Good morning, everyone. It's great to be here. Rumor has it that Tim Cook's in the building, and it made me think about the Apple announcement. The team and I were talking beforehand, how do you top U2 and Bono, and how can we get someone or a band or something here to make it exciting for everyone? Jamie Lerner, our new guy, says, "Well, the New York crowd's really smart. You should get someone really smart and that's respected intellectually like a Princeton graduate or even a Rhodes Scholar, if we could find someone like that." Pat, being the sports freak, said, "No, we need an athlete, an athlete with integrity. You can get the Princeton guy if you want, but you need an athlete, and it should be a Hall of Famer and a world champion.
If you could have it be a New York person like a New York Knick, that would be great." Mosley, international guy, says, "No, we need someone with international gravitas, experience, a senator, someone who was on the Foreign Policy Commission, or even a presidential candidate." I'm thinking, "Guys, we can't afford that many different people." Bill Bradley is in the house. Kate was going to introduce him as a former Seagate board member, Bill Bradley, which I thought that would be the pinnacle of your achievements, Bill. Thanks for being here, and Mike and Chris, thanks for being here. All of you, thanks for being here. Kate gave a good summary of what we're trying to accomplish today, we thought we would just start a little bit with what's going on in the business today.
It's always a funny time to be doing this event, kind of three-quarters of the way through the quarter, it's always an important quarter, September. In the last two years, it's been a tricky time to be here in terms of one was coming out of the flood recovery, last year was still kind of what's going on in the PC industry. It was a lot more relaxing coming here this quarter because business is really strong across all segments. You've heard us talk about this, I think, the last couple of conference calls, especially coming out of the March quarter, where we felt like there was real traction beginning in the world.
In the June quarter, I indicated that I felt it was really the first time in five years that it felt like the traction was sustainable through the calendar year, we were kind of reserving judgment because we were always worried about what was going to happen in May, June came very strong. September has continued strong, now we're getting good visibility in the December quarter. I feel even more confident that this is real global growth and sustainable beyond just a one-quarter event. For us, we gave some guidance for the September quarter, the revenue guidance was, I think, $30 million, $35 million, $50 million. We believe those revenues will exceed by at least $100 million more than that, we expect that the good business environment's going to continue into the December quarter.
I want to make a little pause because this isn't really about tactical quarter stuff, but we knew those types of questions were on people's minds, and just leave it for a couple of minutes of Q&A right now about tactical issues. I'd like to shift everybody's attention to the purpose of this meeting, which is more about strategy. If there's any questions. Sure. There's mics, actually, so let's.
Hi. Katy Huberty, Morgan Stanley. It's somewhat topical because in the last few days, the ODMs are actually talking about order cuts in the PC supply chain for the fourth quarter. What do you think the disconnect is between the visibility you see and what might be going on elsewhere in the supply chain?
Well, I don't know which specific OEMs you're talking about or what they're calling PCs, but across the board, we see pretty good strength across every segment. There seems to be a resurgence in the notebook space, so they may be calling PCs desktop. I think if you look at the details between the first half of the year and what seems to be playing out in the second half of the year, I think in the first half of the year, desktop was stronger than people would have expected, and notebooks were probably weaker. I think that's reversed. They're both still growing, but I think maybe what happened with some of the OEMs is they trended out the desktop growth for the second half based on what was going on in the first half. That's probably not being realized, but the notebook growth is taking off.
Net though, it's growth across the board. That may be the subtlety if they said PC. You don't know what anybody means anymore when they say PC. I tend to think when people say PC, they refer to desktop more often than not, as opposed to what we call clients. Client overall has been strong for the year, but there has been a shift between desktop and notebook on relative strength.
To me, what's curious about it is the theme that we laid out, which is the death of the PC march that occurred on the tablet stuff where we said, "No, this is just an ecosystem that's morphing, and you have notebooks that went to phones that went to tablets that went to big phones." We think that these gaps are sorting themselves out in terms of what kind of appliances do people really need, and that with a larger format phone and maybe even a higher functioning phone, the tablet maybe is taking a less important role at this moment. Because of the new designs on the notebooks that people actually have basically started rebuying notebooks, which maybe for three or four years has been pretty stalled in terms of a refresh rate. That may be the subtlety. Yeah.
Hi. Amit Daryanani, RBC Capital. I guess maybe just touch on what are you seeing on the enterprise side, given the $100 million upside you're talking about, between hyperscale and the traditional enterprise guys. Given the revenue upside, do you expect gross margins to be better than the flattish that you've guided for?
Yeah. I talked to revenue for a reason, That's the guidance that we're giving, is that revenue will be higher. Mix overall is, again, it's stronger across the board, You're getting a balancing between a lot of mix on notebook, which would be lower margin, We are getting mix-up on enterprise as well, which is higher margin. Net net, gross margins, I think, are going to probably be where we guided or in that general range, because it's an equal mix, if you will. We have seen traction in the nearline business that we hadn't seen for a couple of quarters. The legacy business has remained strong.
Of course, we've been proponents that, again, it's not a net zero-sum game about, oh, if the cloud's winning, legacy is losing, because there are different application sets that are being served and different customers that are being served. What we're seeing is that the legacy business is strong, We are seeing a regeneration of the cloud-based business, kind of to the theme that we talked about. Once people work through their efficiencies around either time-to-deployment or utilization, that the real demand for petabytes was continuing to grow, I think people are starting to fulfill that now. We're seeing growth across both segments. Yep, then we'll go to this side of the room.
Thanks. Aaron Rakers at Stifel. All of the comments that you've made around the demand environment brings us back to the capacity ship trends in the industry. It's been a little bit slower the last couple of quarters. With the comments that you're saying today, where do you think we re-accelerate to in terms of year-over-year growth in capacity shift? Is there any updated thoughts with regard to the potential for the industry to see any tightness on finished media?
The capacity shifts are pretty dramatic this quarter. I think we ended fiscal year 2014 at something like just over 900 gigabytes for the year, and it was pretty steady throughout the year. It kind of had its fits and starts, but in that year, we had weeks where we shipped over a terabyte average capacity per drive. Maybe it was a couple weeks out of a quarter, and then it edged back down to average just over 900. We've been solidly over a terabyte this quarter. I think all but two weeks have been over a terabyte, and we'll probably finish at, I don't know, what Dave would think, a terabyte and a half or something like that. It's pretty strong.
Again, a lot of that is because, to the other gentleman's point, it's both in terms of what's going on in client and in the cloud. The cloud companies are pulling a lot more 4 terabytes than they are 2. Interestingly enough, the notebook business is really strong at 1, and the client business, Katy, is also really strong in 2, 3, and 4 terabytes. It's interesting in terms of the overall lift of capacity. In terms of media, that's more of an issue around notebook. Because the biggest constraints are around glass. I guess there's probably constraints on aluminum ultimately as well. On the glass substrates, that's where we would see the first constraint, and if business stays this strong, I think there'll start to be some challenges around glass substrates.
A couple over here, and let's get to the topic here, sir.
Mark Miller, Noble.
Yeah, Mark.
Hi. A couple years ago, you demonstrated a HAMR drive. I'm just wondering the status there. I've been hearing talk more from one of your competitors about a transitional new head technology. I'm just wondering if this new, I think they call it dual tracking type head, I'm not sure.
Yeah
if it, you know.
Yeah.
Are we going to see areal density re-accelerate, and where is HAMR?
Dave's going to cover both of those topics in his presentation on technology. I'll just defer till that. Great. One more over here, and then let's get rolling. Yeah. Right up here in front.
Thanks. Sherri Scribner from Deutsche Bank. Hi, Steve. You made some comments that December you expect strength to continue. If you look at street estimates, it's assuming that your quarter will be up sequentially in December. Is that realistic considering we've got the extra week in September at this point?
Yeah. Pat can't wait for the extra week to go away. The extra week thing is I think a little bit confusing only in that, again, our OEMs don't have an extra week from their perspective. It's just a quarter, and since Seagate is relatively more weighted to OEM than our competitor, we feel that the 14th week, net net is not clear to us it's a huge advantage. We may pick up some incremental revenue in distribution, but of course, we carry the expense for the whole week. I'm not going to make a big statement either way around it's the extra week, because I think the extra week is kind of irrelevant in both quarters.
I think the potential in December is if the strength in the nearline continues with this kind of base of client strength, then December is going to be, I think, an okay quarter. The problem with December is you never know if the year's going to end on December 15th or December 31st. What's tended to happen, though, in periods of strength is it's carried on pretty solidly through the end of the month, in which case the industry does pretty well. In times of slowness, that's when people kind of shut the doors on the 15th and say, "We'll see what happens." Right now it's feeling like we'll have a full quarter, in which case December is usually a stronger quarter than September.
The other thing that's going on is Golden Week is at the end of this quarter, too, and there's lots of speculation around what does that mean? Is it good? Is it bad? It's like that's just micro stuff that is frankly irrelevant to the company on a long-term basis, but might have some tiny effect one way or the other for this quarter or next quarter. Thanks. Okay, thanks everyone, let's get onto the show. Again, what I'd like to do is just set the overall framework for how we're thinking about things strategically. The good news is we've reflected on the presentations over the last couple of years, and the trends that we set out a couple of years ago we think actually have played out fairly consistently with what we anticipated.
The advantage, of course, that we have now is we have two or three years more data about what's really happening. I think we have a greater confidence about what the opportunities are for the drive industry and for Seagate specifically. That's what we really want to articulate today. That's really what's driven some of the activity that we've had around either organizational issues, like bringing Jamie on to head the cloud business, or our acquisition activity in the last six months, which both Dave and Jamie will talk to, both with the LSI acquisition and Xyratex. Just to frame it for you, again, what we've tended to do is start this presentation with a review of the last five years of performance of the company, a five-year over year view. We've done this on a rolling basis effectively.
You can see that overall, the financial results have continued quite strong. If we took this back to fiscal 2007 to fiscal 2014, lots of different histories that was occurring in each one of those fiscal years or any one of the rolling five years therein, the company's financial performance has continued to be pretty spectacular, certainly relative to the drive industry pre-2008, 2007. Certainly we would say even against a lot of industries and certainly a lot of technology companies. Again, a lot of operating cash flow and good growth in earnings. Really the capacity for drive issue does seem to be accelerating again. Even at 24%, for that time period, that was in excess of areal density capabilities.
As you all know, if we're driving capacity growth above areal density, that's a situation that obviously allows us to absorb more heads and disk, and for a manufacturing company, that's a good thing. Okay, just to frame what we think is going on strategically, this little visual, which we continue to rework, just sets our framework. Again, I just wanted to make sure that everyone in the audience has an understanding of how we think about what's going on in terms of IT trends and what the potential opportunities are for Seagate and really for the drive industry. If we start at the top, cloud, I think a few years ago was probably pretty obvious, but hadn't been widely deployed. Clearly, I think at this point, people accept that cloud architectures are going to be the architectures of the future. What does that mean?
Scalable, reliable, the debates around, is it going to be provided by OEMs? Is it going to be provided by just a couple of cloud service providers? Is it going to be all public? Is it going to be public and private? I think all these things are starting to resolve themselves fairly clearly now that there will be a mix of hybrid public/private clouds. There will be a number of very large cloud service providers. There will be roles for the OEMs to play, and there'll be roles for distribution to play, and there's going to be roles for startups to play. From a Seagate perspective, it's more about do we have the capabilities, both technically and go-to-market, that allow us to engage with all of those players as they formulate their particular strategies for the end customers.
Both Dave's presentation and Jamie's presentations will take you through the details of what we're doing, both at the device level or just above the device level or at the systems level, to basically facilitate this trend towards a more scalable, reliable, and cost-effective architecture. Embedded in all this, I think the accelerator that we can see now that we anticipated a few years ago, but is actually coming to the fore now, is what I call software-defined SDS. The concept of basically now having this pool of assets, whether or not it's compute or storage or networking or application software, and being able to dynamically manage that infrastructure. That's a huge shift from when you had to deploy some particular rack with server and switch and storage with some particular piece of software to solve an application in an environment.
Only if an application was at a certain scale did you get big economies, and then the problem was if it scaled beyond that, you had a big step function investment that you had to undertake. The advantage that we have now is that those assets are just a pool of assets that with software, you can basically reallocate and redimension the assets that you have dynamically to then either affect a particular application or a workload or a time or a place. The efficiencies around that and the implications that has for cost of compute are significant. Cost of compute, we believe, will now come down by an order of magnitude or more, which means application space and application availability explodes. As a device manufacturer, that's very important.
At the end of the day, if cost of compute comes down, hurdle rates get lower, more systems are basically deployed, and basically more parts are needed. It's not about the cost of the hardware, it's the cost of the running the hardware. It's that software and services labor on top of those core components that are really creating the opportunity. This is why I think a lot of research is now showing that where value is attributing in the technology segment is either at the component level or what I call the technology owner level, or at the software systems and service layer. It's the integrator that basically has to reposition themselves to take advantage of the new environment. Mobile, obviously, just the ability to access data or entertainment or whatever, or communications anywhere, anytime from our pockets.
I think that's a fairly obvious trend that's with us. 2.5 billion or 3 billion connected users in the next couple of years is now not out of the question. A couple of years ago, it was a hope where we're going to get from 1.5 billion to 2.5 billion. Again, the implications for having these devices that require us to be able to access data readily or to basically do capture of high-motion video and high-content video and send it all over the world has enormous implications for storage away from what's in that little device. We've said for a number of years, we continue to say our biggest concern about flash is not that they stay on the areal density curve or the linear density curve they're on. Our biggest concern is they don't.
If they don't, they don't continue to sell flash cheaper and cheaper, which means these devices won't have 128 gigs, which means people won't take 4K video and capture it and throw it on a disk drive. The good news right now is people seem to be able to stay on the linear density curve, but it's getting harder. We watch that in terms of what is the implications for overall growth, but also is there a role for disk drives to play either in the high end with hybrid drives or as a caching device throughout a network. Consumerization of IT is this trend that says the infrastructure that's supporting compute used to be very well-defined between what was deployed for commercial versus what was deployed for consumer. Those lines now are eroding rapidly, and the infrastructure is being leveraged across both environments.
That again has an implication for us as a device manufacturer. It has an implication for us now as a systems manufacturer. Again, the advantages of scale and leverage, I think, are the keys, and it's one of the reasons that I think the cost of compute is coming down is because you get these applications that can initially be driven by consumers at very large scale and then deployed into enterprise more affordably. Internet of Things obviously is the idea of expanding beyond the devices we think of as mobile in terms of cell phones and tablets and notebooks, and basically having communications and compute capabilities in any number of devices.
The explosion that that creates in terms of how do you manage it, how do you measure it, what happens with all the data that gets collected, and what does that mean from an overall systems perspective, and what does that mean from a device perspective? The final one, which we haven't really talked about before, but I think it's a trend that's emerging pretty quickly and one I think a lot about, especially in terms of where do we spend our time and focus, is the fact that I really believe that the content world and the tech world are actually really intersecting now.
For those of us who've been doing this for a while, we've seen a couple of fits and starts of this over the last 15 years in terms of what's the role of content, or I might even say L.A., versus what's the role of tech, or I might say Silicon Valley and New York, and actually New York playing in both areas very well. What's that intersection all about? What's the benefit of it, and what's the implications of it? I think now it's real. In fact, I argue that there's maybe more interesting things going on in technology in Los Angeles today than there is in Silicon Valley.
Because the role that content plays, especially on the consumer side of these applications and infrastructures, which again implicate ultimately commercial, is so significant and so much of the content is controlled in L.A., that how content is being deployed through technology is at a completely new level. Given the academic resources between USC and UCLA, they're pumping out these engineers that are thinking about this stuff all the time. This new mantra of how do I take technology to distribute "old business models" is just phenomenal, and it's what's so different about 2000. I think it's important for all of us to start paying closer attention about what is it about Comcast or Time Warner or News Corp that implicates Seagate or Apple or Microsoft. We're spending a lot of time thinking about that.
We think it's obviously a good thing because that form of content is extremely rich, and it moves around the world very quickly. Circling all these trends are these other things, machine-to-machine security and big data analytics. I think the one that we, Seagate, continue to underestimate, just because the math is so hard, is the machine-to-machine side. Again, as the architectures become more leveraged and the software becomes more sophisticated in terms of its ability to manage how machines talk to one another and how they create data on their own and analyze it on their own and then suggest changes. Of course, in the trading world, this has been going on for a long time. In the consumer world, it's starting to get really interesting.
The potential for how much data is created by machine to machine versus content that's physically created by us, either as entertainment or spreadsheets or other things, is most estimates are saying now as much as anywhere from 30%-40% of the total zettabyte requirement in 2020 could be machine to machine created content. This is an area that we're getting more involved with with some of our CSPs and some of the OEMs around what does that mean in terms of the device technology, where is the CPU? What does it look like? How does it communicate from drive to drive, if necessary, without going through a server? These are really exciting things. Security, obviously, the implications for security on protecting data, at the end of the day, has a software implication, has a systems implication, has a device implication.
We're getting more engaged there. Obviously, big data analytics. At the end of the day, to the extent that people get more value out of analyzing this complex, unstructured data, there's a greater need, therefore, to keep it, and that's a good thing for the drive industry. A lot of our efforts are really focusing on what happens in the video world, what's happening with high definition or 4K capture video, and what are all the applications that can come off of that that create value. Value can be meaning selling more stuff, or it could be meaning keeping the world safe.
To the extent that we can help facilitate those technology transitions, that's a good thing for Seagate because, again, it causes models then for people to say, "There's a reason for me to keep a couple of exabytes of this particular content because I can do something of value with it." This is the world we see. It's a world that's on trend line to what we described a couple of years ago. It has interesting implications for us, in all of our business lines and in our new business lines, and it should be the framework that you keep coming back to if you're asking questions about what do we think Seagate's going to look like in the future, or why is Seagate doing the things it's doing today.
Just one final slide, and then I'll get it to Dave to take into the specifics about our core technologies, is with this explosion of data and compute around the data, really every layer of computing is under change. Whether or not it's the component layer, the device layer, the systems layer, the data center layer, they're all benefiting and challenged by the new trends and the new architectures. A lot of the focus in the last year or two has been more on the data center, I'll say systems like the cloud. There's been an obsessive, oh, the cloud, what does it mean, and who's winning, who's losing? I think that's all great, and there's lots of interesting things going on there.
The client side is also really, really interesting because when you think about the potential of what the client can become in this world of a highly leveraged pool of resources, it's completely different than the opportunity the client had before. It's super high value, it's complicated, it's technology-rich. I think it's important for all of us as investors or shareholders, and certainly Seagate as a company, not to lose focus on the fact that while the current trend is about what's happening in the data center, just like what happened with tablets, it trended up to tablets and then to big phones and now to notebooks. This is going to have an implication for client, and there's going to be a trend back to technology and applications and sophistication around the client, which is going to have an implication for us as well.
It's easy to focus on the right-hand side of that chart, but we actually think that the device category is getting really interesting. As a technology owner, or what some people call a component owner, we actually think, obviously, that we're the key building block to making all this happen, in terms of fundamental technology, and then again, software and delivery of software on the application side, and it's the merging of those two technologies that we're particularly excited about. We think the drive industry is uniquely beneficially positioned for, in part because our technology is so important, in part because it's a consolidated technology, and in part because unlike most industries that are going through a transition like this, normally your core is challenged, which puts all sorts of pressure on you in terms of can you actually make investments to go to adjacencies that are high growth.
In the case of the drive industry, the core is fundamental to the growth, and it's in a consolidated industry. You see WD that made a bunch of announcements this week, and even though we don't agree tactically with what they're doing, I think strategically it talks to the point that the drive industry has a lot of opportunity around how do you take advantage of the shifts in compute and architecture. We happen to have a different tack. It doesn't mean necessarily one's right or wrong. It just means that both of us see a big opportunity. They're pursuing it with a particular set of objectives that maybe leverage their competencies. We're pursuing it with ones that leverage our competencies. I think the drive industry is uniquely positioned. I think the silicon industry is almost as well-positioned.
They're not quite as consolidated as we are, their technology is fundamental. It gives them a lot of leverage points. Their model is, again, one that's actually growing as a result of these transitions not being challenged. I think that's what's interesting about the phase of technology that we're in today, if you compare it to when we went from mains to minis or minis to PCs or PCs to network. It's why we're excited. It's why we've been able to attract people like Jamie and Phil Brace and a couple of other people that we'll be announcing shortly. The company is really on a pretty positive trend. We're attracting some incredibly talented people. We think the opportunities in front of us are multibillion-dollar opportunities that are accretive. We think it's absolutely critical that we pursue those in a rational and managed way.
I'm going to introduce Dave, have Dave talk a little bit about the core. Even in the context of the core, that message is getting bigger now that we own a lot of VLSI capabilities. We'll take a break, and then Jamie will talk about systems. Great. Dave?
Thanks, Steve. Good morning, everyone. It's good to be back in front of this crowd again. Pat and I were talking before the session started this morning, and we were saying, "What's our message again this year?" It's almost the same as last year. If you participated last year, welcome back. A lot of the same themes inside the core business relative to investments in technology, where we see the markets going, and so on. My presentation's split into two sections. The first one, I'm going to talk about storage market dynamics, what we see going on, and how it's building inside the industry, and what are some of the key watch points, and then I'll talk about the technology at the end, answer some of the questions that came up earlier as well. You saw this slide from Steve.
The way that I would break this down is really to start to say, from a market perspective, how do we look at each of these and say, how do we forecast how many units, what type of units we're going to actually be able to produce in order to satisfy these trends? There's also a technology angle to it as well. What kinds of technologies do we have to develop to really enable each of these different applications, if you will? Or how do we leverage our core competencies to be able to come up with a one-size-fits-all for as many different segments as we can to be as efficient with cash and so on? It's a complex world. That's the main point takeaway of this slide.
To just start with some IDC data right now, this is not Seagate data, to tell you where we think we're going in the near future. Six years away, 2020, there will be about 44 zettabytes of data being created in the world. Today, it's about four zettabytes. If you believe in this kind of growth, you can see the pressure that's coming at us and why we're seeing exabyte growth that we are, that Steve made reference to before. It's the same story. We're actually, for the last year, there were some doldrums, and I could speculate as to why that happened, but we really see that it's going to continue to break north and stay on the curves that we've been on for the last 20 years. 44 zettabytes of data that's actually created in the world. Now, how much of that gets stored?
The IDC data says 13 zettabytes. The install capacity, if you will, the capacity of the hard drive industry and some of the other industries to actually solution this, will be about six and a half zettabytes. You can see the supply and demand break that we have right here, there's a lot of devil in the details. One of the biggest trends that's going on, and it wasn't true probably five years ago, is a lot of this is going into cloud drives. Cloud storage is the thing that's really exploding. Not necessarily the number of units then that the hard drive industry would ship, but the number of heads and disks we have to make in our factories, which is roughly two-thirds of the capital allocation and two-thirds of the R&D that we have to do.
Those are the metrics that we have to look at as we plan for this. As you break underneath these numbers, now switch a little bit to Gartner because Gartner doesn't actually forecast out to 2020, IDC didn't have a chart like this. This is the forecasted growth of two different technologies that can actually solution that space. The two biggest technologies, actually will. You see the hard drive growth in exabytes up against a pretty big curve. If you look at calendar year 2014, 535, that's probably trending right, maybe even a little bit low. Calendar year 2015, you can see the just year-over-year growth that we're going. Big step up as we see some of the technology transitions forecasted out there in 2017. They're looking pretty good right now.
This is the capacity of the industry under normal circumstances with a 6%-8% capital investments that we would make and so on and so forth. You see NAND. This is the growth of NAND and how many fabs and what technology nodes that they'll be transitioning through as well. Keep in mind that not all of this NAND is actually the same grade, if you will. Some of it's consumer grade and some of it's compute grade NAND, which is needed in some of the higher-end applications. I'll talk about that a little bit more. As we really think about the spaces that we have to solution, I think this gets back to one of the points Steve was making. The cloud is really differentiated from the consumer and the enterprise right now. The cloud, the hyperscale architectures are dramatically being changed.
It's really hard as an end purchaser of cloud gear right now to find out where exactly things are going. There's no incumbent architectures, if you will. It's really hard if you're designing in a period of architectural disruption, if you're designing boxes to know exactly what to make. There's a lot of stuff going on in hyperscale. The client and the enterprise, the mission-critical, if you will, spaces tend to be a lot more stable right now. I contend that when we figure out some of these lower cost architectures that'll actually come out of what's going on in hyperscale, they will ripple across to these other segments, it'll be really interesting. You can see the way we break these things down, it really comes down to workload, there's lots of different workloads, even inside of the cloud.
There's an archival workload in the cloud, and there's a 24/7 banging on the drive all the time workload in the cloud, just a lot like mission-critical. There are big block transfers, let's say object storage transfers with enormous size blocks that are going down to drives. Yes, there are still those little atomic units that I've been complaining about before, the 512-byte block. They're really more memory kind of bite sizes, if you will, rather than blocks that are more conducive to be stored on rotating storage. There's a lot of different architectural implications of all these different applications, that the cloud is where a lot of these things are happening. I will say just really a couple of key points about big data and high-performance computing just to have everybody take it away.
The way we see it already is that most of the data that's actually hitting the drives in those applications are already the long sequential data. The drives that are being used today tend to be in workloads that are streamed onto the device, and the device is performing quite well in those environments, getting the data back into the host, if you will. The hosts today don't tend to do very small block transfers to the drive. Already, people who are doing architectures in these spaces have already taken advantage of memory tiers way above them. Let me talk about storage technologies as we think about those workloads, and I will come back to these memory tiers as well. This is an IDC chart that really says, how are we going to solution that, they said, 6.5 zettabytes worth of data.
Who's going to be able to provide 0.5 zettabytes, 2 zettabytes, 3 zettabytes? What's the other technology that's in play there? For example, if you take DRAM. DRAM, my estimate, would be about 12 exabytes of DRAM is what the world's capable of today. It's not growing a whole lot. The DRAM market's pretty stable. You can do the math on that, but it's not going to go by a factor of 10. I think we all understand that. DRAM is used even on storage layers for really fast input-output from the devices. Then there's NAND. Then there's HDD, and you can see how big HDD is relative to units shipped and the capabilities that we have. Then there's optical and tape that are further on the more archival spaces, where the data's less active and in play.
Then there's cost that has to ripple across all these things. A good way to think about it that IDC has come up with, we put a little green box around these three technologies on the left, and the reason for that is that we actually ship a lot of DRAM today on our box. We actually ship a lot of NAND on our boxes today. Seagate's pretty active and has been for many years, managing a little thing that looks a lot like a PC or a compute node. We've got a microprocessor, we've got memory on board, and we have the mass storage. At our level, we've been managing that pretty well. Now, in some of the slides that you'll see and some of the discussions Jamie's going to talk about, you'll see we're stepping out with some of the same technologies to the next level.
As far as DRAM, NAND, and HDD, how they play, how they interact with one another, it's very critical when you design architectures to be able to find what the right solution is for the right application set. DRAM's obviously the highest cost of these three technologies, but it's also by far the highest performance, if you will, into and out of that memory location, by far the highest performance. Like I said before, the hard drive industry has been using a lot of DRAM for years, not a lot on each, say, four terabyte or six terabyte drive, but nonetheless, when you aggregate it across 600 million drives, then you get a lot of DRAM use per year. NAND is coming in as a technology just like it is in many other spaces and making it very easy for us to grow the capabilities of our caches.
Make our product, depending on the market segment, a lot better. We're taking a lot of advantage of that at Seagate. We just did an announcement yesterday, I don't know if you've seen it, that we've shipped over 10 million hybrid disk drives now. This quarter, we talked about this on an earlier call. This quarter, we're actually going to ship over 3 million. We're getting good market traction, and I think it's still our job in engineering these devices to be able to make these caches cheaper and cheaper. Make the end product run better, be able to solution if workloads do trickle through, and they do, that are still the 512 bytes at a time, that we can use that cache for that and a host of other things to be able to make our devices better.
Better power, reliability, lower cost, so on and so forth. It gets into just plain hard drive technology, which I'll talk about now. Just last year, I said we'd be at 20 terabytes by 2020. We're still on the same plan. I'm going to go into the technology as to why that is, but it's a nice round number to be thinking about. The subtle change this year is that we think it's at least. We definitely think we're on this plan, and we think we can actually beat it. As an industry, there's a lot of optimism, I think, out there right now. At The Magnetic Recording Conference that just happened about two months ago, all the companies and capital equipment suppliers and university professors all get together at this TMRC conference, and talked about where they thought various technologies would be.
There was a question already about how's the hard drive industry solutioning this. Let me talk about them. These are the most popular ones and the ones that the poll of the participants in that consortium actually came up with as the most likely paths that we'll be going down. First, shingled magnetic recording. Seagate's been shipping that for a couple of years now. We've shipped almost 3 million shingled drives. We talked about that a little bit in this conference last year. We actually have learned a lot from this segment, from being able to ship the shingled magnetic recording drives into the various markets that we participate in. We learned how to properly address the shingle. It's a very different world than historically we've been in.
We've learned how to properly, if you will, hide that by various layers of tiering and caching from the end-user experience that might have a problem because it's shingled. There's a lot of FUD out there about shingled right now and how end users will have a problem. In all the applications we're shipping, we're not seeing complaints. We think we can even get better at that as we improve our caching and tiering algorithms. We're pretty excited about shingled. We'll just call that one done and being used in various market segments. Not every market segment, but we may choose to do that over time. The question, I think specifically, was about two-dimensional magnetic recording and HAMR, what we call Heat-assisted magnetic recording. Two-dimensional is really thinking about readers and read channels.
How do we actually get a bump by being able to scale our read geometries even smaller using a couple of readers or an array of readers, if you will? It's actually going quite well. I talked about it in this conference last time. We're still on the same plan, about 2016, for this technology. We're fairly conservative in what we're estimating with what kind of bump we'll get with this. Suffice it to say that the technology demonstrations in the last year have actually been quite good in the labs. HAMR's the one that the industry's been talking about for a long time. It's the next big speed bump that magnetic recording would go through. Been huge investments in this, and all I will say is the last fiscal year was a really good year. It's had its detractors.
I've probably even been one of them from time to time as the technology development a few years ago was really slow. In the last year, and we reported this at this conference that I made a reference to, in the last year, we've actually shown areal density demonstrations that are higher than anything we can do conventionally. We seem to have made a lot of progress on the reliability aspects. To put it in a nutshell, you have to get a 20-nanometer laser spot on the disk, and then you're writing it as you're flying 80 miles an hour down this track, 20 nanometers wide. That's the challenge. We were able to do it five years ago. The problem is the thing burned up after a few nanoseconds. Now we've been able to demonstrate over 1,000 hours, okay?
That's not quite productizable yet, we've got a little bit of work to do still, but it's orders of magnitude of improvement in the reliability of the technology and while demonstrating that we actually can make this progress in areal density. We think we're on the cusp, and productization is still out there at the back of the year. We also think that with some other tricks that aren't even on the chart, we're still going to be able to migrate the industry so that we're not waiting for a big step function, but we'll be able to continue to release drives in the meantime. Another big piece is the electronics components that we haven't really talked about this very much in the past, and I don't think people have an appreciation of hard drives, the entire hard drive industry, what Seagate does in particular.
I'll just tell a really quick cartoon of a computer, if you will. I think it's instructive to have this architectural diagram to just highlight a few points. First of all, if you look at the bottom, Seagate's products have almost all been-- All 2.3 billion drives that we've shipped to date have all been hanging off this SATA or SAS bus, the storage bus that's down at the bottom. You look at that HDD, that's right in our bailiwick. Now, the hybrid drives, the 10 million that I made reference to earlier, those are off to the right of the diagram. Then there's been SSDs as well. We've shipped some of them. The volumes of SSDs are still pretty low in the industry, but there is a compelling technology play in a lot of different market segments for the SSD.
If you think about this at a really high level, the electronics that has to service it, let's just talk about what's on the devices themselves. We have these controllers. Seagate, for example, has controllers that we've got a lot of IP in that have been able to address the SATA or SAS bus for many, many years. Fibre Channel before that, PATA before that, so on and so forth. We've actually invested a lot in that. As we went off to do hybrid, obviously, we have to be able to control NAND. In order to control NAND, we had to do our own development, then ultimately, when the LSI acquisition came up, we said we really want to control NAND better than anyone else.
We got an asset inside of the LSI acquisition that was really what are called flash controller division, the FCD. If you think about that, it's kind of the orange box that sits over the top of the SSD box. Architecturally, I believe that over time, a lot of things are going to go to what we call hybrid. As a matter of fact, we're starting to think about it too much of what's NAND and what's HDD. I think we should think about it as what solution for which market that we're actually trying to engineer. As you think about it that way, there may be more NAND and there may be no NAND, but I think all the architectures have to be able to go anywhere.
Another subtle point, though, is that one thing that the LSI teams had also done, which is really compelling business, was to take some of these controllers and gang them behind a bigger controller and build what's up in the green, labeled PCIe card. That demonstration is out, actually, in the back of the room. If you get a chance on the break, please go ask questions about it and go see it. It's actually a totally different space. It's a lot more about the compute space, it's a lot more about applications acceleration, firmware, working with the ultimate partner. It's not just a hardware device. It's not just a piece of silicon or a printed circuit board that's stuffed with some silicon.
It's actually a lot of firmware and architecture around controllers that you actually have to do to optimize to the specific application that you're in. Companies there, like cloud service providers, are very interested in the Nytro product that's in the back of the room, and we're working well with them. If you think about the LSI acquisition in these two phases, if you will, the FCD division, the flash controller division, and the accelerator card division, this is a good pictorial way to think about it. The way we look at it is we, at Seagate, pull stuff out of Mother Earth and build the technologies from the ground up. We actually talk to smelters and we use 51 elements of the periodic table in today's disk drives, and on and on. I can give you those stats.
Not a lot of people are doing that kind of base level things. Whether it's heads or media or other components, we actually have to care about those core elements, building the devices. All the devices are not the same, different markets, different devices, different performance requirements, reliability requirements, power requirements, so on. You can see them across the bottom. We're actually still doing some very interesting work to develop even those core devices with the hybrid drives that I talked about, and then with this Kinetic product, where we've actually, instead of SATA or SAS, we've actually put an Ethernet interface. What's really exciting about that to me is we can actually break free of some of the stuff that the host has traditionally done wrong.
We can actually talk in a different language, what's called a key-value language, across that interface, and abstract even more some of the things that the host doesn't really need to be doing wrong. We can also take some people out of the value chain that are not providing necessarily value in some of those applications. For example, if you're spending a lot of time translating data as it's going down to the drive, time and money, then it's wasted in that application. This key-value architecture across an Ethernet bus may be a much better way to do it. We've got some customers that are pretty excited about that. Anyway, we stay focused at the blue level in the core. We've been doing that for a long time. I've made the argument that we've been building systems a long time.
They're called hard drives. We have a lot of technical skills in that, and we have a lot of supply chain and customer relations and so on and so forth that's leveraging that. We're going to take a break, and when we come back, then Jamie is actually going to tell you about what we're building on top of that core foundation. We'll call you back in a few minutes.
If everyone could please take their seats, we're ready to get started again. Thank you. We'll have more time in the demo area afterwards, and the reception will have executives back there to speak with you as well. At this time, I'd like to turn the stage over to Jamie Lerner, who's our President of Cloud Systems and Solutions. Jamie, please go ahead.
That's right. Thank you. Hi, everyone. I'm Jamie Lerner. I'm the President of our Cloud Systems and Solutions business, and I've been at Seagate now for five months. I was joking with Steve that I've had almost an out-of-body experience. I'm really more excited today than the first day I joined here because I've gone on a pretty interesting tour, if you will. I first met our board about my third week in the company. The board asked me to come back in 90 days with a strategy. The first thing that I did is I went out and did just what comes naturally, and I met with our customers. I sat with the world's largest storage manufacturers.
We went out and talked to telcos, we talked to cloud startups, we talked to cloud operators, we talked to oil and gas companies, retail companies, and all of them came back and told us two things. They've come back and said, "The architectures that we use to collect, manage, and analyze data are fundamentally inappropriate. They're just not working. They're not scaling. They're not usable. We have to shift to totally new architectures, predominantly cloud-based architectures." The second thing they said is the imperative for them to analyze data is becoming the core competitive plane for their business. When we talked to retailers, it was analytics that they were going to use to compete. When we talked to oil and gas companies, it was analytics and data, it was how they were going to find oil.
When we talked to our defense contractors, it was analytics of how they were going to train and operate the next generation of war fighters. The thing that came out of that is what surprised me the most. Our customers came back with almost an assumption. They came back and said, "We are assuming, there's an implication that we want Seagate to help us solve these complicated business problems." They want us to evolve from supplying core storage technology to helping them solve their most complicated business problems. I was so surprised by that. I came back from our customers, I dug in internally, and what we started finding is we have an unbelievable amount of capability to actually do that.
What I'm going to share with you today is a series of programs that we're going to be running to begin to align with a series of new architectural plays that we're coming out with, and a set of market adjacencies that we're going to begin to exploit. I have to tell you, I believe more today than the day I joined that we have an unbelievable amount of capability and technology that we've never expressed, and a hard disk is probably going to be the wrong way to express that set of capabilities. We're going to continue. I often refer to Dave Mosley as the Michael Jordan of building disks, and we are not going to stop that. We're not going to take our focus off that. We're going to begin to layer on a series of new adjacencies to build our business.
What I'd like to ask you today is, I'm going to ask you to think about Seagate in a very different way. This is a company that is changing, and we're not just a component supplier. Again, we are the Michael Jordan of supplying components, we're going to begin moving to be a company that helps our customers solve their most pressing business problems. That's a big change for us, and I'm going to walk you through how that's going to develop. This is the model that we as an executive team are thinking about. This is the model where we are looking at our core business and beginning to find market adjacencies by huge market transitions, the transition from traditional computing to cloud computing, right? The transitions from traditional telephony to BYOD.
We're looking at $50 billion plus market transitions, we're getting out of our laboratories, we're getting out of our design centers, we're sitting down with our customers. In sitting down with our customers, we're saying, "How can we come out and design a new solution with you?" Whether it's the large storage OEMs and ODMs that we've worked with over the last 30 years or the world's largest cloud companies, we're sitting down with them and jointly designing totally next-generation systems with them. We're innovating together. We're working on patents together. We're designing next-generation solutions where we're integrating the disk, the storage enclosure, the firmware, various different expertise, and the software together, and open-source components to solve new problems.
When we get that right, we're now working with those huge storage companies that we work with to accelerate across that when we build this solution, we're going to show you some today, we're going to walk you through some case studies, that we can then accelerate tens of thousands of customers across one of these transitions. The transitions that we're looking at and the use cases are these six. I'm going to walk you through a couple of examples. One of them, one of the core use cases is performance, an imperative where we have to be able to manage process data faster than we've ever done before to solve a very pressing business problem. There's a system in the back of this room called ClusterStor.
It is used by the three largest oil and gas companies in the world to look for oil. The stories that we're beginning to tell about us as a company are fundamentally different. We're now engaging with these oil and gas companies and helping them find oil. We're doing that because a drilling platform is actually a gigantic data collection engine. There is a wellhead that is drilling and is sending back petabytes of data. There are as many as 12 boats around a drilling platform that are doing different geospatial analysis. They're very often dropping different forms of explosives, pulling data off the ocean floor, they are pulling in petabytes of data. The speed at which that they can collect that data, transfer that data, analyze that data, determines their success in finding oil. They're partnering with us to do that.
That's a very different body of work for our company. What we've done with the ClusterStor solution at the back of the room is we've taken our media, our disks, the firmware. We've specially optimized them for that use case. We've built boxes that you'll see there, enclosures with specialized firmware, specialized processing, specialized embedded computers to do analytics work, predominantly through using things like MapR, Pig, and Hadoop. We've built a whole rack system where these racks can be put together as a hive, as a completely converged infrastructure that's completely horizontally scalable to help these large oil and gas companies find oil faster. That translates to a very different profitability equation for those companies. That's very different work that we're doing. We're still doing custom manufacturing. We've acquired the Xyratex business.
That is a 22-year-old business that has been working with the same customers that we work with in our disk business. We help them build their enclosures. We work with the big six or eight storage companies. There's something changing there, too. There are large cloud companies that are coming to us saying, "We have such large data centers. We are processing so much data that we actually have enough volume for you to build a specialized box just for us." We're seeing a new breed of customers coming to us saying, "I would like you to build a specialized box for a very specialized application in our data center, we can't buy that from the traditional storage vendors. We need you to build that specifically for us." We also have a high-performance computing business.
We work with the largest genomics companies in the world that are actually mapping human genomes, genome sequencing. We're working with the largest governments and the national labs. We're working on a large system. One of our latest wins is a system called Blue Waters that's doing research for the U.S. federal government, going out and doing research about how does disease spread. Can we predict earthquakes? Can we predict weather patterns so that we can do more economical farming? They're going out and doing research on behalf of our nation to figure out how do we organize as a society more efficiently. Those are fundamentally different types of systems that we're doing, again, using similar technologies that you can see at the back of the room. We're helping companies build clouds. There are companies coming to us saying a cloud is composed of three things, right?
Storage, network, and compute. The storage piece represents a very large challenge. It is an architecture that has to fundamentally change to work in cloud. It has to be multi-tenant. It has to be elastic. It has to be able to be stretched, shrunk, moved between locations, and the traditional architectures don't support that well. We're beginning to take technologies like Kinetic, putting it into our systems, working with the new open-source vendors, working with the Ceph, OpenStack, but also commercial vendors, Scality, Amplidata, and others, to begin help companies stand up these clouds. We're not doing any of this work alone. We're doing all of this with the same co-travelers that we've worked with for the last 30 years. They're working with us on all of these solutions.
The same people that we sell disks to today, they're partnering with us to build these solutions, partnering with us to take these to the market. We're in no means, and some people often think, are we competing with our traditional partners? Not at all. These are just new set of solutions that we're doing to partner with them and bring into market with those co-travelers. To do this, we also have to be open to being multi-vendor. We obviously want to put our disks inside these systems, but there are cases where some of these people want to make sure there's plurality, that there's different vendors in there, and we're completely open to that. We have to be as we move into this space. We're also helping companies do backup and disaster recovery.
We have 25,000 customers that we back up every day in the EVault cloud. We've been doing that for 12 years. We've shipped 17,000 petabytes of systems through Xyratex. We're bringing those businesses, we're bringing them into Seagate, we're integrating them. We're moving through that. We've got several more months to do that work. The backup and disaster recovery system, you'll see in the back of the room, that we've just launched. We're also doing what I would call data lakes. There is an amazing set of use cases for data lakes, where companies are collecting huge amounts of data. A good example is we have a big box retailer that came to us and say, "There are people that come into our big box, some that spend $70, and some that spend $70,000. Those people who are spending $70,000 are often general contractors.
They're in the construction business. They come in. We want to identify them. Can you help us build a system that identifies basically by their cell phone when they come into our parking lot so we can find them and help them, identify them, and get them almost a concierge level of service when they come into our big box?" That is a fundamentally different type of system, and they have a data lake. Just huge amounts of data of what's happening in their parking lot, the cars driving around, what's happening in their stores, what products are being bought in these gigantic data lakes. Then we pull data out of that data lake, analyze it, and extract it, and turn that data into actionable information that these business leaders can use to run their business.
These are the six use cases that our customers have come back to us to say, "Not only do we want you to help us solve our problems, these are the problems we think you're best suited to help us solve." What the business looks like today is a series of these stacks. What we do is we purpose-build the underlying devices that sit within the enclosure for the use case. There are certain devices that we make that are awesome at saving power. One of the things we're doing is taking notebook drives, putting them inside of huge racks, and it's amazing how much weight they save, how much power they save, how dense they can be.
Looking at moving technologies for different use cases, we have Kinetic drives that happen to be very well purpose-built for a converged infrastructure that's running an object store. Someone building a next generation cloud. We have other infrastructures that are very good at building just gigantic data lakes. Specialized disks that are just cheap and deep and big and the ability to store very large quantities of data cost effectively. We have all-flash systems that we're able to go out and say, "We just want the fastest performance, and we almost have much less sensitivity to the cost of that system, the heat, the electricity. We just want the fastest system that we can get." We take these underlying devices, we put them into an array. These arrays are hybrid.
Some of them have disks, some of them have flash, some of them have mixes of those technologies. Some of them reside behind the customer's firewall, some of them reside within our cloud, in the EVault cloud. We allow those workloads based on a policy to move. Some workloads need to be in the fastest place, some in the cheapest place, some, because of security reasons, need to be in a very secure place. We're able to use a variety of different devices that are put into different types of arrays. Those arrays are software-defined. We can use software, and that is the generational change. The software and the hardware are no longer welded together in a rigid system. By removing those two, now you can use the software to take the underlying storage infrastructure and shape it on the fly, like clay.
We can shape it. I want to shape it for backup. I want to shape it for performance. I want to shape it for a database application. I want to shape it for just storing photos. Instead of you saying, "Okay, we're going to spend three years and build a box, and that box just does backup." We're able to dynamically use software to create backup boxes, database boxes, SAP purpose-built boxes that are really good for Cassandra next generation databases, all through software-defined arrays. We then assemble those arrays into racks like the one you see in the back of this room that can be put into a large company, large cloud, large organization at scale. To give you a little bit of shape of this business, we are in our second quarter as a non-synthetic business.
We are on target to do $650 million-$700 million in this business this year. We will move over a $1 billion run rate on this business within 5-6 quarters. We'll have our first $250 million quarter within 5-6 quarters. This is a business that is at scale. It is a business that is growing relatively quickly. It's a double-digit growth business. We're feeling really good about it. It's early days. This is exactly the type of business that we're going to take our traditional storage business and begin to layer these kinds of multi-billion-dollar market adjacencies on top of that and grow our revenue in ways that we couldn't do as purely a component supplier. They're all technologies that build off our core.
One of the things we talk about a lot is we are here to defend the core. We're going to go into market adjacencies that build off that core strength. What do we do differently that no one else in the world can do when it comes to doing this? What is our unique value proposition? Just like Dave said, we dig into the earth, we pull up rare earth metals. We forge that into a disk drive. That is something that only two or three people in the world can do. When it comes to people who are doing that and building systems, we're the only one doing that at scale. We think that is very unique. With that, I'd like to introduce Pat, who's going to show us what this means financially. Pat?
Thanks, Jamie. I want to be able to go down. I've never done this. Finance folks follow rules. I have to stay up here. I'm not Jamie. I just thought that was kind of cool. You heard from Jamie, you heard from Dave, you heard from Steve that there's a lot of disruption, a lot of changes in the storage ecosystem. What that does is gives us a unique opportunity. What I want to show you in the next few charts is really how we model that out. Why are we doing this? If you take a look at Seagate's core business, what are our real strengths if you really break it down? One, we have deep technical expertise. We own the IP, we own the factories, we produce them. We distribute them.
We can leverage that across these platforms. Extensive go-to-market capabilities. We have OEM, we have retail, we have partners. We can take these technologies that we're acquiring and leverage them there. As you've seen for the last several years, it's a very resilient financial model. What does that all mean? We could take these adjacencies and do it in a very efficient manner with the existing infrastructure that Seagate has. If you take a look at what Jamie's talked about, from some of our M&A, if you take a look at what Dave's talked about from an internal investments, all these internal investments, whether it's through our OpEx, whether it's through M&A, we believe over the next couple of years, that's going to generate $2 billion of revenue. That's from our cloud-based and our flash platform businesses. That's a tremendous amount of revenue growth.
That'll be the first thing you manifest in the P&L that you see, is that we're getting that top-line growth at a much higher rate than we would, as Jamie said, traditionally through just HDD. If you fast-forward to 2016, the gross margins probably is accretive level. We don't really quite get the scale of 2016, but by 2017, though, you now have an operating margin that is accretive. I'm going to show those numbers in a little bit, but you take a look at that and say, what does that really mean? These businesses are going to start cash flowing, contributing to the business within two years. That's a lower risk equation, but it's very managed. As Steve said, it's balanced, it's less risky, and it's leveraging existing infrastructure. While we're here at 2015, I'll give you a summary of 2015.
The nice thing about this chart that Steve talked about, some things have played out amazingly well. When we sat here two years ago, we updated our model. We went from a 22%-24% gross margin to a 27%-32%. I can say within those two, three years, we have not missed one of those, and we think that's going to continue. This chart looks almost identical for the most aspect. Revenue growth for fiscal 2015, 3%-5%. That's all in business. We'll have a 3%-5% growth. That's a good top-line growth. We haven't seen the growth in the last few years. We think that growth is actually coming also HDD this year. As Steve talked earlier, we're seeing that strong in the front end, and we see that continue at least through the first part of the year.
HDD this year is not a story of non-growth. That's a story of growth as well as the new businesses. The non-GAAP gross margins, 27%-32%. Last year was sort of the same number all four quarters. I know it was boring to some folks. It was amazing to us. I expect sort of the same this year, we'll stay sort of in that same range. You might get some, depending on some mix, as Steve talked about earlier, you might have some a little higher, some a little lower, but on average, the gross margin should be somewhat very comparable to fiscal 2014. The only thing that's really changed on this chart from what you've seen a couple of years ago, we expanded our range of OpEx.
As you saw from the chart earlier, we're making investments to grow the earnings power of this entity within two years. We think this is a very solid investment. You can see it's not bouncing out of lines on a big way, but it's 13%-15% for 2015, probably with a bias towards the higher end. If we got more scale quicker on some of these other businesses, that might play out. We're integrating two companies right now. We're going to be mindful, thoughtful. As Jamie said, we're going to engage with customers. We're not going to shortcut that investment for the sake of just the model, 13%-15%, we're not changing it drastically, but it's probably on the high end for 2015. You can see the resulting operating margin. Capital expenditures, we still believe 6%-8% is an appropriate level.
You might think that with our operations and Dave's team's done such a good job the last two years that there should be a new standard. What we have been is very cautious of investing into capacity, but we continue to invest in technology, we continue to invest in facilities, we continue to invest in areas that of the new business we have to do that we think 6%-8% is still the appropriate level to maintain this business and grow this business. When we do all that, what does that really mean for capital deployment? I know one of the big stories over the last several years has been capital deployment. One of the things that most folks should take away from here is that Seagate's not a company that husbands and shepherds cash on the balance sheet. We deploy it.
We put it to use. If you look the last few years on the gross cash balance, relatively the same, because we're going to do it in a form or fashion of one of the four dividends. That's probably the fixed one in this because we're very committed to our dividend. This year, about $550 million. We're committing to a 10% increase year-over-year. We've talked about that for the last couple of years. That's at a minimum. We can foresee that for the next couple of years as well. With the cash flow of the entity, we can see growing that dividend. That dividend is the fixed part that we would say, and we feel very comfortable with the cash flow of the entity that we can continue to grow that dividend. Share redemptions.
We're certainly taking a lot of shares off the books in the last couple of years. The market didn't recognize it, we took advantage of that for our shareholders because we just said, "We'll take those shares in." Now we'll look at it as value-driven. We still have an active share redemption program. We still have authorization from the board, but we'll be mindful of that and that'll be more value-focused, as we go through, looking at that. Debt retirement, opportunity dependent. If you've seen the last three years, what we've done is extended our debt and lowered our cost of the debt. It's an amazing story, even to the point we've gotten investment grade. We're very proud of that, but we'll continue to work our debt as much as we work our capital.
If we feel that we need to do something else there, we'll do it, but in the backdrop that we really want to maintain investment grade solidly. We'll continue to work that. In M&A, you've seen us the last 120 days, make two acquisitions. They're opportunity dependent, but at the end of the day, as Steve talked about, they have to really drive value, they have to leverage from the core, and they have to really take advantage of the things that Seagate does very, very well. Yes, they might not be the riskiest ones, but I don't think that's what you want from us because we're not really transforming it in a big bang. We're taking steps at it, and like Jamie said, having the building blocks, and I think he laid out very well how we're doing that.
Normally we haven't gone out a few years, but I want to just take you out to 2017 because I opened up with that chart of saying what the opportunities on the revenue are. As exciting as the $2 billion of revenue are on that, realize we have a large business of HDDs still around $14 billion. It's not going to shape the financial model overnight, but what it does show is the big takeaway of this chart is we're into a revenue growth scenario in a much bigger way. The revenue growth of 6%-8% by 2017 that we see in our numbers, and that's through the core, that's from the new ancillary adjacency businesses, and we see that driving. We also have a point of view that that's going to drive the margin up.
If you take a look at that 27%-32%, it's probably going to move you to the higher end of that range. At $2 billion, it's not enough to really reshape the whole P&L. Over time, and as we continue to update this audience and the markets, you'll probably see us tend to maybe even break out of that. Because the model that Jamie's going is a lot of customer touchpoints. It's going to be a lot of value add, but that probably changes the model both on the margin and to some extent on the operating expense line. At 2017, given the scale that we're getting fairly quick, we see ourselves falling back to our long-term model of 12%-14%, and probably going more down than up. We see that coming forward, and then the resulting non-GAAP operating margin.
Capital, we're not taking off that. If you see the charts that Dave talked about how much zettabytes we need, in fact, a lot of this industry is going to have to step up investment if we get there. The way we're approaching that, we're going to be very cautious. We'll wait till we see it, but we want to do certain smart things to shorten lead times if we get there, but we're not going to go heavily lean into that. We're very mindful that we don't want to lose revenue opportunity, but we certainly don't want to get ahead of it. We'll play that, but we really think that 6%-8% for the next couple of years is appropriate. At some level, we have to make certain investments, to meet the zettabyte demand that Dave talked about, but we'll address it then.
The real meat of this chart is, there's really three things. Revenue growth, operating margin growth, cash flow growth, which turns into EPS growth. The company will be growing earnings and revenue by 2017, and we think it's a very structured way to do it, a very balanced way, and a very risk-balanced. I think we're very, very comfortable with these numbers here. In fact, there might be some surprises that could push that up if we get more traction in some of these accounts. We're really not moving off our long-term model. Like you said, the main framework's there. As these new businesses develop, it could push that model up, which I think would make everyone in this room, including myself and the management team, very happy with that.
Just closing on this, it really sort of where I started. We have deep storage technology expertise as Steve talked about. We're a company that owns IP. We're a company that continues to generate it, and we're going to continue to invest in it. Whether it's through the factories, whether it's through development, we're going to do that. We're going to do that with the new business as well, that's going to be a strong foundation to grow into. Expanding our cloud systems and solutions, you heard Jamie on that. There's a lot of opportunities there, probably more than we can bite off chew right now, but we'll be metered about that, and we'll really attack that pretty aggressively. A resilient financial model. Not a whole lot of changes.
Like Dave said, if you were here last year, you probably saw somewhat of the same model. If you were here the year before, you saw the same model. We're pretty proud that it's been fairly consistently achieved, and we'd expect next year for most all vectors being achieved. At this point, I'm going to turn it over to Steve to be the moderator and invite the management team up for Q&A. We'll have folks with mics, and we'll just take your questions.
Okay. I have the first question. Yes, Steve? If that long-term model says you might be more towards the high end on gross margin and more towards the low end on operating margin. Wouldn't that mean, on operating margin, wouldn't that mean that your operating margin should be higher? Can we go to that chart there? He knows I'm a finance guy. They're not going to give that away. I'm not going to ask you, I'm going to ask you. Can we go to that chart? Are we trying to go to that? You go back one, Steve, I think. Oh, I do it. When I do it just doesn't really do anything. Yeah. If we're really going to be, one forward next one. Seven, yeah.
If we're really going to be maybe more towards the high end on that gross margin and more on that low end on the operating expenses, shouldn't that mean that your operating margin should be higher than 13%-15%? Because if I'm near 32% and I'm near 12%, doesn't it get to 20%, Mr. Luczo? It's an excellent question. The answer is yes, it does. I think it's the opportunity we see, and it's the transition that we're trying to manage, and we think we're managing it in a really responsible way that says, without really sacrificing EPS when you do the models off of fiscal 2014, we have a multibillion-dollar revenue opportunity over the next 2 years, which we think will result in gross margins trending more towards the high end.
Depending on how big that business gets and when it gets big, actually maybe giving us the opportunity to change that range, but also getting a lot of operating leverage, which should actually result in operating margin expansion in the 18%-20% range. When that plays out and how that plays out obviously is a function of the deployment of the new technologies, as well as Jamie's success in taking the technology head, adding whatever we need, and being successful in some of these go-to-market opportunities. I think that's the excitement that we see. Again, it's all being leveraged off of a core that's pretty stable. The other point is if you go back even to the core model, which is what fiscal 2015 is, because again, Jamie's business or the opportunity around these new opportunities is still pretty small.
The leverage inside that model is enormous. We were telling some people yesterday, no matter how much we model increase in volume or change in mix or anything that absorbs more heads and disk, and we obviously have the best models in the world understanding how our factories work at a very detailed level, we're never right. We're never right in that we never quite understand the full positive impact of what happens in a manufacturing company when volumes start to ramp. With very, very little marginal changes to either unit growth or mix inside of this model, being a manufacturing company and the absorption related to it, the cash flow implications are pretty significant. Which is why when you're on the front end of what feels like a period of growth, we get pretty excited about what some of the leverage opportunities could be.
Since I know there was a lot of sharp people out in the crowd, someone was going to ask that question, and that's the answer. All right. Other than that question, why don't I start on this side since we pounded that side early, and go over here on the edge if we can, or.
All right. I have a mic on here.
or to that gentleman.
Sorry. Your question was all right. I want to ask a question on flash, if I could, which I know you get it every year, and I appreciate your answer, but I'll just keep asking it anyway. It obviously plays a bigger role in the market and also in your own roadmap. Two parts, I guess. One is any updates or anything with the relationship from the flash side with Samsung that went right back to your deal a couple of years ago? Then looking broadly, and this is a question always comes up, but it's not going to stop coming up, I think, long term. If you look at particularly things that SanDisk is doing, for example, and Samsung themselves, how do you offset the fact that you're not vertically integrated in flash, you're not going to build fabs?
I do realize you've said in the past, for example, it's not like the media guys want it in hard drive, so it's just one component of the solution, but it is a pretty big component of the solution. Given that Seagate's own history has been so successfully built on vertical integration, given you're not vertically integrated in flash, how do you deal with that? Thanks.
I was thinking that's really. That was like one question had 17 questions in it. Well, I think, first of all, I think that the analyst community, and frankly, the silicon community, has been a little loose with terms about where flash is deployed and why and where is it successful. There's a lot of mixing and matching of saying flash in storage. The reality is that's why Dave did the chart. You have to separate where flash is being used between what's being hung off the storage bus versus what's being hung off the compute bus. SSDs, to the extent that they're being hung off a storage bus, it's a pretty expensive solution for a storage problem. There may be some tier of storage that you would like to have the performance attributes of hundreds of thousands of IOPS.
As flash has gotten relatively more expensive to HDD again, and as the demand requirements for the flash that's required off of the compute bus have increased rapidly, the architectures are actually saying that tier 0 storage that goes off the storage bus is probably smaller than people thought three years ago. When we were here three years ago, we thought probably 5%-7% of the amount of petabytes stored would have to be facilitated by flash at Tier 0. If you talk to most of the big systems companies today or the IT shops themselves and even the silicon companies, that number now, people are saying 2%-3%. 2%-3% of compute-grade flash against 13 zettabytes is like a tenfold increase over the amount of flash available for that market today.
Where are those fabs going to come from exactly at $12 billion a pop? By the way, they're going into a market that services that revenue at 6% capital per revenue, not 30% capital per revenue. I think, again, when you get down into the real deployment of this stuff, when you think about it off of the storage bus, we believe that hybrid is going to fill that gap because it has the same performance, but it's leveraged against all the HDD cost and capacity advantages. Or we play with it with our controller technology, which is why the LSI acquisition was important because we now have a controller technology that's flash agnostic, and it basically allows the people who are integrating SSDs to take a controller, put it on whosever flash is cheapest.
Really a large part of the deployment at the SSD level is done that way today. Even though some of the companies like Samsung or Toshiba have their own devices, a lot of SSDs are still basically people taking a controller, putting it on who's ever got the cheapest flash, and then deploying it into a particular application or customer. I think when you move up to the compute bus, it's exactly right. That's a great home for flash. Really, that's where most of these flash-based software companies play is off the PCI bus. We have that capability with the product backed here, which is from a market share perspective, number two to Fusion, we think from a performance perspective, better. We also have a controller technology that again, that can deploy against anyone's product.
I think the story is less competitive and more complementary than people want to make it, whether or not that's media or investors. I think if you were engaged in the type of dialogues that we have with our flash customers, you would appreciate it because those dialogues are much more strategic about what are we both doing to solve capacity problems that we're both challenged with in order to deliver the performance at the right level for these customer sets. That gets to your first question, which is how is those partnerships going. I think the big difference between three years ago and today when we forged the Samsung relationship is back then, there was this requirement that said you had to pick your controller technology and your underlying flash in order to optimize performance.
It was a risky game because if you had the wrong controller technology or the wrong flash partner, you would be behind whatever other competitive product was out there. It didn't even matter if you controlled the silicon, because if you control the silicon, but you were still late on the controller, didn't matter. Unless you could just afford to price below the market, which some of the big companies do every once in a while. I think today the difference is that these controllers are flash agnostic, you can basically go to the cheapest flash in town and leverage the controller over that.
Again, we need to understand how much exabytes are being produced in these fabs, how much of them are allocated to consumer-grade flash versus compute-grade flash, what's the controller technology goes on top of it, how much of that flash capacity has to go into cell phones and tablets and Internet of Things, then what's left over really. For a market today, that's a $35 billion-$40 billion market. If you take 30% of capital requirements on our market, that means you can build one fab, and one fab outputs about 10 exabytes. That's a real big problem when you're trying to solve 13 zettabytes. We're still, I think, that the change for Seagate has been where we had a very deep relationship with Samsung. Our relationship now is getting deep with all the flash companies.
Hynix, Toshiba, SanDisk, and Micron are all companies now that we engage with actively around technology, around supply chain, around optimization at the systems level, about where do you use all flash, where do you use hybrid. To Jamie's point, we have some customers that say, "I need all flash performance." We have some that say, "I need hybrid," and we have others that say, basically, "We just have a bunch of data that we have to make sure is accessible in milliseconds, not picoseconds or whatever your smallest measure was down there." I think it continues to evolve, but we are definitely still in the camp that this is a complementary technology that has to be solved together as opposed to some marginal competition at the edge. I just think that's silliness. Yeah.
Hi-
I think the market has actually proved it out so far.
Hi, Keith Bachman from BMO. One for you, Pat, and one for you, Steve. Pat, on the metrics that you gave for FY 2015, they were mostly on the income statement. Is there any metrics you can give us around the cash flow and how that might translate either days, cycles, or how you want to think about it? On the buyback as part of that, I think your comment was value-driven. Is the bias now more towards the dividend side over the buyback, or how should we be thinking about that related to the cash flow?
I wouldn't say a bias. I'd say we're committed to the dividend, we're committed to the growth. If you want to say that's a bias, that's a footprint we have that's fixed. I don't know if that's a bias because there's more cash flow. I don't think I'd say let's get more dividend. The dividend's a structured payback. When we look at the cash flow from the entity, like I said, we generated $2 billion of cash. We're not going to sit here and say we'll have $4 billion next year. We're going to deploy it. We're going to deploy it either in through an M&A, a share redemption, or even if it's a debt retirement. We don't really look at it in the sense of What are we doing with the cash? Because those are the four areas.
We're going to invest in our business first, then we're going to go deploy that capital as we see fit.
Think of it this way, if you have free cash flow in the two-
Yeah, two and a half
two to two and a half, depending on, again, the leverage of the model. The board directive around deployment was always kind of 30%-50%. The company executed to over 100%. All of you said, "Are you going to keep your dividend, your allocation strategy the same?" Well, which one? The 100% of the 30%-50%? 100% was because we were trading at three times cash flow. Yeah, if we went back to three times cash flow with the same perspective on the future opportunity that we had then, we would do the same thing. We'd buy another 40% of our shares back. Today, the company's trading at, whatever, six and a half, seven times cash flow, which in today's capital markets is actually relatively cheap still.
Earnings has gone up to 12 or 13 times, which from a P/E ratio fundamentally maybe is okay. Relative to the market, seems odd to be at a discount to the market when you're a technology company. They're certainly a long way away from where it was obvious that this company was trading cheap, if you believed in its future. I think the point is, with $550 million growing at 10% for the next couple of years, that gets to be a big number pretty quick applied against $2 billion-$2.5 billion. You're in your 30% range to begin with. The question is, what do you deploy? That's opportunistic. If we see the stock go to values that we think are attractive, we'll increase the share repurchase.
I think we'd like to make sure that we're offsetting stock option accretion regardless, so at least we're kind of anti-dilutive in that sense. We all have the other opportunities around us in terms of are there some relatively small M&A opportunities that we want to take advantage of or debt. Again, I think the company, it's remarkable if you think about what the company has done financially in terms of its debt structure over the last two and a half years in terms of extending maturity by three years and reducing interest expense by probably something like $100 million a year. Credit markets obviously gave us that opportunity, but so did our performance. I think we're still on the path to manage things the same way, which we think is very attuned to what the shareholders want.
We have latitude around where the value opportunities is.
Okay. Fair enough. My follow-up, if I could, was just implicit within your comments, it seems like you're suggesting core grows 3%-5% for the next couple of years, 3%-5% year-over-year for the next couple of years. The last two years, in 2013 and 2014, total revenue growth was down about 4%.
Yep.
You've mentioned that your customers feel better. What is it you think that's really changing? Is it the tablet fatigue that seems to be setting in? Health of the PC market? It's a pretty meaningful change.
I think it's macro growth. I've been saying this for a year.
More macro?
I think it's macro growth. I also said two years before that that I didn't think it was the death of the PC nonsense. I think it was macro lack of growth. Look, the world was in a tough place for the last five years, has been really not fun. Not fun for you all so much. Well, maybe more fun for the stock market. For running a business, it's been really hard. In the last year, I'm not saying it's easy, especially with a couple board members here. Let's say no comp committee meeting members. No, maybe. It's definitely changed, where I think most CEOs are waking up every day with more confidence about future business opportunities, which just makes it better to think about R&D investment, capital investment, and deploying assets into areas that can produce marginal growth.
I think that's what's happening on a global basis. U.S. is strong beyond the coast. China is strong. I think China will remain strong. Clearly, North Europe is strong. South Europe, still spotty. Japan is strong. You're definitely getting pickup around the world, and I think that's what's driving it. I think the lack of growth in those areas or the spottiness of growth, for example, in the U.S., yeah, there was growth before that, but it was all on the coast. I think that's why things were pretty flat. Are there things going on underneath that like I'm buying tablets this quarter versus PCs? Yes, there are. Are there things going on beneath that like I've increased my utilization or now I can't increase my utilization? Yes. I think the general economic environment is better. That creates a better business opportunity.
On the technology side, though, again, are the trends of population growth is increasing, per capita income is increasing globally. Number of connected people is increasing and richness of data is increasing. That all translates into the need to store more data. Again, we're in the business of providing that technology. I'm going to shift over here. Rich?
Hey, thanks.
Oh.
Right here.
Is your name Rich, too? Okay. Just call yourself Rich. I don't look bad.
This is Rich. Otherwise known as Ananda Baruah from Brean Capital. This is for both Steve and Pat. M&A was described as being opportunistic as well. Would just love to get some context around where you would envision acquiring over the next few years and to what extent, if any, might the cultivation of the software business something more along the traditional lines be included in part of that strategy?
In the core business, there's always the opportunity, especially if it's against a rising environment, whether or not that's average capacity for drive and/or TAM, that we look at. I don't think that's huge. I think there are particular areas that if we saw an opportunity that we thought help our cost or our ability to scale or our vertical integration leverage, we would do that. In the software area, that's the one area where obviously, we need to build the capabilities beyond our VLSI and firmware capabilities, and especially as Jamie enters some of these areas where a higher level of software is needed to complement what our go-to-market partners are doing, that there are acquisitions that we think can help fill that gap. They're not large, i.e., they're not billion-dollar, multi-billion dollar things.
I think, if we can find the right property at the right price that we think accelerates Jamie's business so we get to the higher end of that model quicker, we're going to do it because the payback for us from an earnings and an operating margin perspective are pretty significant.
Relatively short.
Rich.
Good morning. In terms of technology, Dave, you had your chart up there about where you were going to, from a next-generation technology perspective, get to the next areal density points. Obviously, helium was not on there. I know you've talked about that in the past. Can you just update us on your view on that, especially given some of the recent product announcements, and whether or not customers could potentially pull you into that space if they do want second sources over time?
Sure. I think it's markedly the same as what we said last year. For everyone's perspective, we have a platform that got us 4 terabytes per drive. It's the highest volume shipper that we have in the cloud right now, and we really like the platform. We've extended it to 6 and 8. Our competition, for a variety of technical reasons that we understand, felt it necessary to bring on this helium enclosure. We didn't feel it was very necessary to do right now. We have other ways to solve the problem that actually get us lower cost and better manufacturing efficiency inside of our own manufacturing operations. The products are all within a few months of each other. Everybody gets wrapped around the axle about whether it has helium in it or not. It's really not a fundamental enabler for us at this point.
We like the technology. We said last year, we've had drives running for over a decade in the labs. Individual drives that are still sealed and still running. We've been playing around with it for a long time. We use it in our manufacturing operations much the same as we understand that they would have, too. At some point, we'll deploy it. We don't think we need it anytime in the near future.
Hi.
Nagel Chawla, Technology Insights Research over here. I got a question for the follow-on from Steve to Steve. You're at a return on invested capital of 26%. If you get that operating margin expansion that you're talking about, would you expect invested capital to proportionally increase, or are you also expecting ROIC to expand as well? If it does expand, are you concerned about customers or existing vendors saying, "This is a really attractive profile. I'd like to get back into this business"?
There is a lot of leverage to the model.
Yeah.
If the growth is coming from the non-core businesses, they're obviously not capital-intensive business. If anything, they're OpEx expensive businesses in terms of the go-to-market assets or the technology. To the extent the growth is coming from the core as well as that business, then clearly that's an ROIC leverage point. Where that gets trickier, I think, is as we get closer to this imbalance between supply and demand, where do we start deploying more capital? Again, I think the industry, and frankly, I think all technology has been pretty careful about that type of deployment too far in advance of actually being able to fill up the freeways, if you will. The question is, do I think our customers or suppliers or someone says, "Oh, boy, that's a great business to be in"?
I think actually there's people that are already saying that, but it's how do you be in it? The capabilities of flying a head three atoms off the surface of something is not a lot of people know how to do that anymore. Dave didn't talk about our wafer technology today. Our most complex head are probably running close to 1,100 process steps. That's probably 400 process steps, at least, higher than what the most sophisticated silicon processes are. That's going to step up to 1,500 process steps and then probably 2,000 process steps by the end of the decade. This is tough stuff. I think there's probably people that are already understanding this value split between value attributing, again, down to the component technology owners or up to the service provider software level.
If I'm in the middle, where do I go? I think a lot of the old technology companies or the major technology companies that started as component owners, they get it, like going down would be great. See, Bill Bradley left and Byron Davis came in. We just had to Bradley left, BD, thanks for answering the phone call because with that whole Tim Cook final thing. I can't be left without a star of some sort.
Yeah.
Love the new hairdo. I think what's happening is that as that value shift changes, the companies that are massive with capabilities and go to market and integration technology, they understand the advantage of the component level. From a business model perspective, they can't get there anymore because their margins have already moved into the, not necessarily a higher operating margin level, but the higher gross margin level. I think they'll have a tough time explaining to the street that the right strategic move is let me go back and own component technology.
As much as they may understand that's the right thing to do long term, there's plenty of big technology companies that we read about in the paper every day today that's announcing, "I'm out of this business that used to be core." It's because they have to move to this higher margin service-oriented model. It doesn't really necessarily, in my mind, mean it's the right long-term thing to do, but it's the thing they're doing. I think there's companies that have already thought about that. The question is, how do they engage in a way that allows them access to this technology without, quote, becoming a disk drive company? Are there other industries or companies that say, "I want to make the bet, and I want to go into that." Especially companies outside the U.S. Potentially.
Potentially, there's companies within countries that say, "This is not a technology that I want at risk," and, "We'll make the 20-year investment to get there." Obviously, starting with a core asset like Seagate might make a lot of sense. I don't know. It's going to be an interesting world. I don't leave it just upon Seagate. I think any core technology provider, whether or not it's in the silicon side or the disk drive side or even on the compute side, has these opportunities and also has the challenges. I guess I'm not calling on people, somehow mics get into people's hands and then
This is Monica from Pacific Securities.
Can you just stand up? Where are you? Oh, great.
The question is, you talked about providing the system solutions, maybe oil and gas industry and HPC industries. Do you think at that time you could run into issues that you are competing with other storage OEM customers?
I'll let Jamie answer the question since we have real-life examples already.
Yeah. All the work we're doing with oil and gas is with a co-traveler. We're not going directly to those companies. We're working with HP, we're working with Dell, we work with Cray, we're just providing a more evolved OEM solution to them than a generic box. It's a box that's deeply integrated with the media and deeply integrated with the software ecosystem around it. We're going to market with co-travelers. We're not out there selling directly.
Are they winning business from other storage companies? Yes, but that's always been the case.
To date, we haven't seen that as a conflict. It's really about our engagement with the OEM in terms of are we providing a compelling offering within their portfolio where they see value? That does vary by OEM. When you look at the deployment of cloud-based storage systems, the end deliverable channel between OEM, VAR, systems integrator, CSP, and new company is really dynamic. There's going to be winners and losers in all those categories. I'm not a subscriber to the theory, all the OEMs are going to lose, all the VARs are going to lose, and it's all about the CSPs and the new companies. I think that's ridiculous.
First of all, the new companies don't have the technology, they don't have the go-to-market capabilities, and there's no remote way for them to grow into those valuations without being acquired, because the scale to compete against someone that has 20,000 salespeople on the ground is just multi-billion dollars in a decade or more of work. On the other hand, there's OEMs that are pursuing different strategies. Some of them are aggressively challenging their legacy businesses, and others are trying to protect their legacy businesses.
The ones that try and protect their legacy businesses are probably going to be less successful than the ones that are being quicker to say, "I'm not going to hang up the deployment of my new architectures because I'm trying to protect some old server margin." The VAR-SI channels, I think the one where I'm probably most at odds with the intelligence people, is a lot of people think, oh, that's all going to AWS, or whatever version of deployment is targeted towards small, medium business. I just think that's nonsense. I think regional differences, application business differences, capabilities of a particular companies. You need people that have specific knowledge that can integrate these technologies. This is still hard stuff to do.
Even if we get to Jamie Lerner's disaggregated world of software and hardware, it's not like you can call up the post office or whoever we're supposed to call today and have drones drop us our equipment. It just plugs in at work. It takes a lot to deploy this stuff. I think this VAR-SI channel has actually got a better opportunity today than they did historically when they really were just moving product and trying to post up a point and a half of margin. Again, there'll be winners and losers in that mix.
From a Seagate perspective, it's about engaging with all of them because we get huge market intelligence by seeing the different models that people are promoting and then leveraging as much of our core technology, whether or not it's the device level, the silicon level, the firmware level, or even at the rack level, to basically have that be an offering that any one of our customers can use. Yes, we're going to customize it as they require at the margin. We're not going to pick winners. We're going to pick everyone, and whoever wins, and then Seagate benefits from that. I think it's, again, I don't think it's as cut and dry as people love to create the net zero-sum game thing, and that's just not what we're experiencing right now.
Since you guys are deciding, it doesn't make any difference if I say give it to this person because someone else gets the mic. Just hand the mic to someone with an arm up and try not to give it to the same person four times in a row.
Sorry, Steve.
Unless they're really easy questions, let's keep giving that person the mic.
Sorry about that, Steve. I'm Mark Miller again. You made mention of this at the beginning, and I'm just wondering if you give us more color. With flash going to three-dimensional structures, with the problems with FinFETs, the problems with EUV, it's certainly possible we could see a significant delay on that transition. How does the storage world do in that situation?
It's a great question. It would be disingenuous of me not to be consistent, which is, if the flash, if they can't get to 3D, we have a problem. The world has a problem if they don't get to 3D, because you can't get to the capacity and cost numbers they're hoping for without that technology. You're right, the performance out of the gate has been not fabulous. In fact, it doesn't really exist. They're actually really just doing a regular technology and calling it 3D in the meantime. I think that's getting peeled off now. Today though, say you're delivering flash at something like $0.50 a gigabyte. We can argue if that's the right number. If it's too high or too low, it's directionally correct. To get to $0.15 a gigabyte by the end of the decade, it takes 3D.
If you're not at $0.15, then the amount of flash that's available just for mobile, if you will, whatever that means, right? Phones, tablets, notebooks, and other devices, is going to chew up everything. That's probably a problem in terms of the growth models that we're talking about. We see disk drive technology going to a $0.005 a gigabyte by 2020. We feel great about that margin if they can deliver it at $0.15 a gigabyte. I think there's a lot of ifs. If you think about traditional geometries, 16 nanometers going to 13, yeah, looks like it's going to work maybe for a couple, but going below 13 seems tough. You're on the right point, and it gets back to this question about who's your flash partner.
This is why we think having active deep dialogues with all the flash partners is really critical right now. These are going to be massive investments they have to make in order to achieve these cost curves, we want to make sure that we're in dialogue with all of them, we can leverage off that technology with our controller technology and with our systems technology. It's something to watch closely because the next couple of years, either they make breakthroughs on 3D or they don't. If they don't, I think we have to rethink what does that mean about our hybrid business, because maybe it means actually that a lot of it has to be solved more with hybrid than less. Again, you get a 10x multiplier on the capacity.
If you throw a 64-gig flash chip on top of a 10-terabyte drive, you basically have 10 terabytes of flash. We're watching it closely, and it's something I think that's important for all the analysts to watch as well, and I know obviously you're deep into it. Do you want to add anything?
No, that's good.
Hi, Amit Daryanani, RBC. Steve, I just had a question on the CapEx side. You guys are talking about explosive data growth. Your areal density doesn't seem like it's going to keep pace with that growth by any means. You guys aren't really changing your CapEx numbers. I'm curious, what prevents you from increasing your CapEx given the fact you see this demand growth? Have you had any discussions with your hyperscale customers of potentially co-investing in CapEx?
Well, we don't need co-investors because we generate a lot of cash. It's the question of when do we want to increase capital expenditure? The answer is when we see gross margins moving up, then we'll increase it. We're not going to do it ahead of it. We manage our capital closely at 27-32 points. We think six to eight is right. We've been running well below that. If we went to the high end of that would actually be a fairly significant amount of capital deployment relative to what we've done the last few years. To answer your question specifically, if we started seeing gross margins pushing consistently in the 30-32 range or above, then we would start saying it probably makes sense to let more capital out.
Everybody wants to go to the default position of, yeah, don't invest any more capital because prices will go up. You leave massive amount of revenues on the table and cash flow. Look, it's one of the things that we're paid to manage, and I think we do a decent job at it. When it goes up is when you see gross margins go up. Question is, does it go up slowly over time, or is there a disruption, step function change in pricing, and then we deploy a bunch of capital. You have a long lead time on some of this stuff.
I guess, Jamie, as a follow-up, the $2 billion revenue run rate that you guys aspire for on the Cloud Systems business, $750 million today, do you think that's going to be an organic ramp, or is it inorganic initiatives that should help you get to that number?
Go ahead.
It's a combination of both. Right now, we have the EVault business, the Xyratex business, and the LSI/Avago business that are there. Right now, the business is growing fast. I do think there'll be additional inorganic work that we can do. What we're doing is we're blending the margins of a series of different plays, right? When you build a system, there's hardware, there's software, there's services, and sometimes there's hosting, and they all have very different margin profiles. It's a mix of those things. The hardware numbers are big, but if you look at the all-flash world, the all-flash world in the enterprise is totally a software battle, right? You see four different companies with the exact same flash inside their box having huge disparity of performance, capability. It is all software. The entire software storage stack has to be rewritten for flash, right?
Because it wears out. You have to build very different software that's essentially like a sprinkler, right? Wearing out the flash evenly. That's a totally different set of software. As we look at the architectures that we're building is we have a lot of the hardware components today, and that's growing quickly, but we're beginning to think about how do we round out the services capabilities, hosting capabilities, and software capabilities around that. I think you'll see both organic and inorganic development in those areas. I wouldn't say that we're done.
On that $2 billion, be clear, that's what we already have in our portfolio. That's our flash platform and our cloud. We already have that line of sight, what's in our portfolio. To Jamie's point, it could be more with other activity.
Great. Aaron Rakers at Stifel. Just two questions. Building on that, Jamie, or even Pat, can you talk a little bit about the gross margin assumption you're making in that business? Obviously, moving up the value stack, stripping out the pass-through effect of the hard disk drive business that Xyratex would've recognized.
Right.
I'm just curious of why it looks like the gross margin wouldn't actually float higher. As a secondary question, just real quickly, I'd be curious, any update on MOFCOM? I know it's immaterial, relatively smaller for you guys, but just curious if there's been continued active dialogue there. Thank you.
Let me start with the margin, I'll let you do Movcom. It's a shaping, right? If you look at the majority of the $750 million that we'll do this year, that is OEM development for the world's biggest OEMs, ODMs, right? That has a lower margin profile that when we sell a full system like ClusterStor to an oil and gas company. It's about shape. Right now, the majority of the sales are OEMs. The fastest-growing part of the business are the solutions that, these solution sets. We still go to the market with an OEM, but the OEM derives most of their margin rather than stacking it on top of our margin. They deliver their margins by selling compute, selling services, selling support contracts around our technology. As that grows, the margins will shape.
We have two quarters of data, I think we're taking a conservative approach on it, and each quarter, we're going to learn how that mix changes. The fastest-growing part of the business are the higher-margin pieces. They're just coming off a smaller base. Just to give you an idea, EVault, that's a 67% margin business. $50 million of revenue there has a very different impact than $50 million of hardware OEM manufacturing. I'm sitting with my sales teams, and we're putting a lot of effort into what should the incentive structure look like, right? I mean, the comp plan drives behavior. What are the behaviors that we want our sellers to have? What are the behaviors we want to put in front of our OEMs? We're starting to OEM cloud services, like EVault, to the large storage oligarchs today.
We're coming to them saying, "Well, hey, don't just OEM hardware from us. Would you like to OEM and white label some of our cloud services? Would you like to white label some of our software so we can begin to play with that mix?" Again, we're early days. We've got two quarters under our belt.
Movcom, no material changes. We're actively engaged with them. We have a trustee that we feel we're in good stead. We feel we've followed everything, we'll just keep on working with Movcom. When they see the right path to exit, that'll happen. Right now, we're actively engaged and no material update.
Why don't we try and answer some questions over here? This gentleman here has had his hand up for a while. Sorry, I inserted myself.
Steven Fox with Cross Research. First, just a clarification on all the adjacencies. Do you see an EPS crossover point that we should be looking at? Like how dilutive is this currently in terms of ramping these businesses? Secondly, we haven't talked too much about the retail side, the client side of the business. It's pretty obvious from the presentations that average capacities on the enterprise side are going up. It's not so obvious on the client side that they're going to continue to go up. Maybe if you could talk a little bit about those trends, that'd be helpful.
Do you want to talk about that, Mark?
Sure.
You guys want to-
First on the EPS, if I think of the model we gave on 2015, you're going to find yourself relatively in the ZIP code of 14, which would say that we're able to continue to deliver the earnings power that we did last year, even taking on the additional OpEx without getting the scale of revenue. If you want to put that in the dilution, that we could have done more if we didn't have it. I think if you extend it to two years, you see the earnings power increase. We think it's a good bet. I think that model we gave for 2015, you're going to sort of line your way in that sort of ZIP code. As Steve said, if things continue, and the macro stays strong through the year, we could have some lift on that.
That's where we're hopeful, but we're just going to manage whatever environment we're in.
2016 looks accretive right now.
Yeah.
I mean, based on our assumptions. 2015 looks flat. That's why, again, what we're excited about is to have that type of opportunity but really not have an EPS penalty and actually a fairly quick return on it is exciting, and as long as we can take kind of relatively low-risk investments to capture that opportunity, we'll continue to do so.
Client side, the exabytes per drive still growing quite healthy actually right now, this time of the year. We talked about PC a lot, but if you break it down to more, I'll call it traditional desktop, understanding that some of those distribution channel drives go into NAS applications or surveillance applications, it's definitely moving up from one terabyte per platter, products that we're getting out a little bit higher areal density than that, you're seeing the 2s and the 3s and the 4s, 4 terabytes go. Even on the notebook side, what's traditionally been just 500 gigabytes single-platter product, you're seeing 750s and one terabytes and the market's definitely moving north. I think a lot of that's around video and what's going on at the client relative to video.
You know the cameras exist, but how you use the video on the client devices is still a little bit up in the air, there's going to be a lot of development there. We're pretty excited about it and looking for a resurgence, not only in total box count but also in component count inside the boxes.
Yeah, I think in two years at this meeting, we could be talking about as much opportunity on new client platforms as we are today about what's going on in the back-end architectures. That was my opening comments. I think the potential of what can happen in the client as a function of what's happening with technology and deployment in the back end is really pretty exciting. It's just a question of which technology provider locks onto it and charges ahead. There's a couple of them right now that are starting to do some pretty creative things on the client side.