers, including our XE9680, XE9640, XE8640 and the R750 and R760xa servers. Customer demand for these AI servers nearly doubled sequentially, and demand remains well ahead of supply. In CSG, the demand momentum we saw in June and July continued into August, but slowed as the quarter progressed. The result was CSG revenue that was down sequentially and short of our expectations. Operationally, we executed well, remaining disciplined on pricing in an increasingly competitive environment. And we controlled our expenses, focusing on profit and cash flow, including outstanding working capital performance. And lastly, we've returned another $1 billion to shareholders via share repurchase and dividends. Yvonne will go into more details on cash flow and capital returns later. AI continues to dominate the technology and business conversation. Customers across the globe are turning their operations upside down to see how they can use generative AI to advance their businesses in meaningful ways. These AI initiatives are being driven at the CEO and Board levels. And as a result, we are at the front of a significant TAM expansion. AI-optimized server mix increased to 33% of total server orders revenue in Q3, driven by strong demand from AI-focused cloud service providers and growing interest from other customer verticals. We drove improved demand margins, increased services attached and incremental unstructured storage attached over the course of the quarter. The XE9680 is the fastest ramping solution in Dell history. In Q3, we continue to see strong demand and big wins, including customers like CoreWeave, a cloud provider that specializes in GPU accelerated workloads; and Imbue, which is using high-performance computing clusters powered by the XE9680 servers to train foundational models. Our AI-optimized server backlog nearly doubled versus the end of Q2 with a multibillion-dollar sales pipeline, including increasing interest across all regions. That said, AI hype is everywhere, and we need to be measured in our expectations. We are still in the early innings with AI as customers continue to work through their AI strategies. Experience over multiple technology cycles tells us that progress won't always be linear, but we are excited about the opportunity in front of us. We believe Dell is uniquely positioned with our broad portfolio to help customers size, characterize and build GenAI solutions that meet their performance, cost, and security requirements. Our AI strategy, AI in our products, AI built on our solutions, AI for our business and AI for our ecosystem partners is the foundation for our actions, priorities, roadmaps, and partnerships. And in Q3, we continue to build our capabilities. We are collaborating with Meta to make it easy for our customers to deploy Meta’s Llama 2 models on-premises, with the Dell AI-optimized portfolio. We are also collaborating with Hugging Face to help users create, fine-tune, and implement their own open source GenAI models on Dell infrastructure. Earlier this month, we introduced the ObjectScale XF960, an all-flash scale-out appliance for GenAI and real-time analytics based on our software-defined object storage solution, which can run on Linux and Red Hat OpenShift on PowerEdge servers. Looking forward, the recovering ramp in PC demand we were expecting in Q3 has pushed out, with large enterprises and corporate customers remaining cautious with their spending. The PC install base continues to age, and there are exciting changes coming to the PC next year, including advances in AI-enabled architectures from Intel, AMD, and Windows on Arm, which will help drive a PC refresh cycle. We are also seeing the beginning of a traditional server rebound, and historically, storage follows a couple of quarters later. We are leveraging our strengths to extend our leadership positions and turn new opportunities, including multi-cloud, edge, and AI into incremental growth. And we are positive on FY '25 and fully expect to return to growth next year, given the expected tailwinds to our various businesses, including AI. Technology is everywhere, and Dell is ready. The amount and value of data continues to grow. And as that happens, the opportunity for Dell Technology grows in tandem. We have proven that over four decades through wave after wave of innovation. And we have proven our ability to capture the growth as our TAM expands and translate that into results for our stakeholders. Regardless of the economic cycle, expect us to focus on growing and extending our core businesses in the areas with the most attractive profits to deliver innovation for our customers, remain disciplined in our pricing, and focus on costs. Now over to Yvonne for the detailed Q3 financials.

Yvonne McGill, CFO

Thanks, Jeff. We're focused on driving a balance of growth, profitability and cash flow in any demand environment. We delivered revenue of $22.3 billion, down 10% with strong gross margins, lower operating expense, and improved working capital management. Gross margin was $5.3 billion and 23.7% of revenue, flat year-over-year. We continue to see increased pricing pressure in Q3 but remain focused on profitable opportunities. And you should expect us to continue to maintain discipline and focus going forward. Operating expense was $3.3 billion or 14.9% of revenue, down 5%, driven by lower SG&A costs and down 7% sequentially as we actively manage our spend. Operating income was $2 billion, down 17% and 8.8% of revenue with the impact of a decline in revenue, partially offset by lower operating expense. Our tax rate was 19.2% year-to-date or 15.4% for the quarter. Net income was $1.4 billion, down 19%, and diluted EPS was $1.88, down 18%. Our recurring revenue in the quarter was $5.6 billion, up 4%, and our remaining performance obligation, or RPO, was $39 billion, flat year-over-year, with growth in deferred revenue offset by a decrease in backlog. Deferred revenue was up primarily due to increases in software and hardware maintenance agreements and VMware resell. ISG revenue was $8.5 billion, down 12% and flat sequentially. Servers and networking revenue was $4.7 billion, up 9% sequentially. We saw server ASPs continue to expand in both AI-optimized and traditional servers, and our AI mix of server demand accelerated again sequentially given customer interest in GenAI. We delivered storage revenue of $3.8 billion, down 13%, with demand growth in data protection and PowerScale. ISG operating income was $1.1 billion or 12.6% of revenue, down 170 basis points, driven by a decline in revenue, partially offset by an increase in gross margin rate. Looking forward, our many #1 positions are proof of our deep enterprise expertise. And with a TAM of $200 billion growing at a 7% CAGR over the next few years, we are confident in our ability to grow the business as the market returns to growth. Our fiscal Q3 CSG revenue was $12.3 billion, down 11%, primarily driven by a decline in units, while ASPs remained flat. Commercial and consumer revenue were $9.8 billion and $2.4 billion, respectively. CSG profitability remained strong in Q3, with operating income up $0.9 billion or 7.5% of revenue. Operating income was down 20 basis points, driven by a decline in revenue, offset by lower operating expense and an increase in gross margin rate as we maintained pricing discipline and benefited from lower input costs. With a TAM of $400 billion growing at a 2% CAGR, we will continue to focus on commercial, the high end of consumer, profitable relative performance, and executing our direct attach motion for services, software, peripherals, and financing. During the quarter, we saw continued strength in APEX and our Data Center Utility and Flex On Demand offerings and added new multi-cloud offerings, including APEX cloud platform for Azure and Red Hat OpenShift. Our Q3 Dell Financial Services originations were $1.8 billion. DFS ending managed assets reached $13.9 billion, up 1%, while the overall DFS portfolio quality remains strong with credit losses near historically low levels. Turning to our cash flow and balance sheet. Our cash flow from operations was $2.2 billion, primarily driven by working capital improvement and profitability. Working capital benefited from an approximately $200 million sequential decline in inventory, strong collections performance and continued improvement in receivables aging. Our cash conversion cycle improved again sequentially and is now at negative 52 days, a 20-day improvement since the end of last year. We ended the quarter with $9.9 billion in cash and investments, flat sequentially, driven by free cash flow generation, offset by $1 billion in capital returns. Core leverage was 1.6x exiting Q3, flat sequentially. During the quarter, we repurchased 11.2 million shares of stock at an average price of $66.55 and paid a $0.37 per share quarterly dividend. Turning to guidance. Enterprise and large corporate customers continue to be cautious in the current macro environment. Against that backdrop, we expect Q4 revenue to be in the range of $21.5 billion to $22.5 billion, with a midpoint of $22 billion. Sequentially, we expect ISG revenue to be up mid-single digits, driven by sequential growth in traditional servers and seasonal growth in storage. We expect CSG revenue to be down low-single digits sequentially. We're seeing pockets of stability in PC demand, but have yet to see a broader recovery in the PC market. And in our Other business segment, we expect to be down in the low 20s sequentially. Operating income rate should be down marginally versus Q3, driven by a more competitive pricing environment in CSG. And for our tax rate, you should assume roughly 20% plus or minus 100 basis points for Q4 or 19.5% at the midpoint for the full year. We expect our Q4 diluted share count to be between 729 million and 733 million shares, and our diluted EPS should be $1.70, plus or minus $0.10. We're increasing our expectations for the full year diluted earnings per share to $6.63, plus or minus $0.10. Turning to FY '25. It's still early in our planning process. However, I recognize you're thinking about next year. So let me share our current thinking. We're seeing signs of stability and inflection in parts of the portfolio, including traditional and AI-optimized servers. We expect revenue to return to growth next year, above our long-term financial framework. The opportunity is the broader IT spending recovery with large corporate and enterprise customers, particularly in the U.S. We'll continue to focus on profitable growth, but we'll be mindful of the competitive environment and inflationary input costs as we move through the next year. Pricing discipline and cost controls will help mitigate these headwinds. Count on us to continue to execute our unique operational model, focused on cash flow, and returning capital to shareholders. And we look forward to updating our FY '25 expectations in more detail on our Q4 earnings call. In closing, we have strong conviction in the growth of our TAM over the long term with technology trends like AI, multi-cloud, and edge in our favor. At the end of the day, our strategy is simple: Leverage our unique operating advantages to extend our #1 leadership position and capture new growth. This strategy, coupled with our P&L leverage and strong cash generation, drives a resilient long-term financial framework and capital allocation plan. We have proven our ability to generate strong cash flow through profitability and working capital efficiency, including $9.9 billion of cash flow from operations over the last 12 months. At our Analyst Meeting last month, we committed to increasing capital returns to our shareholders. Expect us to continue to invest in innovation, be disciplined in how we manage the business and focus on what we can control, delivering for our customers and our shareholders. We are excited about the future and confident in our ability to create meaningful long-term value for all of our key stakeholders. Now I'll turn it back to Rob to begin Q&A.

Rob Williams, Head of Investor Relations

Thanks, Yvonne. Let's get to Q&A. We ask that each participant ask 1 question to allow us to get to as many of you as possible. Let's go to the first question.

Operator, Operator

We will take our first question from Amit Daryanani with Evercore.

Amit Daryanani, Analyst

I guess maybe to go to the AI server demand discussion, Jeff, that you had. I think you essentially said your orders doubled. So does that imply something north of $4 billion right now? And then how do you think that manifests itself into revenues into fiscal '25? And maybe you can touch on how broad this customer base is becoming versus perhaps a hyperscaler; that would be really helpful.

Jeff Clarke, CEO

Sure. Let me pull that apart on it. So I mentioned our demand nearly doubled quarter-over-quarter. What was interesting is every part of the AI-optimized portfolio grew quarter-over-quarter, and we saw significant growth for enterprise customers. So it's important to note that we saw the entire portfolio grow. We saw the number of customers grow, and we saw the number of enterprise customers grow quarter-over-quarter. When I think about the $2 billion that you mentioned, that was a backlog comment that we made in August. And that included, up until that point in time, the August 1st month of the quarter because that was a real-time update of our backlog $2 billion. We shipped over $0.5 billion of AI-optimized servers during the quarter. So when we think about demand nearly doubling, and I mentioned backlog doubled, we'd like you to part with $1.6 billion of AI-optimized servers backlog at the fiscal exit of Q3. Equally important, we saw the pipeline in the quarter triple. I'll say that again, the pipeline for AI-optimized servers tripled quarter-over-quarter during Q3. Lead times remain at 39 weeks, and demand is ahead of supply. We continue to work to improve supply and are working now to convert that pipeline into real sales into orders, so we can continue to ship and benefit from this exciting time.

Rob Williams, Head of Investor Relations

All right. Thanks, Jeff. Next question?

Operator, Operator

We will take our next question from Wamsi Mohan with Bank of America.

Wamsi Mohan, Analyst

I appreciate the early look into fiscal '25, given some of the puts and takes that you called out, both on revenues and margins. Just wondering, when you say it's going to be higher than your long-term range, is the comment pertinent to overall Dell Tech? Or is it also a comment that we can attribute to both CSG and ISG? Frankly, we're coming off cyclical bottoms in so many of your businesses; it feels as though you should be able to materially outgrow. So any kind of characterization of that would be helpful. And also on the cost side, Yvonne, I think you noted pricing disciplined cost controls, but also some headwinds. So just thinking through that, would you say that there is also upside from a margin perspective or EPS growth rate perspective?

Yvonne McGill, CFO

Thanks, Wamsi. I'd tell you, we're certainly at the early stages of the planning process. We usually wrap that process in the January time frame. But I recognize everyone's interested in next year. So I wanted to give you a little bit more context there. We expect to return to growth, as I mentioned on the call, already above our long-term financial framework. We're seeing an inflection point in traditional servers in addition to the AI-optimized momentum that we've been talking about. We expect servers and networking holistically to be a bigger portion of our ISG mix in the next year. If I move to PCs, our growth expectations will be dependent on the timing of the PC refresh cycle. We are also expecting a decline and mentioned it for Q4 also in VMware reseller revenue with no impact to profitability. We're expecting a more competitive environment overall. We started to see that in the third quarter. So we'll expect that to continue, especially in the PC market into the next year. Other things to consider, input costs are expected to be inflationary next year, led by NAND and DRAM. As always, we will continue to be mindful of our cost structure. Regardless of the environment that we're operating in, we will continue to execute our proven operational model. You can continue to count on us to be financially disciplined, all while driving growth, profitability, and cash flow. We're really optimistic about FY '25 and excited about returning to growth. We look forward to giving you all even more context and update in our Q4 earnings call in February.

Rob Williams, Head of Investor Relations

Thanks for the question, Wamsi. Next question, operator?

Operator, Operator

We will take our next question from Toni Sacconaghi with Bernstein.

Toni Sacconaghi, Analyst

Yes. Your tone on the call regarding the demand environment seems very different from what it was 90 days ago, when you mentioned growth accelerating and a rebound in spending that exceeded your expectations. This quarter appears to be the complete opposite. You performed well above normal seasonality in Q2, but fell below in Q3, and you're projecting a lower than normal seasonality again for Q4. Did you misjudge demand in Q2? Was there a shift from Q3 to Q2 and did you misinterpret the demand characterization 90 days ago? What actually changed? I also want to clarify the AI situation. It would be helpful if you could provide an update on your backlog compared to 90 days ago, which was $2 billion. Considering your pipeline, it seems credible that backlog could reach around $5 billion by the end of the year. If there's a 9-month lead time, shouldn't we anticipate $5 billion to $7 billion in AI server deliveries? And if demand is improving for traditional servers, why shouldn't we expect a significant increase in server numbers next year?

Jeff Clarke, CEO

Well, Toni, that's a few questions. Let me work my way through those. So what happened? You recall in Q2, we talked about the improved demand in June and July for PCs. That certainly helped us close the quarter, and we benefited from that. When we were together in August, we had seen that continue in August. Our PC business was up year-over-year during that month. And then things changed. The business started to slow. It slowed in September and slowed more in October. We saw more cautiousness from our customers, particularly large commercial customers, enterprise customers, especially in North America. We noticed the public sector slow down, while at the same time, we saw stabilization in SP. The significant change was the number of large deals slowed over the course of the quarter as our customers again became more cautious and selective. As the market slowed, we've seen increased pricing pressure. So the pricing pressure changed in PCs from August to October. These large deals became more competitive, and we saw promotional pricing throughout the quarter, more aggressive as we exited the quarter. That did change the PCs. We did not see or call the slowdown in our guidance. The guidance that Yvonne and I gave a quarter ago was what we saw through August, and we were optimistic that the recovery would continue. It clearly pushed; things have slowed. While we have now had two consecutive quarters of quarter-over-quarter growth of our traditional or data center servers, and we have the tailwind of AI continuing to grow. As I mentioned earlier, the pipeline of AI tripled in the quarter. Demand nearly doubled quarter-over-quarter. So those are the tailwinds amid the changes we see in the business. The biggest change was in PCs since storage performed as expected. The Q3 exit backlog was $1.6 billion. Orders nearly doubled, and the pipeline tripled. Our job is to convert that. We're working with our field and our customers to convert the pipeline that grew significantly. The interest in our AI products continues to be strong. On-premise deployment and AI interest continue to be strong. We've been working with Meta on Llama 2, bringing that on-prem, and Hugging Face in an open-source environment to bring those models and tools on-prem. I think it reiterates that AI is going to follow the data. The data is on-prem, and we believe the pipeline and opportunity continue to build for us.

Rob Williams, Head of Investor Relations

All right. Thanks for the answer there, and thanks for the question, Toni. I'd like to think that our tone would be transparent and honest, and that's what I hope you hear from us, that we call it like we see it and give you the best view that we can give you at that point in time. So appreciate the question, Toni.

Operator, Operator

We'll take our next question from Erik Woodring with Morgan Stanley.

Erik Woodring, Analyst

Awesome. Jeff, I just want to dig into some of your comments about spending on AI-optimized infrastructure and the impact that might have on traditional hardware spending. Because obviously, there's a lot of money, as you're very clearly showing us, being thrown at AI-related infrastructure. Can you just maybe give us a bit more detail on what gives you confidence that this spending won't cannibalize traditional either general compute spend or overall hardware spending as we look out over the next 12 plus months? What are the signposts that you're looking at that gives you that confidence?

Jeff Clarke, CEO

Of course, Erik. Let me try. We believe what we've seen for two consecutive quarters in our traditional server business is growth that's grown sequentially from one to two, and now from two to three. What we saw for the first time this year was the pipeline actually grow in-quarter for traditional servers. That's a very encouraging sign. We saw the adoption of our brand-new 16G server double quarter-over-quarter. Our sales force and accounts increased their activity, and opportunities in large deals rose towards the end of the quarter. Our conversion improved over the quarter. I think those signals tell us that this eight-quarter digestion period is nearly complete. Data centers need to be updated or upgraded for additional capacity. Those workloads continue; more data is being created. While at the same time, there's a whole new category of computing—accelerated and AI-optimized computing—that's driven by all the market momentum around generative AI. Thus, we believe there's a big opportunity in both areas. We think there are early positive signs, though I won't use the word recovery. However, we believe we've seen an inflection point.

Rob Williams, Head of Investor Relations

Does that get out of your question, Erik? I just want to ensure that's clear.

Erik Woodring, Analyst

Yes, absolutely. Very clear.

Rob Williams, Head of Investor Relations

Thanks. Appreciate it, Erik. Next question.

Operator, Operator

We'll take our next question from Ben Reitzes with Melius Research.

Ben Reitzes, Analyst

I appreciate it. I wanted to ask about PCs with regard to your comments around 2025. What are you expecting that growth to be above model? What do you think the impact will be of the Windows 10 expiration and some of the new chips coming out that harness AI? If you don't mind touching on that, that would be great.

Jeff Clarke, CEO

Sure. Let me set context as we're heading into calendar '24, our fiscal '25 in PCs. We expect the market to close in Q4 being down. It will be eight quarters of negative growth in the PC industry, the longest I can recollect. It's ripe for a refresh. Another data point for you to consider: there will be 300 million PCs turning four years old next year. That's typically a tipping point for upgrades in commercial. Most of those are notebooks. As everyone was working remotely, the notebook mix went up; 300 million PCs that year are now aging. Next year will bring exciting architectures from Intel, AMD, and Windows on Arm to see AI making its way out to the edge and to PCs. It's a pretty exciting time. I've used the moniker that this is the next great application for the greatest productivity device on the planet. So, considering that the PC market has faced eight quarters of decline, along with an aging installed base with over 1.5 billion units, 750 million of which are over four years old, these are all becoming candidates for upgrades. None of them are capable of running the new AI workloads coming out for the edge and for PCs. Therefore, the advances accompanying Windows will drive a replacement cycle as well. Historically, when a new version of Windows is available or one is retired, that drives a replacement cycle, creating opportunities for us. We're excited about that, particularly given our bias toward commercial. 80% of our revenue comes from commercial PCs. I believe the opportunity looks promising, and our internal model would have the PC market growing at around low single digits, about 3% to 4%. Expect us to grow and take market share.

Rob Williams, Head of Investor Relations

Thanks, Ben. Next question?

Operator, Operator

We'll take our next question from Asiya Merchant with Citi.

Asiya Merchant, Analyst

A couple of questions on my end. Just on the AI opportunity that you mentioned in the enterprise, can you share with us the workloads and the use cases that you're now seeing? Is this gone beyond the cloud service providers you talked about? I think you mentioned CoreWeave and Imbue on the call. But if I recall, there was also some commentary on enterprises. Are you seeing an opportunity for your AI pipeline to grow with enterprise use cases? If you can share any anecdotes on that? And then on share buybacks, obviously, you kicked that up this quarter. Should we expect this level of share buybacks to continue into fiscal '25? Or given improving cash conversion cycles and a rising top line, should we anticipate a further step-up in that run rate?

Jeff Clarke, CEO

Sure. I'll lead with the AI question. A couple of specific data points to help understand and characterize what we're seeing. The number of buyers is up, and the number of enterprise buyers is up significantly. All parts of the portfolio saw quarter-over-quarter growth. The reason that's important is that not everybody needs a 9680. Smaller models and smaller data sets are perfect for our other products, aligning with our opportunities in enterprise. We are observing traction across institutions of higher education, financial services, healthcare, life sciences, and manufacturing for our products. This grants us a considerable opportunity in on-prem data and AI/ML work across various sectors. The tripling of the pipeline included enterprise customers in enterprise demand.

Tyler Johnson, CFO

I'll take the share buyback question. We tripled share repurchase this quarter compared to last quarter. As a reminder, cash was weaker last year, and we had to pull back and focus on dilution management. However, cash accelerated this year, putting us in a good position to be more opportunistic. We don't typically guide on share repurchase, but we remain focused on the 80% plus capital return. You may recall that we first started our dividend in FY '23 and were running at a 96% return of capital. I'm optimistic about that and believe that as we look forward, we will repurchase more than just to manage dilution.

Rob Williams, Head of Investor Relations

Thanks, Asiya, appreciate it.

Operator, Operator

We'll take our next question from Mike Ng with Goldman Sachs.

Michael Ng, Analyst

Just have two quick ones on AI. First, on the AI margin profile, you obviously had very strong ISG margins in the quarter despite rising contributions from AI server mix. So could you talk a little bit about how much of a headwind to that ISG margin rate came from the rising mix of AI servers, if that's the right way to think about it? And then second, I was just wondering if you could talk about the relationship between AI servers and networking and storage for you guys. Should we expect some of that AI sales pipeline that you talked about to eventually include selling Ethernet for AI or Infiniband or more storage in the near term?

Yvonne McGill, CFO

Let me start with the impact of margin rates. We did see a little bit of dilution from the impact of the shipment, but really not a material impact right now. We saw nice performance across our holistic server portfolio. The increase that Jeff referenced earlier regarding traditional servers was helpful in that mix. So it does have an impact, but we talked about it being margin dollar accretive but margin rate dilutive. We saw a little bit of that, but not to a significant extent in our third quarter. Expect as that grows, we will see more of an impact. Again, margin dollar accretive and as we have more services attached, as we expand that into the enterprise space, we'll see more margin accretion coming from those AI offerings.

Jeff Clarke, CEO

That's a great point, Yvonne. The team did a great job improving margins of our AI-optimized servers quarter-over-quarter by selling the value, the design, the performance attributes, its thermal attributes, and its connectivity. Additionally, the successful attachment of storage and services enabled us to see improved margins quarter-over-quarter. Regarding your question, yes, there is a relationship between storage and networking with AI. These are typically small clusters, large clusters, and require high bandwidth. We view it ultimately deployed where the data is being created; much of the data created will be outside of the data center. A significant amount is unstructured, and we see it as a tremendous opportunity. Our structured and object assets will play a critical role here, and networking provides the high-speed interconnect fabric.

Operator, Operator

We will take our next question from Simon Leopold with Raymond James.

Simon Leopold, Analyst

Great. I wanted to see if you could unpack what you see going on trend-wise in storage, particularly given that you said it came in as expected, but the revenue was a bit light versus Street expectations and it looks like it's still down year-over-year in the next quarter. My question is geared toward the longer-term trend because I'm wondering if AI is pulling money away from storage. Do you see margins trending better or worse given shifts in input costs and the better margins that some of your peers have called out recently? Just wondering how you're seeing that trend as well.

Jeff Clarke, CEO

Absolutely. A couple of data points—and Yvonne can chime in as well. When thinking about margins, we saw our storage margins improve quarter-over-quarter and year-over-year, which is exciting. We continue to sell the value of our products and our broad portfolio. However, we do see customers being cautious. A significant concentration of the storage business lies with very large customers, and they're absolutely being cautious and selective. Although, in the quarter, we saw demand grow for data protection and unstructured categories, orders increased year-over-year. Regarding the AI impact on storage dollars, I don't believe AI pulls away storage investments. Eight quarters of server decline has impacted storage performance. Our experience tells us that storage tends to lag behind the server market by a few quarters, and that's what we expect. We see the opportunity around unstructured data is immense; we'll continue to focus on that.

Yvonne McGill, CFO

I'd add, Jeff, that the margin accretion you referred to on storage is real and it's great because it'll be recognized over time in the P&L, with high service and software attached. We don't see all that benefit today, but we will see it in the future.

Rob Williams, Head of Investor Relations

Next question.

Operator, Operator

We'll take our next question from Sidney Ho with Deutsche Bank.

Sidney Ho, Analyst

Great. I have a question on AI as well. It's great to see very good momentum in that business. Are there any concerns that some of these orders could be just double booking, given supply is tight everywhere that could ease over the course of next year? I assume you have great visibility into the backlog, the $1.6 billion you talked about concerning timing. But how about the multibillion-dollar pipeline? How comfortable are you with that? Also, what I would ask about is can you clarify what products and maybe services are included in that order number? Is that just AI-optimized servers like the XE9680? Or do you include CTU-only services in there? Any professional services or even APEX as a service?

Jeff Clarke, CEO

The last question first. When we talk about our AI backlog and demand, it's not high-performance computing. It's not CPU-based. It simply refers to the portfolio that is optimized for artificial intelligence: the 9680, 9640, 8640, 760, and 750. Regarding double booking, it's an unusual market. Considering NVIDIA's role helps generate demand and supply. I think there's reasonable fidelity in the demand signal we see today but can’t say there's absolutely no double booking. This new market approach is based on qualifications needed for supply availability. Many control points ensure great fidelity in the demand signal.

Rob Williams, Head of Investor Relations

Thanks, Sidney. Next question.

Operator, Operator

We'll take our next question from Krish Sankar with TD Cowen.

Krish Sankar, Analyst

Just had one for you. I'm kind of curious; you spoke about AI PCs. What is your definition of an AI PC? How much do you expect the ASP uplift to be for an AI PC? And how much incremental growth could it drive from the current $250 million unit run that you're going through?

Jeff Clarke, CEO

Our definition of an AI PC is one that has the performance capability to run the workload locally, not through a cloud service on an app that has been AI-optimized. The upcoming architectures will all meet the criteria. There are performance variations, but they meet the threshold to consider them AI-enabled. It’s difficult to determine whether this drives significant growth in the TAM. However, you don’t want to be a PC user that doesn’t have an AI-enabled chip. It will outpace yours due to its performance capabilities. As a result, this will lead to productivity gains and drive a refresh cycle. Considering the market's scale, with over 1.5 billion units in the installed base—950 million units sold over the last three years—this catalyst could accelerate the market's growth rate.

Rob Williams, Head of Investor Relations

Well said. Next question.

Operator, Operator

We'll take our next question from Aaron Rakers with Wells Fargo.

Aaron Rakers, Analyst

There's been a lot of questions already answered. But Jeff, I just want to ask you simplistically about AI. You mentioned 39 weeks of delivery time or lead times. Are you at all surprised that that's not starting to change? Have you seen any indications that lead times have pulled back with some of the China restrictions put in place? And with that context, would you expect diversity—in other words, another large GPU supplier—to be a factor in unlocking and converting some of that pipeline as we move into next year?

Jeff Clarke, CEO

A couple of things: The pipeline is NVIDIA pipeline today. It's not an alternative pipeline; it's NVIDIA-based. I wish I could tell you, Aaron, that the backlog was less than—in sorry, the lead time was less than 39 weeks. I can’t today. We work every available channel opportunity, with our supply chain capabilities, to improve supply. We've offered our services; we're hoping to inform you that we're improving lead times and supply availability to better serve our customers. Currently, it stands at 39 weeks. Regarding alternatives, options are on the horizon; work needs to be done with those alternatives, as the software stacks must be resolved before adoption occurs. We have a multibillion-dollar pipeline that’s covered by NVIDIA with a 39-week lead time; we work every day to boost availability to satisfy our demand.

Rob Williams, Head of Investor Relations

Next question?

Operator, Operator

We'll take our last question from Samik Chatterjee with JPMorgan.

Samik Chatterjee, Analyst

I guess sticking on the AI subject, you shipped over $0.5 billion of AI-optimized servers in the quarter. Just curious, with lead times holding where they are, how should we think about the trajectory of shipments in the coming quarters? Is it fairly stable in that sort of $0.5 billion range? Or should we expect some level of ramp-up if supply eases? And then just a quick clarification, Jeff. You've mentioned a few times now, the improvement you're seeing in traditional servers, even as you’ve called out that customers are pulling back spending in recent months.

Jeff Clarke, CEO

Well, we don't forecast shipments in terms of specific dollars by product line for the next quarter. Our shipments for AI are implied in the guidance given by Yvonne. So I prefer not to break it down into specific projections for AI servers. But with a 39-week lead time, you can look at the math to assess potential future shipments. Traditional servers and why we believe there's a green shoot are due to what we considered an eight-quarter digestion period, which is the longest I've observed. Data centers have aged product that needs to be updated, with capacity requirements increasing. We are seeing that improvement across multiple areas.

Rob Williams, Head of Investor Relations

All right. If we could see if we can get one last question in.

Operator, Operator

We'll now take our final question from David Vogt with UBS.

David Vogt, Analyst

Can you talk about the thought process if AI lead times come in a little bit more next year? What that means for your networking fabric or equipment business? And then ultimately, the pull forward for storage. If I missed it earlier, I want to understand how that would represent the rest of the ISG business.

Jeff Clarke, CEO

Clearly, as GPU supply improves, the fabric associated with those units will be shipped and aligned accordingly. The capability for our networking business will benefit from additional supply, which is encouraging. As I mentioned, storage attachments are paramount. We're diligently working with our sales teams to identify prospects as this data multiplies outside of the data center. Our deployment strategy ensures we are there, with Dell storage pathways in place. Each opportunity, be it model training or fine-tuning, creates a need for compute assets, necessitating supportive storage. Think of this as the AI estate, representing a genuine opportunity that our teams are currently pursuing.

Yvonne McGill, CFO

On your question regarding component costs and the potential pressure we may face on margin rates, we operate a low inventory model. Thus, when we encounter component cost increases, we endeavor to recover those quickly by raising prices and passing them through into the market. You'll continue to see us apply this strategy. It's certainly not our first experience in this area, so we have well-tuned processes in place at Dell. However, navigating through such conditions takes time to recapture the effects of cost pressures.

Rob Williams, Head of Investor Relations

All right. Thanks, everyone, for joining us. Yes. Thanks, David. We'll see many of you next week at Raymond James and Barclays and also the first week in January at the Consumer Electronics Show. Thank you. Have a good evening.

Operator, Operator

This concludes today's conference call. We appreciate your participation. You may disconnect at this time.