Good morning, everyone, and welcome. I'm Toni Sacconaghi, Bernstein's IT hardware analyst, and I'm thrilled to have Michael Dell, Founder, Chairman, and CEO of Dell Technologies, join us today for a fireside conversation. Before I begin, I'd just like to read the Safe Harbor, a reading from Dell. "Dell Technologies statements that relate to future results and events are forward-looking statements based on the company's current expectations. Actual results and events could differ materially due to a number of risks and uncertainties, including those discussed in the company's SEC filings. The company assumes no obligation to update its forward-looking statements." Now, this is an open Q&A in the sense that investors can ask and vote on questions via Pigeonhole. You should have received those links when you registered for this call.
That would be the best way to vote on questions and ask questions for Michael. So without further ado, Michael, welcome. Happy New Year, and thanks very much for joining us today.
Thank you, Toni. Great to be with you and all the investors that have joined us today.
Well, why don't we just plunge right into it? You know, the topic in technology for, I guess, the last 14 months since the launch of ChatGPT has been AI. So maybe broadly, you've been in the technology industry for 40+ years. Maybe you can talk about, just broadly, how do you see AI impacting the economy and companies over the next five to 10 years?
I am quite optimistic about the impact here. I think it is a big unlock of productivity and will be, you know, a catalyst to overall economic expansion and growth. And I think when we look back on, you know, sort of, the decade of this decade, it will have been a major factor in economic expansion. And, of course, you know, from a technology industry perspective, it's, in a sense, a big unlock of all the data that was accumulating and rapidly growing, and a TAM expansion because it effectively gives technology a bigger role in the economy than it already had. And so, yeah, I'm quite optimistic about the opportunity.
Certainly, you know, we see, you know, a big TAM growing for, hardware and services, which is the place that we tend to play in. And, you know, as I talk to, customers of all sizes, it's really interesting. You know, you have a customer roundtable at one of our Dell Tech Forums with, customers, you know, of all varieties, and this is the number one topic by far. It's certainly the number one topic with the largest customers as well. And so there's a great interest and appetite, and customers need a lot of help in doing this, and we're really just at the beginning of, helping customers make this happen.
Great. So, Michael, maybe I could just follow up on a couple of things that you mentioned. One is: how do we think about... you talked about productivity enhancement. Have you thought through, or what are you hearing from customers in terms of the most promising applications for AI, and are they incremental to existing productivity? You know, when we talk to CIOs, some say, "Look, we've been doing AI in some form for 5+ years, RPA, et cetera. Look at how automated customer service has been relative to 10 years ago." So, when you think about the principal opportunities in AI, what are they, and are they incremental, or are they completely new?
It's a great question, and you're absolutely right. You know, machine learning has been around for a while. There was sort of a big machine learning wave that started around 2010, and a lot of companies have been using propensity models and, you know, prediction models in their systems. You know, one of the interesting things about this sort of generative AI wave is that it's put increasing attention on the technologies that have already been available, that maybe organizations haven't been using. But the areas where I've seen the most, interest and application are in anything having to do with customer operations, content creation and management, software development, and sales. And as to the question of, you know, is it incremental or not, you know, I think it sort of depends on the perspective of the company.
And, what I'm seeing is, you know, there are companies who say, "Hey, the budget's the budget, we're not changing it," right? And then there are other companies who are saying, "Well, wait a second, we can get a 20%-30% productivity improvement here. Whatever the budget was, forget about that. Now we're gonna go do this, because it's too big of an opportunity to let it go by." And everyone will fall somewhere in between those but, you know, as more and more of the reality sets in, that this is a sort of true competitive advantage and game changer, assuming that's the case, right? And, I'll acknowledge that there's definitely some hype in this area.
You know, I think, you know, it is sort of a change or die kind of moment, I think for many organizations and industries. And, you know, as a result, I tend to see it as much more of a TAM expansion.
Great. Well, how, you know, along those lines, I mean, one of the things that I struggle with as an analyst trying to quantify impacts of things, is just how do we think of the potential productivity impact from AI relative to other productivity waves we've had over the last, you know, 40 years? I mean, you were at the forefront of the PC revolution. You know, when I had my first job, and we're about the same age, actually, when I was a summer intern in college, I had a, I think, 128 KB Mac, and I used to have to bring it home to run the software application all night 'cause it took so long. And, you know, I had a rotary dial phone, it was landline.
You just think of the PC revolution, and then you think of the internet revolution, and you think about digital transformation. It feels like we've had these kind of successive waves of productivity. And so, tech spending as a percent of the global economy has increased, certainly increased as a percent of the market capitalization, which is a reflection of that. But is this something that's necessarily gonna be more productivity-enhancing than prior waves? Like, if companies spent 2%-4% of their, you know, their budgets on IT spend in the past, like, is that number gonna be higher because of AI? You know, and, how do we think about that in the context of all the productivity that technology has brought, you know, over your lifetime?
Is this something that's truly bigger and more significant than some of those other waves?
I think it's at least as, as significant as the prior waves. You know, I think it'll vary, you know, by industry, but to your point, you know, if you took any specific organization and you said, "Oh," let's say they didn't adopt the PC or the internet or digital transformat- right? They wouldn't probably be around now, right? And so, there is a speed advantage that organizations will have in terms of adopting, you know, maybe, maybe ahead of others, but these things do get normalized out. I do take the, the argument that this will overall make the pie bigger, because it's effectively going to amplify human capability, and creativity, and productivity, and opportunity, and, you know, I think, you know, can have tremendous positive impact on healthcare and education.
And, again, when I talk to customers, the CEOs and the board members have kind of figured out that this is a big deal. And, you know, the initiatives inside the big companies to make use of this are driven at that level, and that's somewhat different than the prior waves. That might be somewhat of a generational change, but it could also be, you know, that the data is showing some fairly significant magnitude, right? People are talking about 15% productivity gains, 30% productivity gains, for example, in code development. If those numbers are anywhere near correct, you would be derelict in your responsibilities as an officer and executive of a company if you weren't figuring out how to go get that.
Got it. Understood. Now, you mentioned about kind of the intelligence and the... that's going to occur. Like, philosophically, to you, and this is maybe a little off-topic, but nevertheless of interest, like, do you worry about AGI, and do you even think and contemplate about a timeframe for when that might occur?
I mean, I worry about it a little bit, but not too much. You know, for as long as there's been technology, you know, humans have worried about bad things that could happen with it. And we've told ourselves stories for eons about, you know, horrible things that could happen with whatever new, unknown, you know, force in the world there is. And we actually have a great mechanism as humans to worry about things and then, you know, create counteractions so that the bad things don't actually occur. And we can think of a gazillion examples of that. Even, you know, since we're both about the same age, you remember the ozone layer and all, you know. I mean, there are all sorts of horrible things that were gonna happen. It didn't happen because humans took countermeasures.
You know, any tool, whether it's fire, or the wheel, or, you know, AI, can be used for good things or bad things. These tools don't wake up in the morning and say, "I'm gonna be good today or bad today." It's what we ask it to do... and we have to make sure that it reflects our humanity. And, you know, that's our job, is to make sure that it doesn't go crazy, and we don't turn the world into a bunch of paperclips.
one of the-
There was a book about that.
Yes. One of the things that our tech team at Bernstein has been thinking a lot about, or wondering about, is just how big the infrastructure build-out right now is happening for AI. And, you know, we look at NVIDIA shipments, and, you know, consensus numbers sort of point to AI servers being in 2025, being, like, $120 billion, which is, you know, as big as the traditional general purpose server market today. And it's a 5x increase in the server market, AI server market, in, like, three years. We went through the entire internet, and the server market didn't really grow. It was the biggest build-out in infrastructure and data history, and the server market was $50 -70 billion a year for 15 years.
So, I guess my question is: Do you worry that we may have some temporary overbuild in capacity, that will be absorbed, no doubt, over time, given the size of AI and given its capabilities? But, frankly, the magnitude of that server build-out relative to other metrics that I mentioned looks very big. Do you have any of that concern, or do you think it's just a straight line from here?
Well, you know, sort of thinking about your question, you know, what would you do if you had concerns? So I think if you look at where we're investing our capital, we're being pretty disciplined here. I think there's no question that there's a big build-out. Whenever you have, you know, cycles like this, the opportunity for excesses to occur is absolutely there. Nobody quite knows what the demand-supply balance really looks like, particularly when you have a condition like this. You know, I heard estimates from last year that said that the demand, actual demand was, you know, 5x the supply. Who knows? I mean, maybe that's right, maybe it's 4x , maybe it was 1.2x . I mean, but there certainly is aggressive build-out.
Look, I think if you talk to customers, you know, a significant majority of them believe that AI is gonna help them be more competitive, you know, change the rules of competition, help their cost structure, disrupt their industry, help them with product delivery and differentiation, and we're kind of at the beginning of that. We're also starting to see the competitive landscape change as you have, you know, diversity in the hardware and software layers. Also, as we go from just training to inference. Again, I think it's a big opportunity. There certainly could be excesses, you know, formed in AI and in the demand signal in one place or another.
You know, we try to carefully insulate ourselves from those.
Are you... What are you seeing in terms of AI business momentum? Is it just continue to go straight up to the right?
It's continued to be very strong. You know, customer demand nearly doubled quarter-on-quarter for us, and the AI-optimized, you know, backlog roughly doubled to about $1.6 billion at the end of our third quarter. Sales pipeline grew substantially, you know, roughly by 3x quarter-on-quarter. And, you know, having met with a large number of those customers, yeah, there are real use cases behind them. It's broadening out from the sort of AI-as-a-Service, GPU- as- a-S ervice, companies into financial services, government, education, medical, and... You know, I mean, sort of anywhere where there's a prompt in a system is the opportunity for these kinds of technologies to improve the experience in a meaningful way.
Of course, you know, we have a broad portfolio here, and we're just starting to see some demand from the PC or workstation side. It kind of starts with the high-end workstations, where we have leading share. But, you know, some more advanced customers are running these models on their own PC. And of course, there's been a lot of discussion about this at CES in the last few days. There will be a new wave of kind of AI-enabled PCs that have the capability to run these models, just like there'll be, you know, new phones that can run these models. And then you've got all the inference activity that's gonna occur.
You know, once you've trained your model, you've got to put the data through it so you get a better outcome, you make a prediction, you have something, you know, occur more effectively. That's gonna require an enormous amount of computing power.
How are you thinking about inferencing? Do you feel that inferencing may require more infrastructure than training when it's all said and done? Or is inference-
No question. No question at all. So here, here's the... I mean, just think about any activity, and it's obvious that the inferencing of that activity is way greater in computing requirement than training over time, okay? Let's take an example. Let's say you have some sort of vehicle, whether it's a you know goes on the ground or goes in the air, whatever. You know, it's a vehicle of some kind. Well, it requires a lot of computing power to train it, okay? But then the actual movement of the vehicle and the flying the vehicle, that happens every single day, a gazillion times a day, every time there's a vehicle. And so the, so you've got to take the data and put it through the model to get the outcome to know what to do.
It, you know, the same would be true for industrial robots or, you know, any kind of activity. So I think inferencing ultimately is going to be far bigger than training. And of course, inferencing will happen in the physical world, in the real world. It won't happen in a big training data center somewhere, right? It'll be embedded in the device. If you've got an automobile and it's going down the highway, and it has some kind of capability to do something different than an automobile from the past, it's inferencing, and the capability is in the automobile. It's not going over some 6G network to get the answer to decide, you know, what to do while you're driving at 80 mph .
But do you think, I mean, if we think about the compute intensity more broadly, I mean, is there, you know, is there not the possibility that the algorithms will be radically simplified, that model sizes will actually, you know, over time, actually get smaller, and so the inferencing is being done, like, much more locally in a much smaller models, that the compute infrastructure doesn't necessarily follow, you know, the prop- it's proportional kind of number of queries, whether it be, you know, on the inference side or on the, on the training side, but that, you know, you have fundamentally different, changes to, the compute algorithm as it is today, and then also the inferencing model?
You know, and that, you know, arguably, that's what we've seen historically in the form of Moore's Law and virtualization, and that's what's kept the lid historically on server spending. The world has got a lot smarter in either designing those products or designing the algorithms necessary for it. Is that not also inevitable here?
Yeah, no doubt. I mean, there will be enormous innovation in the models themselves, and you're already seeing this with the proliferation of open source. And so you'll see models of all shapes and sizes, you'll see refined models, distilled models, but ultimately, if you calculate the amount of computing power required to process all the data in real time, it's gonna be more than the training. It just is. But you know, unless you don't actually do much with it, right? But I mean, think about, you know, Tesla, for example, right? How many cars do they have? Every single car is doing inference every time it moves around, right? That's a lot of computing power.
And you know, if you know, let's say all. Let's say in, I don't know, 10 years, you know, all the car companies have some capability approximating, you know, whatever, right? The amount of computing power required to do that is enormous.
Last question on AI, and then we'll talk more specifically about Dell. Do you think that AI, or if you had to sort of put a percentage, how much AI training, and if you want to split it out, training and inference will be done on- premise versus the cloud. Our CIO surveys are saying the vast majority believe that it will be done in the cloud. I think the vast majority of Dell's AI infrastructure sales today have been to Tier 2 cloud vendors. How do you think about that? And if you were to hazard a guess, you know, what might percentages look like at end state and why?
Yeah, I mean, look, I think as people get into actually doing this, what they're finding is that it's very hard to move all your data to the AI. And just like, you know, in more traditional infrastructure, there's a hybrid approach. And you know, vast majority of customers also prefer a hybrid approach. And so, you know, most companies already have on-prem colo and public clouds. And again, I think with the innovation that you're seeing with Retrieval-Augmented Generation, and the desire to keep proprietary data inside an organization, you know, I don't think it's likely that all this data is gonna move outside of companies and into at least a public cloud environment. Certainly, some of it will, but you'll also have...
I mean, we have a lot of customers now that have innovated with us in a model where they keep their data in a colo that is one hop away from all the public clouds, and they can access all these various services that they want without having to replicate their data, you know, many, many different times. And again, you know, it's I would argue that it's more likely that you'll bring AI to your data than data to your AI.
Just from a profitability perspective, in terms of Dell itself, you know, the company has commented that, you know, right now, AI servers are not as profitable as traditional servers. And, you know, is that a function of the customer? Meaning, you know, these are cloud-like customers and, you know, Dell's experience in selling to Tier 1 cloud vendors was it was very difficult to make money there, no support, you know, contracts that were, you know, often considered loss leaders by, by OEMs or ODMs who are taking them on. So how do we think about the profitability for specifically AI servers, today? And if it is lower, is it because it's largely a Tier 2 cloud customer, or is there something structural about, an AI server that is intrinsically lower?
Or do you actually think, "Hey, as this moves more towards on-premise, our margins should be the same?" So maybe you can just unpack why the margins are lower today, and is that a cloud versus an on-prem consideration, or are there other factors at play?
So a lot, a lot of pieces there. Margin percentage is going up, but, you know, currently, the margin percentage is lower than the average for servers, but it's definitely accretive to margin dollars. And we're, I think, doing a better job of adding more services, and storage, and networking into that solution. And as it goes more toward enterprise, you know, it, it, we will be able to capture more value there. But, you know, a lot of the early customers, I'd say, there's a lot of buying power, and a lot of the, cost is concentrated with one particular supplier that you mentioned earlier.
So, you know, it hasn't been at the same level as you know, traditional servers, but we're finding ways to improve it.
Just on the support side, Michael, support historically, support service has been a large part of the economics of selling a server, and attach rates in enterprises are very high. What is that like with an AI server today in terms of support attach to a Tier 2, you know, cloud provider? And is that a contributing factor to the profitability of Dell as well?
It is, and it's... you know... We're also starting to see some improvement there, as these companies sort of get into the running of these things over time. And, you know, the way we like to say it is, you know, you're not gonna buy a machine that is critical to your business without the service that goes with it. And so, I think we'll be able to improve that, and the trends are promising.
Okay. If we, if we think about Dell more broadly now, or more specifically, I should say, how do we think about sort of the right industry growth rate and margin level, again, industry, for PCs over the next five years?
... Yeah, so, you know, for PCs, I'll say we've been somewhat surprised at how elongated this down cycle has been. And, you know, look, the PC has generally been a defined life appliance, where at some point, the new one gets that much better that you're just compelled to replace it. And, you know, as we keep stacking up new capabilities, you know, and the install base gets older, it's not a question of if people will upgrade or replace, it's a question of when and with what. We also, you know, keep adding with peripherals and services and, you know, additional categories, you know, to the TAM.
You know, as we talked about, we expect, you know, revenue growth kind of in the low single digits, and as we laid out at our Shareholders' Meeting, you know, back in October, we expect, you know, an operating income framework of 5%-7%. We've been a little better than that, you know, in many cases. But that's kind of the framework that we laid out. There is a big TAM expansion for us in peripherals, and we've done a great job in displays and docking stations. Organically, we've been building a very nice business with, you know, all the other accessories, keyboards, mice, headsets, cameras. And, you know, of course, our services attach and much more emphasis on commercial enterprise.
You know, we have the best franchise in this business in terms of share of profits and share of revenue.
Right, but-
And so, you know, the refresh is coming. Nobody knows exactly when, but we'll be well-positioned.
Yeah, I mean, maybe I can push you a little bit, Michael. It feels like there's almost a disconnect between low single-digit revenue growth for Dell, and you're kind of saying, "Look, like, there's a refresh coming." I think you believe you're a share gainer. You know, you talked about expanding your adjacent businesses and peripherals. And so to me, it either feels like, well, either you're not gaining share or the market's not gonna grow. But or Dell's gonna do better than low single digits, but you know, sort of the calculus of what you laid out doesn't really end up at low single digits in my mind, unless I'm missing something.
Unless you actually believe the market could be worse, or there's gonna be significant ASP pressure, or you know, maybe that's just the floor and you're hoping to do better. But it... You know, if I'd listened to your conversation and then said, "Okay, well, what's the implied growth rate for Dell?" I would not have said low single digits. So maybe you can square the disconnect.
Well, Toni, look, I think, if you know, you've been following us long enough to know that we're sort of a under-promise and over-deliver type of company. And so, you know, I'm not gonna change our framework. And, you know, we definitely have a history of gaining share. We've gained, you know, 10 points of commercial share over the last 10 years, but we don't gain share every quarter. And, you know, it does ebb and flow. You know, if the market does better, I think we'll probably do better, as well.
Got it. Shifting over to the enterprise business, Michael. It does feel that storage margins are... You don't report them explicitly, but you can kind of impute them if you made comments about sort of where server margins are. And it does feel like storage margins are quite a bit lower than EMC's history and some storage pure-play peers. Why is that, and how might this change over time, and is it a priority for you?
Well, let me just say, yep, we agree. You know, this is an opportunity, it's a priority. They are lower than they should be, particularly given that we have a number one position in so many areas. We have been taking a number of steps to, you know, simplify the platform. But, look, if you listen to Arthur Lewis, you know, we're very focused on improving overall margins here. And I think we've got an opportunity there. We haven't delivered on that to our aspirations, but it is clearly an opportunity that we're focused on, with a more simplified product platform, with more software-defined capability, like our PowerFlex...
software-defined infrastructure solution, which has done quite well in the last several quarters. And, look, I think you know, the keys are gonna be growing share, better monetizing our software assets, while continuing to simplify the portfolio.
Is storage the biggest margin improvement opportunity at Dell, or is it AI servers? Or if you think about what could move the needle most in margins, is it specifically the storage business, or is it broader overhead, you know, and further cost reduction, which recently you've done a nice job on? How do you think about, you know, the biggest levers for margin improvement?
Yeah, there's probably a factor of what is the market growth, but all the ones you mentioned have got a lot of focus. And, you know, we have, I think, done a nice job with our operational cost management, and that's gonna be an important part of our ability to generate strong free cash flow and earnings, for the, you know, the next five years and beyond.
Okay. I'm gonna turn to some of the Pigeonhole questions, and again, I encourage people to vote or add to existing questions. So I'm gonna start with the ones that are most voted on. So Michael, what's the biggest growth opportunity for Dell that is underappreciated by the investment community?
I would say, you know, probably, you know, the fact that we have had an incredibly consistent long-term track record of being agile and generating strong cash flow. And, you know, that's not only a combination of our portfolio capabilities, but our culture, and sort of the internal capabilities that we've built. And, you know, over the past five years, we've delivered more than $36 billion in adjusted free cash flow. Q3 was roughly $1 billion, and, you know, roughly $6 billion in capital return in the last seven quarters. That's, you know, more than 90% of adjusted free cash flow.
So if you sort of dissect the culture, and the set of businesses that we have, and the way we run the place, you know, this is just a cash flow machine that knows how to adjust and make, you know, changes as required to continue to do that, and I think we'll continue to do so. I think it's also reflected in our capital return and dividend policy now, fully ensconced and understood. And yeah, we'll continue to do it.
Now, so I think that's a good characterization of elements of Dell, the security that people underappreciate. If you were to say from a market growth opportunity, what do you think is the biggest opportunity that Dell has that is underappreciated?
Yeah, I mean, a lot of discussion about AI on this session today. I think Edge is a significant opportunity. You know, Telco infrastructure, Open Telco infrastructure, those are two new areas that we've been growing and investing in, that are, you know, relatively new domains for us. And you know, I don't know how much the market knows or understands about those. We'll certainly, you know, continue to highlight them and talk about them in our investor communications.
Right. I mean, clearly the market believes, and you've shown that the AI business could be a multi-billion dollar business. You know, can, can Edge and telco add, you know, $5+ billion in growth in, you know, over the next five years? Are those opportunities like, of significant scale? Obviously, they're quite small right now, but, can you dimension them in any way in terms of, you know, how big they could be for Dell?
Well, put them together, and they, they easily, I think, fit in that kind of category. And, and to be clear, you know, we already have businesses in those areas. But if you, if you think about what's going on in, in the world, we just talked about, you know, AI in the form of, of inference, right? You know, we, we, we have a, we have a fabulous business inside Dell that we, you know, refer to as our OEM business, and what this essentially means is our machine is the brain of a bigger machine that is made by, you know, Emerson or Motorola Solutions, or, you know, pick your company. We got 5,000 customers, and it's an enormous business, and it's sort of totally uncorrelated with, with, with the, with the traditional IT definition, and it's growing very nicely.
We're continuing to build out more and more capability in Edge. Telco is interesting because as you move to a sort of disaggregated, virtualized, software-defined telco networks, there'll be more discussion about this at the Mobile World Congress coming up next month. You know, all that infrastructure is sort of heading in our direction, and we've been well-positioned for that. Put those two together, yeah, that those are nice opportunities that are, you know, many, many billions of dollars, you know, in a multi-year timeframe.
Okay, great. Another highly voted question on Pigeonhole is: What are the odds of Dell doing a large deal greater than $8 billion in the next five years?
Not very high. I think we've been pretty clear about this, but, you know, you know, if you sort of think about where the company is. You know, we've talked about our return of capital strategy. I mean, if you sort of think about big deals that you could do, not that one would have a strategy to, "Let's go look for big deals," right? That's generally the wrong way to do it. First of all, a lot of things, you know, sort of, aren't practical from a regulatory standpoint, okay? A lot of things aren't practical from a valuation standpoint.
And, you know, we're just not focused on, and don't see an opportunity to use our capital effectively in that way. And so, you know, that's kinda why we've laid out our return to capital strategy, the way we have. I mean, you could see us do small, you know, acqui-hires and, you know, tuck-ins like you've seen, you know, various small things, but I would not expect a large acquisition anytime in the next several years.
Michael, you know, Dell was more acquisitive in kind of the 2010s era, you know, about products and services, various software assets, you know, had VMware. A lot of those were sold. Is Dell today... So you went from arguably a more comprehensive solutions provider, software, services, and hardware to something that's arguably less in terms of breadth. Do you feel you have the right portfolio today? And was that, in hindsight, was that foray right at the right time? Was it a mistake, given that you divested it? But you know, arguably, the trajectory felt different in 2010 to 2015, where Dell was going in terms of its capital use, in terms of, you know, moving into software and services.
So I don't know if you can reflect on that, but more importantly, like, is the portfolio where it needs to be today?
Well, so, feel very comfortable with where the portfolio is today. If I go back to sort of 2009, 2010, 2011, 2012, I think, we made some mistakes. You know, we, we bought, some businesses that weren't leading businesses. They were difficult, in that they weren't close adjacencies, and, you know, some of them were more tools than platforms. And, you know, that sort of informed, you know, something much bigger that came along a little bit later, you know, everybody knows, the EMC, VMware. That all has worked out quite well. And yeah, I feel great about the portfolio. And, there you go.
Okay. You talked a lot about capital allocation, and Dell's been very explicit about that, and-
Yeah, let me-
... the targets
Say a little bit more. So, you know, today what you have in Dell is you've got a business that has a very broad portfolio of number one positions, with a unique operating model. And while we did divest the former Perot Systems, we have a fabulous services business that is actually very profitable. And, you know, it doesn't do everything that the system integrators do. That's totally fine. We have fantastic partnerships with the SIs. One of the reasons we divested it was because, you know, EMC had fantastic relationships with the SIs, and so we've continued those, and we don't really wanna compete with them and didn't really find much revenue synergies anyway in trying to do that ourselves. So, ...
You know, on and on, on the software side, obviously, we're continuing to invest in and expand our own software IP, like software-defined infrastructure with PowerFlex and data protection. And again, a set of businesses with number one positions.
Now, Michael, you talked a lot about the capital return, and it's substantial, and you also have excess cash, which the company has acknowledged. You know, gross cash is about $10 billion, you only need about $4 billion or $5 billion to run the business. You know, mathematically, the capital return plus the excess cash could effectively buy out Dell again, and make it go private. Is that a possibility?
That's not our plan at all, and, you know, let me start by giving you, you know, 14.1 million reasons. So if you were paying attention to SEC filings, you'll see that I transferred about 14 million shares of Dell stock to charitable entities last year. I think a significant majority of that stock is now, you know, sold, free-floating in the market, you know, et cetera. I would expect that will continue. This is something I signaled at our October meeting, and, you know, to be very clear, we have no plans to go private. We are planning to stay public, and, you know, we're gonna use our capital to repurchase shares and provide dividends.
You know, one should expect that I will continue to distribute shares to charitable entities, and to sort of fund those activities.
Michael, how about Silver Lake? Obviously still have large share in Dell. How... Why should investors not worry about sales of their shares, and how should they think about the likely timeframe for exit, and why that couldn't pressure the shares?
You know, I can't speak for them, but I just observing them, they've been super thoughtful about this. They used a fund-to-fund transfer to roll their Dell investment into three newer funds. You know, they're looking to put capital to work. It's kind of hard to find an investment as great as Dell. This is what Egon Durban will tell you at the Silver Lake annual meeting, and you know, I think you gotta listen to what he's saying. Having said that, they have roughly 90 million shares now. They started with 137 million shares, so they have very thoughtfully distributed about 1/3 of their shares to their LPs.
You know, I'll also tell you that from time to time, we've gone to them and say, "Hey, can we buy some of your shares?" You know, and they don't wanna sell them to us, so there you go. But I think they've been very thoughtful, they're long-term shareholders, and we don't have any other information.
Speaking of maybe not having other information, but it's topical, S&P 500 inclusion for Dell, I think since April of last year, you qualify to be included. How do you think about that and its likelihood?
Well, this is a better question for the S&P Selection Committee than for me. We've done our part, and I think, I think we just wait our turn. You know, we meet all the criteria, as far as we understand them, and, you know, we'll just keep being a great company and generating strong cash flow and earnings, and eventually our number will come up. That's kind of the way it works.
Okay. I'll ask a couple more just from the Pigeonhole, and then we'll wrap the call. Understand the AI reports from two cloud folks. I'm just reading the question. Not quite sure that's 100% accurate, but nevertheless, high concentration of cloud customers. Are enterprises taking a wait and see approach to on-prem AI? What needs to happen for larger on-prem AI enterprise investment?
I don't think they're necessarily taking a wait and see approach, I think, I think they're just at the beginning of it. And they're kind of, you know, developing the architecture, a ton of proof of concepts going on, some deployments are starting. But as I mentioned earlier, as customers get into the whole data flow and system architecture of using their existing data with, you know, generative AI models, and retrieval-augmented generation, and all the things you need to do to build a whole pipeline to successfully do this, it's gonna require a lot of infrastructure. And, you know, we're starting to see that.
... Okay. Just seeing if we got to everything. I think we largely did. Michael, maybe I can just, or we can just finish on, are there important things that we didn't talk about on this call that you wanna highlight? And is there anything more that you wanna say? You talked a lot about the underappreciation of cash flow of Dell. Is there anything more you want to talk about as Dell as an investment? So two separate things. Anything fundamentally we didn't address that you want to underscore, and then, B, anything else you want to highlight about Dell as an investment?
No, I think we covered it, Toni. You know, I think more investors are starting to understand what we have in this business. And again, very long-term track record of being agile and figuring out how to grow cash flow and earnings successfully. Number one franchises in a broad set of infrastructure areas and you know, a very disciplined leadership team. And I think tremendous opportunities for us with all that's going on with the incredible growth in data, AI, Edge, et cetera. So very excited about the opportunities ahead.
Okay, Michael, I'm sorry, I have one minute, and I can't resist a follow-up. You did talk about the leadership team.
Sure.
How do we think about succession for you and Jeff? You did bring someone in from the outside as a co-CEO. They're no longer at the company. Jeff and you have been leading the company for a long time. How do we think about your timeframes for transitioning leadership of the company, and how do we think about succession?
Yeah, well, you know, I think I've got a pretty long runway still to go. And look, we do have a strong internal bench. You know, you mentioned Jeff has been, you know, been here 37 years, I think. Yvonne, our CFO, over 25 years. Sam, who leads our client business, 25 years. Arthur, 20 years, including his start at Alienware, interestingly enough. A very experienced leadership team and lots of depth, you know, inside the organization, and we do spend a fair bit of time continuing to develop our team. You know, if one looks, we've been quite successful in continuing to create new leaders from within the business.
I have every expectation that we'll be able to continue to do that.
Right. Well, Michael, thanks very much for your time today on behalf of Bernstein investors. We really appreciate it. We wish you all the best in the new year. And thanks again for doing this today.
Sure thing. Great to be with you, Toni. Take care.