Thank you. Good morning, everyone. Thank you for being here today for Seagate Technologies 2015 Strategic Update. We appreciate you coming out this morning for our product showcase and talking with Dave Mosley and Phil Brace. Just for everyone's benefit, folks will be back there during our break and during our reception to talk more about our products.
So please make sure you go back, introduce yourselves and learn more. With that, I'm just going to get started with our forward looking statements. I just want to remind everybody that in our presentation, we will be making forward looking statements and we refer you to our risks and uncertainties that are outlined in our SEC filings and on our website. We will also be using non GAAP information and the reconciliations are in the presentation that will be posted on the website today. So our agenda today before we get started, I just wanted to let you know who from the Seagate management team is here.
In the front row, we have Dave Mosley, Phil Brace, Pat O'Malley, Dave Morton.
All of you
know Jingjing Chen as well from the Investor Relations Group. And from our Board of Directors, we have Chris Hongkin. So thank you for being here. Our agenda today, we're following what we have done in the last couple of years where we have some interaction with the crowd during our presentation. So our first presentation will be from Steve Lusso, our Chairman and CEO.
He'll be talking about some global trends and technology trends. He will then be hosting a Q and A as we have done in the past. We're mixing things up a little bit. We have a special guest moderator today. So we're looking forward to having the questions that he's prepared for Steve.
After that, we'll be taking a short break. We'll have some videos that are about halfway through. And then we will be turning the stage to Dave Mosley, who will be talking more about our technology portfolio. After that, we'll have Pat O'Malley talking about our financial model. And then we'll go be going back into more executive Q and A.
With that, I would like to turn the floor over to Steve Lusso.
Great. No, I'm going to get this done. Yes. Okay, great. Thanks, everyone.
Good morning. How's everybody doing? Good. Everybody excited to get off the Hamptons after this? We did discuss all wearing white.
We didn't know how that was received in New York and our PR people didn't think that that would play well on the video. So imagine me all in white getting ready to go to Hamptons with all of you this weekend. Thanks for being here. I want to cover a few things and then I'm going to ask Harold Levy to come up. He's going to do an interview for us.
But just to frame a few things, I think as you know this is our 5th year doing this. And how we tend to think about the world is technology trends, how does our portfolio stack up against that? What does that mean in terms of our financial model and ability to return value to our shareholders relative to where the stock is trading. And then that's all usually encompassed in some sort of conditioning around macro environments and the various options that we have against macro trends. So we'll continue to kind of talk about that throughout the day, but that's how we always think of the world.
And in that context, this 5 year view is really kind of interesting. 10 or fiscal year 20 11 when we were here, a lot of the same, yes, buts were in play. Yes, but the client is dead. Back then, I think it was dead maybe because of tablets. I can't remember why it was dead then.
Yes, but SSDs are going to take over the world. Yes, but the world is in terrible shape. And it was all against our thesis of that we thought we could grow in the 3% to 5% range, maintain margins and return value to shareholders, especially relative to the stock valuation at the time. And we kind of find ourselves looking at this 5 year view and what you see is that, as we expected, the nearline growth has really accelerated from about 6 exabytes to 57, so big growth in terms of exabyte shipments. And again, that's how we think of the world as exabytes.
How we package the exabytes is probably the among the more easier things that we decide, it's actually delivering exabytes that's harder developing the technology to deliver the exabytes. And in fact, the transition from client to higher densities and higher capacity drives is actually quite favorable for the drive industry for a variety of reasons. 1, more heads and disc per platform. 2, the content in those heads and disc and channel is much higher, much more difficult to do, more value. 3, the test time tied up with those is much higher and of course a lot of our capital goes into test.
So you're absorbing more head capacity, more disk capacity, more channel capacity as well as more test capacity. And then the usage of those drives is quite different as well. Dave is going to go into this in his presentation on technology. But I think one of the things that the investment community and the general media is not kind of really up to speed on is how these systems are being deployed inside real environments. So what's really happening with the workloads?
How are the drives really being used? What does it mean not just about number of units or capacity, but what's happening to each unit at each capacity. And the workloads are very different than on the client and they're different in a way that again favors the drive industry because they're more intense usages. And against that 5 year review, we also grew revenue by 4%, which is kind of our forecast going forward. As you can see, earnings per share at 6%, so in excess of the growth rate of revenues.
And again, we did that mostly through buyback and some OpEx control. But of course, going forward, as you know, we're making big adjustments to our operating expense as well. So as we look at the next 5 years, this is why we have confidence that we still believe that the mid single digit growth rates are achievable. Margins should remain stable given the stability of the industry where supply and demand is being matched fairly well. And in fact, with Toshiba kind of adjusting to their accounting situation, there's always the question of what happens to the Toshiba Drive business.
And that's some market share that's potentially at play if they decide to exit that business, which is not improbable. Against this, of course, we've generated substantial cash flow and against capital expenditure budgets that are certainly adequate for us to advance our technology and our asset base in a way to meet demand. So again, 5 year snapshot that we're pretty proud of. And in the course of all this, I think we returned something like $10,000,000,000 to investors in dividends and stock buybacks. So just briefly, and then I want to have Harold come up and touch on these topics and other.
Again, our framework is what are the technology trends from an architectural perspective? I don't know if there's a whole lot different today versus what we've been saying for the last couple of years, which is the shift to cloud based computing and mobile devices accessing those data sets is continuing to accelerate. It's being driven probably more so by consumer applications today than commercial. But clearly, commercial applications are coming into play as the costs decrease and then people start to disaggregate or disrupt commercial applications. As long as the flash companies can stay on the linear density curves, we're confident that the mobile devices will continue to have the performance characteristics that will result in an experience that allows those devices to continue to deploy globally, which then, of course, drives the need for compute, networking and storage fairly substantially, particularly as those trends shift towards large amounts of unstructured data, which is in our opinion high definition or 4 ks video.
Those trends seem to be in play. The cloud service providers are still the significant purchaser of cloud equipment. But as you get into hybrid models where people are going to basically deploy on-site as well as in some of the bigger clouds, we're starting to see that customer base broaden. That customer base is in transition, whether or not it's the CSPs or the new companies in terms of how they purchase their technology. Depending on where they are in the scale of things, they either almost buy all of their technology from individual technology we believe that will be the go to market channel for probably the next 5 to 10 years where there's a mix of people either buying through OEMs or buying technology directly depending on how much scale they have and what their internal competency is.
We are not believers that all the OEMs go away. We are probably believers that OEMs consolidate, but that's why we still believe that an OEM channel is important as evidenced obviously by our transaction with Dot Hill. Inside of all of that is a clear shift that says, how do you provide technology at the device level, the subsystem level and the systems level that allows that customer whether or not it's a value added reseller, an OEM or an end user to have choice of buying any one of those pieces to stack in the way they want with a software overlay that they want. And so for us, as you've seen in the back, it's really about having a broad portfolio with HDDs, hybrids, SSDs, performance accelerator cards at the device level and making sure that we lead across each one of those categories as well as how do we package them either as subsystems or as systems depending what the customer wants. So we feel that our technology portfolio is evolving nicely relative to the technology trends.
The financial model I talked about, we feel that the company creates enough value that we can invest in our R and D and our capital and do M and A as needed and still have excess returns that we can then distribute back to shareholders in the form of dividends and or stock buyback depending on where the stock is trading relative to our expectations or relative to cash flow multiples and interest rates. And then the global economy, I think is in many ways the most interesting thing right now. Starting 9 months ago, I think Seagate was one of the first companies to start to challenge the thesis that China was growing at 7%. I think I know I was on record several times saying from my perspective the growth rates are closer to 3% to 4%. And now of course there's a lot of smart people in the world who are even challenging that in terms of what's real growth in China.
It's a concern. It's not a concern because some minor percentage of people that own equities in China have lost a lot of value. It's a concern because if that reflects a real shift in economic growth, what's the impact on the region. And of course, the region is important to every multinational, but obviously particularly to technology companies where anywhere to 50% to 60% of the revenue is generated in region. Strong dollar we were talking about 3 quarters ago that given the rise in the dollar over the short period of time that it has strengthened was unprecedented and the ability to really understand the implications of that from an overall demand perspective as well as a currency risk perspective and cost perspective was hard to calculate.
We did our best in terms of what we thought the impacts would be. But I think, again, as we've kind of lived through this, people are getting better now after the fact understanding what the implications are of the strengthening of the dollar, which to me is the biggest risk on an interest rate rise in the U. S. It's not about 25 basis points on the debt outstanding so much as is that priced into the dollar or not. A lot of smart people think it is priced into the dollar, if that's the case, great.
And if it's not and there's another rise in the dollar, I think that's problematic for all multinationals, but particularly for technology companies because we're dollar cost based. And it's really hard for us to do anything obviously to hedge against the strengthening dollar when you're dollar cost based. So I think with that, I'd like to introduce Harold Levy. He's Founder and Managing Director of Iridium Asset Management. I am.
And I asked Harold to do this because for those of you who know Harold, he's a tough guy. He's been an investor in Seagate for I think going on 7 years I'm thinking. And it's been a really I mean this in the most positive way challenging relationship. And I mean that in that like some of our other investors that have been in our portfolio, they challenge us to be better. And Harold and his team are in active dialogue with us in terms of really pushing our thinking about trends and contingencies and how we manage the business.
And sometimes it's very frustrating. I think once I completely lost it with Harold and then after I hung up, I was like, but he's right. And it reminded me of when Dave Bonnerman used to say, we spent too much money on capital. And I said, what does that airline guy know? And then afterwards, I was like, but he's right.
So this is unscripted. I told Harold I didn't want to know the questions ahead of time. He could ask anything. And what we do is ask Harold to come up and we'll start with some of the questions he had. And if they're too hard, then I'll go to the crowd.
And if he gets forward with my answers, then he'll go to the crowd. But so please welcome Harold Levy. Thank you. Which side do you prefer?
Well, I have my That's
a good side. I have
my list of questions. I've got a nice introduction and my list of questions for not so nice introduction. So I'm going to go to the nice introduction.
I think I only yelled at you once, right?
You lost it once. You lost it once. I'm not sure whether you hung up, but you No,
I didn't hang up. You came close. She came close. Well, the point of contention was whether or not we were returning money to shareholders quick enough. And I was telling Harold that look, we're doing at this pace, but if the opportunity presents itself along your thesis, then we would do it faster.
And but Harold's viewpoint was, well, but my thesis is right, so do it faster. And it turns out he was right and then we actually did it even faster than faster. So then we ended up to be okay for that little time period.
I think we also had an issue about dividends. And I kept telling you that when you buy back stock, I don't recall how much money that actually goes in my pocket that I can spend.
That's right.
It's kind of a theoretical good thing.
Yes. I mean that's a good point. Anyways So we better keep raising the dividend.
The nice thing about being an investor in Seagate is it's a good lazy man stock because we can ask the same questions today that we asked 5 years ago. So I will ask the question I asked 5 years ago. Are we making is the market telling us that we're making buggy whips, sophisticated, highly engineered, complex, essential today, but long term have to question the future. That's essentially what the market is saying because you have a market value of we all know the history. We have a market value today of $16,000,000,000 plus or minus we have very modest debt.
Management is well regarded. And we're selling at 10 times free cash, right? So the market's telling us implicitly that this is not going to continue.
Yes, Or 6 times free cash flow. But yes, I think it is what the market is saying in some form or fashion. And therefore, there's the value play. There's your chance for an arbitrage. I think the world has trended towards misinformation statements that aren't based on fact and kind of quick sound bites that then events happen as a result of that.
And then shortly thereafter someone can analyze that and say, but that's not true, that's not true, that's not true. And the answer then is, okay, but who cares? And you just move on. And it's a dangerous thing. I've done lectures on this that the lack of editorial control and the lack of cause and effect that the fast social media world has started is not a good thing.
And it's not a good thing in a lot of ways, including mispricing of assets, which means then capital flows aren't probably occurring in the way that they should. And how does it self correct? Well, it self corrects over time. We know that, but the question is what's over time. And like I said, if you think back to fiscal year 2011, the stock was probably at, I don't know, 21 or something then and it self corrected to 70 and a lot of people made a lot of money as a result of that.
And I think the same type of opportunity exists. The reality is the shift in client is something that we recognize and have recognized for a long time. And that in many applications, if you're just consuming data, you don't need a lot of storage unless your consumption is video, in which case then you have a latency issue that you decide you want to manage or not. But all that compute still requires bits to be stored. And people don't store on flash solely.
I mean they can store some, but at the end of the day all those bits go on a disk drive too somewhere either in the form of backup or it's in the cloud or whatever. And if you're consuming only, that's an interesting world. In that it's, I think, kind of frighteningly boring and dangerous, because someone has to be creating value. And therefore the question is where's the knowledge worker and where's the platforms that are designed for knowledge workers. And those need more CPU, more bandwidth and more storage.
Why? Because the problems that we're trying to solve now aren't about analyzing structured data. Those problems, the compute that's being applied to those, I think, is fairly linear. In that, if the data sets are structured, it means that the data sets have been defined according to an algorithm that people understand and what they're trying to do is basically do it faster. And so fiber optics or more CPU or more storage with more SSDs to munch data quicker.
And that has a linear aspect to it that I think will continue to grow. But it's the unstructured data where all the value is. And unstructured data by definition means that you haven't actually figured out the algorithm yet that says, if I have an event that I know has caused an effect, How quickly can I analyze what caused that? So it's all about latency. But really what it's about is prediction, right?
Where you really want to get to is prediction. And to get to prediction on unstructured data, which is basically video, it's analog, it's not digital, you're nowhere close to what we need in terms of compute bandwidth or storage. And I think this is what's exciting to us about data science and Dave is going to talk about this. When you see how our drives are used in data science applications and for all you in the financial services market, probably the biggest case study is Renaissance. No one really knows what they do other than they make gobs of money.
And they make it because they're analyzing unstructured data. They're looking at variances. They find them before anyone else. Then they look at the pricing around those data and they are the pricing. And they're basically creating more efficient markets because of that.
So I think there's huge drivers to it. So I do believe that if you're reading the newspaper and you say, oh, SSDs are taking over the world. If you want to believe that, but that isn't how the technology is being used. Client, we believe, will continue to probably decrease to some point, we believe then increase. That may be different than what our competitors think, including our disk drive competitors.
Why do we think that? Because we think there'll be a new class of clients. If you say what's a client today, it's a desktop computer, a notebook, a tablet, a small phone and a big phone. And if you say that's the world, then yes, it's probably a net zero sum game that there's a trade off between what's HDD and what's flash depending on are people buying big smartphones today or are they buying notebooks or are they buying desktops. But if you peel back life every kind of 5 years, one of those devices didn't exist.
So are you really saying we're going stop inventing now on those 5 clients? I don't think so. I mean, I think there's going to be new clients. I think the 3 d printing is going to be bigger than PCs. And I think that because a 3 d printer needs 2, 4, 10 terabytes of storage to actually render specifically what it's trying to print.
But what is a 3 d printer doing? It's disaggregating the supply chain where there's 1,000,000,000,000 of dollars of real dollars being sent. What did the PC do? The PC disaggregated knowledge workers. I mean, that was a pretty soft measure.
We're going to make knowledge workers more efficient. Printing is going to basically take dollars out of the system. So I actually think PC printing 3 d printing itself will be bigger than PCs. What about surveillance? What about robotics?
What about drones? What about cars? I mean, if you don't call that a robot, which is self driving car to me is a robot. So I think there's a whole new class of clients that start to evolve in the next couple of 3 years that actually even grows that aspect. So I think, yes, I do believe people think it's the buggy whip, but I think they just don't really take time to understand how technology is being deployed and used.
So in 5 years, will I not ask you this question? Or will anyone else not ask you this question?
I think in 5 years you won't ask this question. Because I think in 5 years, this is what I've said, the inconsistency of how the market is pricing assets around the cloud is really interesting. Massive valuations, right, if you're a cloud company. I mean, so massive in fact that actually the Holy Grail is not to make money. And trust me, I want to get to that land.
I want to get to the land when the day I tell you guys we're not making money is when we're worth $100,000,000,000 That's a cool place to be. But it's kind of crazy, right, because the promise of all that and look there's lots of promise to those companies that are being valued that way. But the promise of it says that you're collecting all this data and information. And but at the end of the day, it has to be used somehow creatively and that has to be used at the endpoints, not at the data side. It's going to be used at the endpoints, which I think obviously in terms of some of the security issues and bandwidth issues and cost of bandwidth is actually causing people to spread that data out more and more.
But it's really how do you have platforms to access that data to do creative things That's going to be interesting on our thesis. And I think over the next 5 years, if the promise of the cloud delivers, which I think it will, then implicitly the promise of a new client is going to deliver. And the question is who does that? Now thank goodness in the last 2 days, Intel raised up their hand and said maybe we actually should be back in this business. Now we just got to wait for Microsoft to get there, because to me the big failing has been Microsoft and Intel have not actually thought about what's the next great client.
And if it's not led by them, who's it going to be led by? You need someone to do the software and to do the silicon, at least Intel stepped up yesterday, which is great. But I do think and then I think we're going to continue to have 18 months to 24 months of this shift from client to nearline. Again, it's all good for kind of Seagate. But I do believe there's going to be a new client.
And if it happens, it's great for us. And if it doesn't, we're kind of skinny down to the fact that if it doesn't, it doesn't. I mean, we're kind of managing our business that way that we'll scale up with it. So translation, we have higher earnings and a higher multiple because we won't have
a going out of business multiple anymore?
That's right.
All right. We can end it. I think that's all I got. Yes.
Well, and in the meantime, since we believe that, right, what do we do? We stick to our thesis, which is invest in the core technology, which today, again, a lot of the investing public doesn't want to kind of go for 3 or 7 layers down. But the reality is, as our technical lead on the competition right now, there's never been a drive company in the 30 years I've been doing this that's had the lead that we have in media, heads, channel, firmware and manufacturing. I mean, it's really pretty staggering. And so we're going to keep the pressure on it.
And by the way, part of that in fairness to the competition is because they've had the whole separate agreement, because the MOBCOM regulators not allowing them to merge. And because they can't merge, they can't define a product set. And because they can't define a product set, they can't pick a technology path on some of those underlying components, which allows them basically to make the capital investments that they need to that we've been able to part of it. And part of it is, they've always kind of been a fast follower and we've always kind of been the leader. So we have to keep investing.
But the beauty of the drive industry is between R and D and capital as a percentage of revenue, it's a really efficient device. So we can do that and play the optionality of if the client takes off again, we'll be able to add that capacity pretty quickly. On the other side, if the client doesn't take off, then we still have optionality, which is we can reduce our footprint. And again, it's something we can do pretty quickly and create a lot of operating leverage.
Is 3 the right number of competitors?
Is 3 the right number of competitors? Probably, but we don't have 3. I think it's a weird thing. First of all, we have 5 competitors in notebook. So again, a lot of people want to point to this world and say, well, it's you're already down to 2, how can margins get better?
Which again, I take back to supply and demand management. It's not about pricing control. It's about supply and demand management. And if you have too many providers of anything, the issue sometimes can be around people's ability to leverage price. But when it's a technology based product that's as difficult as this one is, it's more about did you mess up in terms of how much stuff you make.
And the history of the drive industry, which is why it's different than the flash industry entirely, is it was the consolidation of the components that happened first. And how that played into Seagate's hand is that we were always vertically integrated and the rest of the industry wasn't. And for a long time, we paid a price for that, because the rest of the industry would pick off the cheapest component kind of that cycle. But where we won over time is that we knew that the technology was going to require a tight integration of the science behind each one of those as well as the ability to ramp and have lower cost. And so what happened is that over time the component industry consolidated.
And once that happened, WD realized where it was headed and then they went vertical. And that happened in kind of 2,002 or whatever it was. As soon as that happened, we knew it was going to 2 or 3 players. We didn't know if it was going to take 5 years or 7 or 10, but the writing was on the wall. And then of course by 2,008 or 2,009 it started to whittle down.
But because of the hold separate, there's really 5 players in notebooks still because Samsung is still separate brand as well that's priced separately and marketed separately. Toshiba is obviously in Hitachi, WD and Seagate. In some markets, there's 3. In some markets, there's 4. In some markets, there's 5.
And I don't think in any market, there's 2 anymore, maybe like 15 ks or something like that. But is it the right number? I think given the long term view we have on zettabyte growth and the industry's ability to meet that, it would be nice to have 3 2 or 3 players to service that. But 5 is too many. And I think, Toshiba has a real challenge to decide what they want to be now in light of the financial issues.
They're bundling assets and selling assets. I think it's clear the last thing they want to sell is the fabs. And they have to answer the question, why are they in the PC business? And if they're not in the PC business, why are they in the drive business? If that happens, I think, obviously, that's opportunity for us and WD.
And I think it's not non zero probability that opportunity exists. And at some point, we're not as optimistic as WD, but at some point, we expect Mavcomm will allow us to merge our companies and allow them to merge their and then you're going to get some more efficiency on the notebook side. Is 5 or 6 too many in Flash? It depends how well capitalized they are. I think there's going to be shortages across the board in memory.
I've said this forever. And of course, we always view us as storage and flash is memory. Of course, we have memory products now based on flash. The tricky thing about flash is like I said, the difference is because a lot of people say, we'll just consolidate like the drive guys did and look how great that is. But it's not about there's not a supply chain to consolidate.
I mean their equipment set supply chain is already pretty much consolidated. And the problem is, it's just really expensive. Spend $12,000,000,000 for 10 exabytes, it's insanity. So how you use that those exabytes and it's exabytes that can't be written to a whole lot even at compute grade. So I don't know.
It's such a massively capital intensive and R and D intensive business that you'd kind of say consolidation would be good because you get efficiencies. And on the other hand, maybe the consolidation comes just through someone winning market share versus M
and A.
The weird thing about the flash market is you've got Samsung, Hynix, Toshiba, Sandisk and Micron. And unlike the drive industry where there's people leaving it and you have Intel coming in, right? So if anything it's going the other direction. And the question is, if you're not a technology leader and you're not a cost leader and or a cost leader in that scenario, how do you really survive long time? When the biggest player is the technology leader and is the cost leader and has a business philosophy that extends years versus quarters.
So they'll price their things differently against that. It's I think it's tough. I think it's tough.
While we're on long term, short term demand has been probably less than anyone would have thought a year ago, right? Are you convinced it's cyclical not structural?
Depending on the category, I think some are structural and some are cyclical. And I think some of them are cyclical within structural. And so I think like I said, I think for us probably the PC market starting last June was probably not as strong perhaps for structural reasons, as people thought. And people could have been OEMs or people could have been Intel. And so I think people were building inventory expecting a back half calendar year that ultimately didn't materialize.
And then the question is when you take the inventory adjustments. And I guess for us, we always joke that December is the month of December is a really interesting and informative month for us, because sometimes December ends on the 15th December and sometimes it ends on the 15th January. What I mean by that is sometimes it just shuts down and you know like stuff slowing down and sometimes you're going flat out till the end of the month. And you could have even sold more. This year clearly it rolled over early.
And for us, I think that was the first signal that especially relative to 2014 or 2000, I guess, 2013 that it was something was kind of a miss between supply and demand. And I think the PC guys kind of took shots at it in March June. But again the PC market has lots of suppliers still too. There is an interesting question. How many do you need in that market?
And I think the answer there is probably 3. And right now, there's still kind of 7 in certain flavors or forms. And so I think it's just taken a while longer to adjust the inventory in the channel complicated by the strength of the dollar particularly in Europe. And but I feel now it feels to us like it's probably it feels to us like it's kind of getting stable again. It feels like PC client desktop is starting to show a little bit of firmness.
ODMs are starting to talk about stable builds for this quarter and maybe even rising from next quarter. So I think most of the PC companies and Intel have probably taken the hit that they maybe should have started taking earlier. We'll see. Again, structurally, I don't know that there's huge deltas to the curves that we've drawn. I think for certain classes of clients, the demand is going to be good.
It's just a question of what's the quality of the product. Notebooks have gotten a lot better. People are starting to buy notebooks again. Now people decide big phones are better than tablets. So now the tablets are dead.
And I think there's obviously desktops, especially if you're a consumer or a creator of a lot of video, there's a lot of reasons why desktop computer is a good device to have.
Are there huge efficiencies that still are going to happen on the enterprise side, so that that demand is not going to there's a cyclical issue there?
Yes. I mean, I think the answer is yes, but that's not the whole equation, right? The whole equation is, are the efficiencies going to be enough to offset the data growth? And the answer is no, absolutely not. And that can be applied to CPUs, bandwidth, storage, anything.
I mean, yes, all of those technologies are getting better. They're getting more efficient. They're getting lower cost per whatever your measure is on those technology fronts. But relative to the amount of data that's being ingested, not nearly enough. Again, you're going to have cycles.
For us, on the nearline side or cloud side, Today the world still is, if you have 10 or 15 or 20 cloud service providers, they're a big percentage of the drives that we sell into a nearline application. And it's not a big enough number of customers yet where they can smooth each other out. And in fact, a lot of them are kind of driven by the same cycles in terms of their capital expenditure budgets. So you still get a fair amount of cyclicality within that demand set. Even though the general curve is pretty solidly shifting, I think what's the shift in exabyte stored on nearline today versus 5 years ago?
57% up from was it 6% to or 6 exabytes. But what was the percentage? We have that well, you'll see it later. We have a chart. I mean, there's huge percentage growth, right?
But within that customer set, things are still really cyclical. And each one of those customers is big enough that I think that impacts the drive industry. As that customer set broadens, we get smoothing. And each one of those customers is doing a few things. They're driving efficiencies in that they want the ability to have higher utilization rates.
And they kind of get those achievements and step functions. We talked about this I think last year or the year before that kind of every 6 to 12 months whether or not it's Google or Microsoft or Amazon, there'll be some sort of advancement for their workload that allows them to push their existing architectures longer and harder. Secondly is they want to decrease the time the deployment. So the day they from the buy an asset to the point that it's actually productive, it's still a big gap. It's not as big as it was and there's still a range.
The very best maybe is doing that in 2 months and maybe the worst now is doing it in 6 months before the worst was doing it in 12. That collapsing of buying an asset to getting into production is that's inventory. And not only is an inventory, it's risk and forecasting because you're making assumptions about your data growth rates and then you're saying, and this is how long it's going to take me and my assets to get in place, so you buy against that. And if you're right or wrong on your data growth rates, you either have too much or too little inventory. And of course, you're actually taking macro risk in doing that.
So they're obviously trying to improve that and they continue to improve that, which takes some volatility up. And then the third thing is the devices are just getting a lot better. The utilization of hard drive is the reliability of the drives is getting better and better. So some of the math around when was I going to have to replace that thing is changing. So I think there's a lot of things that are still very cyclical within that nice growing demand curve.
I think the good news is as you get into more of a hybrid cloud world, which again big investment thesis, Seagate believes in a hybrid cloud world. We do not believe in a world where 20 companies control all the data. I do not believe there'll be one copy of Star Wars that we all watch. And I don't think that's happening for a long, long, long time if ever. And so as you get a broader customer base, which has its own challenges by the way technically and from a go to market perspective for us, But as you get that you do get a smoothing of I think those cycles.
So the lifespan
of a hard drive changed?
Yes. It's gotten longer. Well, in the reliability has definitely gotten better and longer. On the other hand, Dave's going to talk about this, the utilization of the devices. You take a drive in a near line in a cloud environment, those things are getting big 20 fourseven.
I mean, so it's not just about capacity. It's about the amount of data that's being taken in and out of those devices. And it's pretty staggering. So they are getting more reliable, but they're also getting used a lot more. So net net that's kind of what we're always fighting in terms of trying to help our customers.
So 5 years ago what was the lifespan? What's it today? Has that changed?
On the products themselves, product cycle? The product cycles from 5 years ago today are I don't know, probably about the same product cycle. I think it's I think on this next cycle, it's going to get pretty long, which has cost advantages. So we're going to roll into a bunch of platforms that we announced today, I guess. If we didn't, I guess, I'm announcing them now.
We have 1 terabyte per disc on a 2.5 inches disc, 2 terabytes on a 3 inches disc. That platform, you could see being around for a long time because you have 1, 2, 4, 8, you kind of have capacity points that fill a lot of gaps, especially given our capability of putting lots of disk per device. So I think that technology probably has a longer run than maybe the one before.
Actually our numbers agree. Since you've been at Seagate, we calculated approximately you've generated about $15,000,000,000 of cash. You've spent about $4,000,000 on CapEx. And then the remaining $10,000,000 $11,000,000 depending on how you count, you spent you gave us $2,000,000,000 in dividends, dollars 7,000,000 in repos and then you did about $1,000,000,000 of M and A. And that's after spending, I think, about $6,000,000,000 in R and D.
In hindsight, would you say should have spent less in
R and D, more in R
and D, more M and A, less M and A? How do you kind of look back?
There's I I mean, I think we did a decent job, notwithstanding the argument amongst the owners of the stock of how do you split between dividends and buyback, which I think is a completely fair argument to have and that we take input constantly. And to me that's always just a function of where is the stock trading as a cash flow multiple versus borrowing rates. The long term goal clearly is to get the share count down to a level where you can sustain a dividend growth rate that's extremely high or get to a dividend absolute dividend number that's extremely high. So that when you do it as a yield, again, the stock is kind of very undervalued today. But you have to be able to build a strong business to keep the cash flow sustained, which is all about R and D and capital and M and A.
But for us, do I think that there's I ask this question to Dave all the time. If you had another $500,000,000 to spend in R and D, what would you do differently? Would we be advancing technology faster or making better products or whatever? And I think we have it about right. I mean, we're spending $1,000,000,000 a year in R and D and we have a lot to do, but I don't feel like we're starving anything important.
And in fact, I think there are things that we can do with our resources even better. I think there's opportunities in front of us. That Seagate is actually phenomenally well positioned for it, maybe not classic storage, but it does fall into our expertise of high volume manufacturing or managing complex supply chains or and or technology based products that are servomechanical, I. E. Robots, which I do think is the next $1,000,000,000,000 industry.
And I think there's things that we can do with our R and D resources that could give us exposure to those markets that are not going to cost more money. And by the way, we're probably the best robot maker in the world, which no one realizes, because our robots have to do things at the micron level, not at the millimeter level with complex vision systems and control systems and everything. So I don't think there's an issue there. In M and A, I think relative to whether or not it's SanDisk or WD or whatever, I mean, I think our M and A track record is really good. I think the assets that we bought have been better assets that have the dollars or less than what the other guys have bought.
And now we have to put them together and grow them. But it's not like there's something I could say, boy, I wish I bought that 3 years ago. Maybe as an investor there is, but not as a Seagate, not as part of the corporate assets.
If the rumors from, I don't know, 3 or 4 years ago were true about private equity and you were sitting here as running this business for a private equity firm, would you be doing anything differently?
Well, we wouldn't be having this meeting probably. I yes, Having been one of the few people who have like lived in both worlds, it's easy when you're in the public world to say the private equity world is a better world to be in for a lot of reasons the myth of longer term perspective or an easier investor base to deal with. At the end of the day, I don't know that any of that's really true. What private equity did for us then was and that was just because of the incredible arb that existed with the Veritas asset is it allowed us basically to capture that arb immediately as opposed to over time. And even though we kind of laid it out for everybody again, it was the same thing.
People didn't believe it even though it was right there. The facts were there. And so they allowed that transaction to happen and there's kind of no way to do that otherwise. Today, our ability to take advantage of what we view as a mispriced asset, we can kind of do it on our own. It's not like they bring anything.
We have plenty of equity. We get cash flow and we could borrow cheaper than they'd ever lend to us. So I don't know. And from a management perspective, the funny thing about that buyout, the first buyout was we were already 2.5 years into a 5 year transition on Factory of the Future and it was another one that the Street didn't believe that we could actually get automated. And the private equity guys, that's what they benefited from.
I mean, they just went and paid attention and they sold they've already done they're already almost done. They took 50 people out of the clean room and now there's 2. And so that was a huge transition that they benefited from. And when we looked at it a couple of years, it wasn't a rumor because remember we actually did disclose that there were discussions. That was in retrospect, they all the parties that were involved didn't admit that it was the biggest thing that they missed, because that buyout would have actually returned more than the first one.
And what they missed was they missed SSDs. They were convinced that every data center was going to be filled with SSDs, just like a lot of these people in this room are. And they didn't do their homework. And I remember sitting in a meeting investment committee meeting with 1 of the firms and one of the partners and the firm had changed their structure where instead of the principals actually being the ones deciding on investment they decided to let everybody decide. And this one partner says, well, my brother's uncle's sister runs the IT department at Ford and she says, blah, blah, blah.
And I remember looking at the investor, are you kidding me? Like that's the question I have to answer. That's not how these technologies are being used. And right then I knew that deal wasn't going to happen. And but it was a huge miss.
Do you run the business any differently? No. Would you be higher levered? Yes. You'd be high well, which is why I didn't want to do it.
So the mistake I made, if I view it personally, not for you all, it was actually the right thing I did for you all, if you're investors. What they wanted to do is they wanted to lever up the company, pay down the debt, re lever the company, do a dividend. And that was the way that the DCFs worked for them to get the IRRs. And my thing was no way. I said there's no way we're going to lever up the company, pay down the debt and then re lever up the company.
I said I won't do that to Seagate. And because of that it changed the cash flows right for them, because they were going to have this big bullet coming in after year 3 when they could refinance. I just said no. In retrospect, I was completely wrong. Obviously, it would have been easy to do that because interest rates went from like 6% to 0 and we wouldn't have been burdening the company at all.
So those are just the way things work out. But do you run the company in any business? We don't run the company differently. We run the company for the long term. We run as if it's our company that we own that we're paying back the debt and we're paying dividends.
And so what gets easier is you don't have to do any of the public stuff and that is easier. It's 20% maybe of my time.
Your tax structure is very favorable. Is there some thought that maybe again 5 years from now, are we going to be saying, well, that was something that was a huge asset that we didn't pay attention to that turned out to be pretty interesting for the owners of the business?
Potentially. Our tax structure is the best tax structure in the world. And it wasn't reverse engineered. I mean, it was completely it's completely legitimate in terms of how we got here. There are aspects to our tax structure that have to be managed in terms of where we have our holidays and when those come up and what do we have to do in order to make sure that we can keep those incentives in place and those incentives obviously relate to economic activity in the countries that provide the incentives.
But in terms of the overall structure, it's a favorable structure and there's a lot of people who will promote kind of doing M and A to take advantage of it. My answer is a little different. I can't remember who told me this is some professor that I like to basically don't ever let tax kind of motivate you one way or the other. That being said, I think there's opportunities that could be strategic and or tactical that give us an advantage because of our tax structure and we should take advantage of that. We shouldn't do it because of that.
But if a particular acquisition or merger looks good tactically or strategically, and then in addition to that we have this great tax benefit, then that is value add that we can bring to the table, which is quote gives us a bidding advantage, which I think is interesting. The M and A world is getting really interesting, especially for kind of non we don't have to make a profit companies. In that, there's lots of assets that are going to be for sale, B assets. And B doesn't necessarily mean they're not good. It just means they're not the ones that don't have to make money.
The ones that don't have to make money, only the companies that don't have to make money get to buy, kind of by definition, because then they could they get even more valuable. For those of us that have to make money, we have to buy companies that make money. And if we buy companies that don't make money, you all get mad. And whether or not they make money right away or over 1 or 2 years, we can usually get some flexibility on if you believe we can deliver. So what's interesting though is how many people how many companies can really buy those B assets?
A really small group of companies. So what happens? They're all going to private equity. This last transaction on Veritas like Carla is going to knock it out of the park on that thing. That's a great asset.
It's sticky. It's great cash flow. They have a play of getting into software defined storage. But there was no strategic buyer, because all the strategic buyers are selling stuff. And we could have been a buyer, but it would have been way too in our OEM space right now.
Maybe in 3 or 4 years when the OEMs have kind of left that market, they'd actually not be threatened by us doing something like that. So for Seagate, as you kind of think of this lots of cash flow kind of sustain, we believe in the core technology, keep bringing the share based on to holders that actually get the long term story and do the work, you kind of become a private company with a public stub. And what I like about that is it means if the only company I'm bidding against a private equity company and I have strategic value or tax value, I can win those. And we can all do very well by that. So I do think the M and A world is shaping up really nicely for companies like Seagate.
So we agree that we're all not going to be asking the same questions 5 years from now. Is it a high likelihood that whoever is asking the question is going to be asking to you?
A high likelihood? Well, I hate to challenge whatever God has in store for me. Yes, I don't know. Maybe. I'm having fun.
The team is great. They understand I have we have 5 babies, 4 babies, 5 babies, Agatha just freaked out. 4 babies. I keep pushing for a 5th, but and they're great to let me spend time with them and make sure I'm home and take the kids to school and adjust my travel schedule. So for me, work is good, the team is great.
I'm having fun. I'm more excited about the future of technology than I've ever been. Maybe it's just because I'm getting old. I'm worried about the state of the world, but I do believe that technology can solve a lot of the problems if we focus more on the important stuff versus the social media stuff. Not to slam social media, but there's just bigger problems to solve than should Twitter go from 140 characters to 280.
And that's what we do. To me, one of the things I worry about not worry about, I think about is the shift from consumerism driving technology versus enterprise, whether or not the enterprise is a company or government or a large organization, is that the technology becomes a bit trivialized, kind of by definition, it's solving consumer problems as opposed to big important problems. And I worry that if that's what's driving the market, then that's how technology start getting gets optimized for that versus getting optimized to solve the really big important problems. And it's our job I think as one of the key providers of important technology along with the CPU guys and along with the networking companies to make sure that we're providing some sort of environment to consistently push our technology in solving those bigger problems. And if we don't that worries me, because then maybe we are buggy with.
If all people are going to do is look at a tiny little screen that has a thumbs up, then yes, you don't need to just drive for that. But if that's the state of the world, we're all in trouble probably for a lot of other reasons. What a high likelihood you'll be here? On the planet Earth, I sure hope so, yes. And on this stage, yes, I hope so.
I'm having fun.
Yes. Kate, do you want to throw it open? Should we throw it open or keep going?
Go ahead, Steve. Yeah. Whoever gets the mic.
That's a tough act to follow, but I guess Steve a question of 2 parts. You're talking about the 5 year CAGR that you had there. Part 1, if the math works, but client keeps getting smaller and smaller for you guys, both from revenue exabytes and shouldn't sales start to accelerate as that mix starts to go more enterprise heavy for you? And then secondly, if I look at those numbers, it would imply the price per gig cost down that you give to your customers is in the mid to high teens. I'm curious why do you think you have a reason to do that especially in enterprise where you don't have 5, you might have 2 players really?
Yes. You probably have 4. You have Seagate, WD, Hitachi and Toshiba. And there I would agree the pricing why you do it is because you can afford to do it right now. The margins on those products are strong enough that given the erosion of the client, you can actually increase the price erosion at the high end of dry volume and net net end up delivering the model that you want to deliver.
I think that's not smart, which was kind of what I said 3 calls ago, which was meant to be a particularly strategic dialogue, which got into interpretation of a tactical dialogue about pricing. That was my point. My point is that to deliver a 10 terabyte or 15 terabyte or 20 terabyte drive by the in the timeframe that we want to do that. The R and D and capital is significant and it's going to require a margin structure that's above the 28%. And so if you give all that away right now, how are you going to recapture it?
The only way you can recapture is with some sort of supply dislocation, which might happen because there may not be enough of those devices, particularly if the data science world takes off and says there's value for us keeping every piece of 8 ks video we can find. But my point is people should be smart enough to figure that out now. Part of that again is I think driven by the structural oddity of Hitachi and WD, which is being forced by a regulator, because they go after each other really aggressively. Hitachi is particularly strong in 2.5 inches and WD and 3.5 inches and so Hitachi has been aggressive in the 2.5 inches marketplace, which has done more damage to WD than done anybody. And like why that happens, I don't know.
But the way you subsidize it is basically by paying for it with the higher end drives. And I just think that doesn't make sense long term. So it should get better. I think it is getting better by the way. But it's a strategic discussion.
It's not again, this call got caught up in some tactical quagmire afterwards, which is not the point. But I think your general thesis is right over the next 3 years. Yes. I'll repeat it. There you go.
Okay.
Yes. It's Rod Hall with JPMorgan. I just wonder if you could talk a little bit about your how much you're thinking strategically about supply of the cloud providers, who to supply, who not to supply and whether that thinking is changing at all in the last couple of quarters, so we've definitely seen some dislocation in that market.
Who to supply and not to supply? I don't know what to sell to anyone. But I
mean, do you strategically think about who's going to be the bigger provider in a couple of years? Who's not going to be?
No, we don't have to. Do we think about it? Yes. Do we have our own opinions? Yes.
Does it change our behavior, which is maybe at the margin that there's if it's obvious. There are obviously obvious CSPs who are going to be buying a lot of disk drives for a long time. And yes, we take resources from OEMs, let's say, who we don't think will survive long, long term and reallocate not person for person, but dollars being spent. But it's probably the biggest challenge we have. Like I've said, to me, it's not the technical stuff that I worry about.
I think the technical attributes of what we have to do, whether or not it's with our silicon based device or whether or not it's with our magnetic based devices, we're pretty firm on what we want to do and we have a high degree of confidence that we can deliver to it. The go to market side is where I worry more in terms of our ability to make the transition, because it's not the same people. It's not selling to a procurement officer at 1 of 20 companies. It's selling to a CIO or to a BU at 1,000 companies or to VARs. And the only companies that have ever done great with VARs is Microsoft and HP.
Somehow we have to actually crack the VAR channel effectively, because they will deliver to smaller companies that are deploying cloud architectures. That's so that's a challenge for us. But do we make bets? We don't really kind of make bets. I do believe the OEM market will consolidate drastically.
But I don't have to pick the winners and losers, right? We sell to all of them. And as this company starts winning, we just sell more to them, right? And we don't favor them. We don't where it's going to get interesting and where we will have to make that discussion to that decision is what happens when people actually start architecting stuff differently, which is happening.
So the implementation at cloud service provider 1 versus cloud service provider 2 is very different, because their workloads are different. But they're both so big that we put technical resources on it to say, okay, we'll make a device or a subsystem or system that addresses that workload to either drive your performance or your cost or whatever you're trying to achieve. The question will be interesting is, what does that starts happening with other customers and you're limited in resources, how are you going to make the choices? Then we'll have choices to make. That's not happening yet, which is maybe why some of the OEMs aren't maybe growing as quick as they can, right?
Because what they're still caught in the old world of more suppliers is better. If I go to a defined architecture that results in me being sole sourced on some technology whether or not it's drives or anything else, They feel like they're losing some sort of leverage when really what they're gaining is a better product. And so it's going to be there are companies that address that differently that are more or less comfortable with sole sourcing technology, again, whether or not it's us or someone else. The ones that are more comfortable doing that, we'll put more resources to, because we think they'll win.
Could you just make a quick comment on what you think the recent changes in China and the economy might be doing to Mofcom's timeline? Do you think it's extending the timeline? Do you think it has any impact at all on what's going on from a timeline point of view?
As I said, there's people that write like books, really smart people that write books like this thick on trying to understand China. And I'm not one of those people and they haven't written a book like that. I have no idea. Although, I don't see anything happening before October, because that's when they're having their big Congress and they're going to go set the next 5 years. As important as it is to us or WD, in the grand scheme of things, is that important to China to decide some disk drive merger that happened 4 years ago?
I just can't imagine that's like showing up on the top 100 things of the government to deal with, which makes me think it won't get dealt with. I mean, I think the lack of understanding about how decimated the middle management is right now and whether or not that's middle management at SOEs or at in the government itself, it's substantial either from a capability perspective or people have been removed. And so the decision making is kind of non existent. And those remaining, I think don't really know the conditions set under which they're supposed to be making decisions. Tells me it's probably a longer time than not.
That being said, as Mosley said, if we woke up tomorrow and they told us we're done, I wouldn't be shocked either. I think it's completely random. Therefore, I don't guess. You've had your hand up for a while.
Aaron, I think
I've done the person right in front of me. Go ahead. Me? Yes, go ahead. Okay.
Thanks.
So two real quick questions, Steve.
So first of all, as you look at the weakness from a demand perspective, are you concerned at all at the upstream supply chain as you see particularly the guys participating in a per unit level? Does that concern you at all? And then the second question would be, as you look at I think a year ago, you talked about upward levers to the model, even talking about the potential for an 18% to 20% operating margin. Do you still believe that is something attainable over the next few years? And if so, why?
Well, the first question, yes, we do worry about it, but it's one of those things that get again, I probably like the word think better than worry. I guess I think about the stuff I worry about. But we think a lot about our upstream supply chain. I think the one of the huge outcomes of the flood was forced you to think about your upstream supply chain even more than you did. And clearly, if you're in a single part per unit world, even though for some of those parts they're getting more sophisticated, right?
The motors inside those drivers are a lot more complicated as well. But it's more challenging for them. And then the question is what's the structure of their industry and how do you think? Yes, we think a lot about that. And we've done a lot of things to kind of protect ourselves and them as a result of that, because it's important technology.
I mean, you need all 200 parts to make that thing work. And some of it is stuff we control directly, some of it is investments, some of it is engineering help. But in general, we feel fine about the supply chain right now relative to the demand levels where they are in terms of units. And on the second piece, I still think, yes, we do still believe there's leverage to the model, probably and mostly driven by we think that 57 exabytes goes to multiple 100 of exabytes and that's all in devices that again absorb way more heads into this could have way more content that has way more sophisticated channel, lot more test time. So you're basically you're getting just massive absorption.
And at the end of the day, we're a manufacturing company. So the leverage on absorption is huge. Yes. Oh, sorry, we're going to go here. You still have a quick sorry.
Just quickly, the demos this morning focused a lot on nearline and flash based products. What are your thoughts on mission critical after EMC talked about that low teens decline both from a growth and an investment standpoint?
We've said kind of consistently that legacy is going to last a lot longer than people think, and I still believe that. Where the cloud based systems have been most effective is in the applications that you can kind of pick off. It started whether or not it's CRM or sales or I mean HR. But ERP and financials are starting to convert a little bit. But the big legacy systems, it's just so much cheaper to add to the installed base to deliver the functionality than it is to transition.
And I think those transitions are 5 to 10 year transitions. So I think the legacy systems are going to be around for a long time. Maybe they're not growing at 15%, but I don't we haven't seen any decline in legacy. If anything, it's exceeded our expectations every quarter. It just seems to be really solid and we don't see any change to that for the foreseeable future, notwithstanding the fact that people are starting to move more applications into the cloud.
Go over here. Go here and then we'll go to Rich.
Yes. Thank you. Steve, you talked about flash is still very expensive. We also talked about the leaders cost technology leaders and cost leader. The one in Korea Samsung is actually driving down prices because of technology and cost.
Retail level SSD is down to $0.30 to $0.40 per gigabyte. Enterprise level is at or below $1 using 3 d NAND. In that context, what is your contingency plan? Let's say for just hypothetically, flash gains more momentum, SSD gains more momentum and you have more adoption of flash arrays combined with caching that would capture a bigger market.
It's not storage. It's not storage. I hope it happens, because if it does, then whatever data is being processed in that fast memory is going to have to be stored somewhere. And at $1 per gigabyte, we're at a penny, okay? We think they're going to go to $0.15
per gigabyte. But this is 32 layer. As they migrate to $0.40 and $0.60
We expect them to go to $0.15 which is a very aggressive projection. If you believe in 3 d and there's a lot of people by the way they don't believe in 3 d who are really, really smart. At $0.15 we're at less than $0.01 So if it's for storage, it doesn't make any sense to store anything on an SSD. If you're using it as a fast processor or a Tier 0 device that's feeding something that's hanging off a CPU bus, then yes, there's basically a hierarchy that says here's how data is going to move. And we believe that happens and we hope it happens and we hope it happens in data science, because if it happens in data science using unstructured data, then you're going to have to store exabytes that are feeding those gigabytes or terabytes that are feeding the gigabytes of DRAM or whatever the replaces of DRAM.
It's a whole architecture. So why do you There's no use of SSD today to speak of on the enterprise that replaced the disk drive, a boot drive. Okay. See, that's like 1% of the market.
You're talking about a consumer driven change.
No, I'm talking about enterprise.
Right. I'm talking about enterprise. Enterprise is also changing given the change in consumer. What if the consumer is not going to store for years years? What if we're just going to be focused on storing unstructured data for 2 years, 3 years and then you just flush it out and then move on.
And that's a bunch of unstructured data that they're not storing today that they will store tomorrow. So that's great. It's all net positive relative to the world today.
So when are we going to see it in the model? Is it just we're at the verge of that transition where as data growth accelerates the need for storage capacity going to increase too?
Just at the beginning, have you seen the 5 years of data? Our average capacity per drive has gone from 400 gigs to 1.2 terabytes. The number of exabytes shipped into the cloud has gone from 6 exabytes to 57 exabytes. 57 exabytes is the total amount of Flash produced that's compute grade.
But earnings are also declining. Excuse me? But earnings for the company on a fiscal year basis are declining.
For the last fiscal year? Yes, we have cycles. Sure. We have cycles. And that's why we're making adjustments to our OpEx to adjust to the revenue line.
A lot of our OpEx was on stuff that we were promoting that we decided wasn't working the way we wanted to. I mean, now you're arguing about Seagate's financials. I mean, you want to argue about financials, I think our financials have been pretty good relative to the rest of the world. But if you're talking about how our technology is being used, we hope that the flash guys can stay on linear density curves. If they can't, that's my biggest concern.
I've been saying that for 5 years. These are not competitive technologies. If you look at where the devices are used, they are not competitive technologies. 1 is being hung off a CPU bus, the other is being hung off a storage bus. If people solve the problem of compute of solving like analyzing a massive amount of data and they need Flash to do that, God bless them, because you got to store the data somewhere.
That's on a disk drive. It's the only place it's going to be. It's the only place cheap enough, stable enough that, oh, by the way, you can write to. And our drives get banged on 20 fourseven with reads and writes. You're not going to do that in Flash ever.
You're using Flash for a completely different purpose. And it's complementary. That's the argument for what happens to consolidation between flash and HDD. It's complementary. It's not competitive.
Yes, it's competitive in client notebook. It's competitive like we were never in a cell phone. It's competitive in that 0.001 overlap of a boot drive. But what we gained from the fact that there's 3,000,000,000 devices connected that are capturing video and sending it all over the world and being stored somewhere on a disk drive, I'm happy to give up some incremental amount of drives that we lost that we're getting paid 5 points of margin on. It's just that's the same argument for the last 5 years.
It's not a war with flash. In fact, our drives will have gobs of flash on it. We're going to have 200,000,000 devices that are going to have somewhere between 8 32 gigs of flash and we'll be probably one of the biggest customers of flash.
I want to go back to the point that you mentioned about Star Wars copy that we're not going to have one copy. We're going to have many of them, which I agree with. But the problem, they're all on Amazon. I mean, then you have like only 3 giant guys who control the industry. You have Amazon, Google and Microsoft who have this hyper data centers and you are a supplier to them.
So their ability to efficiently add capacity is a negative overall. We're not going to have 10 of those people. And we've seen that in the semiconductor industry where you outsource everything. And for the equipment suppliers have been a nightmare.
So Yes, we don't believe that.
We don't believe
that. We don't believe that. And that's why I said, look, we have thesis. It supports our models.
But why you don't believe that? Because we've seen Netflix both and give the keys
to Because
we don't believe that those 3 companies will control a significant proportion of compute bandwidth or storage. We don't believe the architecture makes any sense. We don't believe they can finance it. What's the revenue model? Right now, each one of those companies is subsidizing their infrastructure through a different mechanism than what they're selling.
So we give storage away, but we finance it with search, okay? How long do you get to keep doing that? We give storage away, but we don't finance it with anything. How long do you get to keep doing that? We give storage away, but we finance it with retail transactions, okay?
The point is and by the way, I don't think Baidu would agree with your assessment. There's only going to be those 3 people. I don't think eBay would agree with that. I don't think Yahoo! Would agree with that.
I don't think Tencent would agree with that. We believe that it doesn't look like the pharmaceutical industry, which is kind of what you're painting, that it probably looks more like the banking industry that there's 20 to 30 really big CSPs and then there's probably thousands of smaller ones. And then you probably got another set of 5,000 corporations that have hybrid models because they're not going to want their data off-site or they can't legally. And so we do not believe in the model that it's 1 or 2 or 3 or 5 companies that control all the data in the world. It doesn't make any sense for like a 1,000,000,000 different reasons in my mind, whether or not it's security or whether or not it's just latency and functionality.
But we believe in a hybrid model. And the consensus of most people that are deeply involved in technology is that it's going to be a hybrid model. And that means there's opportunity for us the other way. You're kind of saying, yes, if your world is you substitute HP, IBM and Dell for Google, Amazon and Microsoft, does that really help you? No, it wouldn't.
Because the mentality maybe slightly because they like stuff for their workloads whereas the other companies don't. They wanted everything to look the same, so slightly. If you're engineering driven, you might be able to get better market share. But we don't believe in that world. And if you believe in that world, yes, you don't then yes, you wouldn't believe any of our numbers.
The boss is
That's the boss. Sorry.
Yes, sorry.
Well, I shouldn't I'm so rude to my guess. Do you have any more questions? Were those interesting questions to you?
We already got
higher multiple and better earnings.
So I was done after the first one.
So first, I'm going to let me again thanks Harold for doing this. We're going to move to a quick break and then Mosey is going to go into some technology. But we have actually 2 sets of gifts for you depending on if you were a nice guy or not. So we basically we had disk drives because that's what we do. So we have that.
But more interestingly and important, we have this is not whiskey. So we have a case I have a case of this is this great olive oil that we make on this organic farm that my wife and I have been involved with. There's a couple of you in the crowd actually that know this Senator Bradley and Rick Schutte. Where's Rick? Rick didn't hurt his arm by drinking this stuff.
So anyway a case of the olive oil from our farm and sink. Thanks. I'll be back. Thank you very much. Thanks.
All right. Thanks everyone. So we're going to take a short break, a 15 minute break and then Dave Mosley will come up and talk about products and technology. Thanks.
Kate to take her seat. Thanks, everyone. Welcome back from the break. It's very good to be in front of this audience again. I especially enjoyed the interchanges this morning back in the technology demo side talking about some of the products.
And I'll give you some highlights again in this presentation. So sorry to be redundant there. Just to start off though, I'm going to talk about the markets. There were some Q and A about some of why we believe in the growth that we do. So I want to underpin the rationale with some data.
This is a slide that we showed last year at this time. Fundamentally, it hasn't changed. We still believe we're on the same trajectory for amount of data being created in the world and the amount of data that is going to be useful if tagged properly and how that data is actually being processed is a very hard transfer function to get from there to how many devices do we have to make in FY 2016 or 2017. For the people that are running factories and large scale supply chains like me, that can be a really tough challenge. That's one of the reasons, as Steve talked about, we do make investments, but we're pretty conservative about it.
In general, though, the trend that we said about we talked about last year is still definitely happening, which is most of the data is moving to the cloud or new applications that we'll call cloud like applications. And there were some questions about that and I'm sure we'll continue that dialogue throughout the course of the day. So what do some of the new applications that we're servicing, what do they really look like? I'll start with just the general theme of video. 4 ks video is a video that's rich games that are rich in high definition video, virtual reality on the bottom right there.
We're seeing a lot of movement in, for example, DVR businesses that we have, surveillance businesses that we have towards a lot higher definition file sizes that are much bigger, so on and so forth. And what that means is higher capacity points for better value proposition. So if we can get more gigabytes, terabytes right out in some of those applications right out at the point of sale, right out at the point of being analyzed, it's a good thing and it improves the experience for the consumer. I'll talk about Internet of Things just really briefly, but we do believe that that's also a key driver for the data growth that we're seeing in the world. It's probably still early.
And of course, we're looking at things that create more than a single bit of data. We just don't want a temperature reading from out from in your house. We want devices, call them robots for a second, that are full of sensors and all that sensor data to be captured. And we're pretty enthusiastic about some of the customers that we're working with on some of those new file systems, operating systems to be able to capture and process their data. Steve mentioned 3 d printing as well as another new client that we're very focused on, not just from a how do we service it perspective, but also how do we use it in some of our manufacturing processes, which I hope to in a video later give you an appreciation of.
Mobility is still there and the world as we all know people are taking more and more pictures. They're taking more and more videos. Some of the platforms for capturing those pictures and videos are getting better. They're not intuitive yet and they're not the data doesn't flow from device to device that easily yet, but we still believe that's a trend that over the next few years will be solved. And then those videos will that very personal diverse content will be easier edited and shared throughout the world, and we think that that's a great opportunity for us in a lot of ways.
And then big data analytics, and I'll talk about workloads there a little bit. But big data analytics, we believe there are many different kinds of applications that are coming from some of the things that Steve talked about to some mundane things even we see in our own factories where we're using video analytics and other kinds of analytics end to end in our manufacturing process that to reduce costs and improve our quality. So those are the big drivers. As far as the devices that we produce at the very base level, I said just a $50,000,000,000 business here, maybe it's just a $60,000,000,000 business we could argue. But we service all of these applications at the higher level with a fairly small set of devices.
And we can tailor those devices for the specific applications. Those applications are growing very quickly and they're also creating a lot of architectural and business disruption. So it's really tough to know exactly what you need to build at the lowest level. We stay focused on our key components. We should be okay.
If you look at that from a financial perspective and I think Steve made a couple of veiled references to this. As a percentage of revenue, so look at the chart that goes from 37% in FY 20 10 to 57% in FY 20 15. The stuff that's not highlighted, think about it that way, that's PC. So PC is becoming a smaller and smaller portion of specifically where we get revenue. And some of these new applications, surveillance or DVR or gaming or the cloud business critical, the 6 exabytes going to 57 exabytes that Steve made reference to.
That you see that portion of the transition is going. So a lot of people ask us questions about what about the PC. Well, the answer to that question is we've been in the middle of this for 5 years. Our thesis is not that the PC goes away from here, but there has been these shifts already happening within our business. At the top of it all, and I can't overemphasize this point enough, is application layers that are changing a lot.
That's really what the customers make the buy decisions on right now. And there are various layers underneath that of software that can optimize our systems. At the bottom of it all is the components that we actually manage, whether we make them ourselves or we buy them and then manage them into systems. And I think or into individual devices. And I think it's very important to realize that some people make components very well, but they don't make devices well.
What Seagate has been doing for years years is making devices very well. We've been through a number of interface transitions in our 35 years. And we've been the person who shipped the lion's share of fiber channel devices and the lion's share of SaaS devices and so on. We've been through those kinds of transitions before. We've also been through things that are more subtle than that.
Transitions like putting DRAM onto our product and managing the caching and tiering that goes on. Putting NAND on our product, I talked about this morning that we've shipped 19,000,000 hybrid drives and people ask me, well, when's it going to go big? So we're very good at integrating these systems together into products that answer the call for application layers above us. And we do that through a whole different set of customers, compute layers, storage layers, specific storage layers and others. I want to give a specific example just to drill down a little bit on video and what it's doing to us.
And maybe based on some of the earlier Q and A, I'll tailor these comments just a little bit, if you don't mind. It's not just about video, but it's also about file sizes. The file sizes that we see hitting the hard drives are getting more and more sequential. So they're not random small block random writes and reads, okay? They're more and more sequential.
They're getting bigger. The average mission critical, business critical and certainly surveillance file size is over 100 kilobytes right now. And there's some I mentioned Hadoop earlier in the slides. There's some that get up to 100 of megabyte file sizes. What that means is the drive is being used sequentially.
Now there's a lot of people that assume that drives are still being used as random block small block write and read devices. It does happen occasionally. So sometimes workloads leak through. But if you think about video coming at a device, whether it's in a surveillance workload or in a DVR workload or maybe in a consumer client workload that's being generated in order to that we might do at work. I mean, I personally am using more and more video at work all the time.
Those things are large files, very large files. And the way that those workloads tend to hit the drives are sequential workloads. And again, back to the earlier point about, well, but isn't performance can't you wouldn't you define performance as small block? We used to in the old days. We used to we defined these we talked about these things called IOPS, which was really a hard drive metric around how we address blocks and then we talked about our seek times and things like that.
But we haven't done that for like for about 10 years. What we're being pushed really hard for right now is to improve the throughput into our devices. So the devices are being asked to write more and read more. Our workloads are twice as heavy as they were 5 years ago. So not just the capacities of the drives are going up, but the workloads are going up as well.
Steve talked about 20 fourseven in the data centers. The average file size that's being hit in a cloud service provider application is 66% sequential, so enormous file sizes. And we're answering the call for that. We're designing products specifically. We talked about some of the 8 terabyte drives back in the back of the room that are specifically going after those workloads.
And I can't overemphasize that enough. We don't see one application. So there's a lot of people that talk about how good their device is, their system is in one application in the world and then they try to extend that to many applications, many other applications that they may have not be tailored for. Seagate actually sees many linear combinations of different applications. And we try to bring the right tool to the right application combination, So right job, right tool.
And it's for us, we have an array of assets at our disposal. We can use NAND. We can use DRAM. We can use HDD and we do. And we're starting to see a lot of bifurcation in the cloud from high capacity to high performance drives.
We still see our mission criticals. We see some of our mission criticals are largely sequential size. Some of them have to hit a low cost point. And then obviously our SSD one of the best things about some of the products in the back of the room I was mentioning this to someone earlier is that our SSDs are giving us insights into the use cases of the compute layers that the kind of the leading some of the leading thinkers in data centers and enterprise are thinking about in the future. And we may tailor some new SSDs from that, new interfaces.
We actually may tailor some new looking new devices that are new combinations of device. So again, pick the right tool that we have at the bottom and apply that to the application space. That's really Seagate's game. Okay, I'm going to shift over to technology a little bit. The core technology of Seagate is hard drives.
And if we keep pushing areal density, we will continue to monetize the hard drive. We keep pushing areal density. We'll continue to monetize that hard drive the hard drive investment. Aerial density has been kind of stale for the last few years. It really has.
You can see that in client devices. 500 gigabyte notebooks have been shipping for a while, 1 terabyte desktop, so on per platter desktop has been shipping for a while, 3 or 4 years. We've made significant progress in the last year in the labs. And so we're going to talk about a bunch of new products over the course of FY 2016. That fundamental aerial density those gains will continue to keep us at least 10x maybe 20x better than any competitive technology relative to cost.
And we believe that fundamentally, we could have large arguments about it. But that's Seagate's position and that's why we keep investing as aggressively as we do in aerial density. Relative to other cost efficiencies, there's things like reliability that we make parts and make drives that are more and more reliable all the time because we know that's helping the total cost of ownership for our customers out in the field. Power is another vector that we're pushing very hard on in a lot of different ways actually having silicon that goes to lower powers. Power nodes is very important in data centers.
So we're continuing to invest in cost technologies all throughout up and down our supply chain including inside of our own manufacturing. Talk about optimizing performance. I mentioned that a little bit in video before, but every different hard drive, SSD, hybrid drive, so on needs to be performance optimized for the application that it's going into. And so we spend a lot of time the world's getting a lot more diverse. We don't have just a few OEMs anymore.
So we spend a lot of time out with customers right now understanding their workload and the tuning. So that's a lot of applications engineers. And then client mobility, some of the drives we talked about this morning, 7 millimeter thick, we have one that's 5 millimeter thick out in the world. We realize the world doesn't want to be carrying around heavy thick devices anymore, but may still want the ability to carry around a lot of video for good use experience for them. So we're going to go optimize around client mobility as well.
This kind of looks at this same summary slide on a different scale. And you go from devices that we make in Seagate that are pure archive devices all the way over on the right hand side to the highest performance devices, storage and memory devices that are serving PCIe buses into the compute layers. We also have products in the back like our cluster store products that do this kind of thing with combinations of technology at massive scale as well. And then we have all the assets kind of end to end to be able to address the customers the right way. This slide I showed last year and it reinforces some of our LSI acquisition rationale.
We're really happy with the progress here. If you don't have controllers access to controllers that are talking to one another at all different layers, then you may be hamstrung in the future. If you don't have best in class controllers to control hard drives, to control NAND, to control what DRAM technology the DRAM technology that you might be using or any other kind of technology that you might pull from the hardware shelf in order to address an end customer application set then you may be competitively disadvantaged in the future. So we still believe in the investments that we made and we're still continuing to make a lot of progress on that IP development. And then as far as aerial density goes, I've made a few references to this before, but our current products that we're shipping are traditional perpendicular recording and also shingled magnetic recording.
We've shipped about 4,000,000 shingled drives. Now a lot of people again they view that as not successful. We view it as very successful because we've learned a ton about the customer applications and we're ready to be able to hide the shingle behind tiers and caches, so that in most market segments, the fact that it's a shingle drive is irrelevant to the end customer. They don't have to do any tough application changes. Now there are a lot of movements afoot just to be completely transparent here for end customers who want to, to be able to tune the host.
So in the parlance of this, it's host aware and host managed SMR. So they can tune the host software so that it never gets into a problem with SMR. We'll still work with customers on that, but we're also doing drive managed as well. So we'll manage all the SMR in the background. Very happy with the progress that we've made there, shipping it across a lot of different platforms and ready to go big in FY 2016.
And that's one of the big things that we're bullish about some of these new products. In the future, we talked last year about 2 dimensional magnetic recording. We'll be ready to ship that in the next year or 2. Now that is really putting 2 readers into the device. Maybe you could even go more than 2 readers.
There's manufacturing prohibitive to go do that, but it does give you having some information about your off track noise gives you the ability to increase the track density and get some more aerial density out of it. HAMR, we've talked about in the back a little bit. HAMR is still not ready for prime time. I wasn't tremendously happy with the progress last year, but there was progress. So the highest aerial densities that we see today have to be written with HAMR.
And we still have some issues working through the reliability. We've actually solved a lot of the problems, but the whole industry is through various consortiums is really focused on getting the last of the problems solved so that we can get this into products. And then we've done some demonstrations in the last year of bit pattern media combined with HAMR that gets us to 5 terabit per square inch. So that shows you roughly a factor of 5 more of capability of the magnetic media, its ability to store disks that tells us where we can go where we can still yet go in the future. As far as the new products this year, one thing we didn't talk about very much when we started shipping them, but we are shipping 2 new products that are over 1 terabit per square inch and have been for 3 to 6 months have been shipping those products.
That's kind of an industry milestone. I remember when I joined the industry, the big target was how to get to 100 megabit per square inch or something like that. And now we're talking about a terabit per square inch. It's just continued progress along these vectors. And then we've announced that the technology is ready to go and we'll roll out a bunch of products throughout the course of the year for 1 terabyte on a 2.5 inches disk, which is going to be about 1.2 terabit per square inch.
So we're very comfortable that we have solid aerial density leadership right now and ready to put these products into high scale volume manufacturing over the next couple of quarters. And as a matter of fact, the 1 terabit per square inch products have already been in manufacturing and they're already getting up the ramp. Helps us offer better value proposition to our customers whether it's higher capacities in smaller form factors or whether it's taking at the same capacity taking components out so that we can reduce costs and pass that along as well. It was a good year in the labs. So FY 2015 and just the core HDD technology was about 30% areal density improvement and we'll be putting that into the market in FY 2016.
Obviously, there's a lot of other things I could talk about inside of Seagate's manufacturing capabilities, supply chain capabilities and I'd encourage you to go out on our website and make yourself familiar with that. If you're not, I can talk about it in a soft line as well. But I wanted to just give you a flavor for how focused we are on innovating new products and the ability to manufacture those in high volumes so that we can answer our customer demand with a quick video. And with that, I'd like to thank our customers and suppliers for all their support in the last year. It's been quite trying times obviously for everyone economically.
And one of the questions before was asking how are they doing? The one thing I'd say is they've diversified quite well in the last couple of years after the flood. So, but I actually really thank them for the support. You can see how proud we are of what Seagate has put together and how ready we are for to go into production with these new products. So with that, I'll turn it over to Pat.
Thanks.
Thanks, Dave. You heard from Steve sitting here and you heard from Dave. We sorry about that. We could talk a lot about technical stuff and how proud we are. But what we're really proud about is how we think the products do really make the world a better place and allow a lot of things to happen.
So we're heightened by doing this every day because as we make our products better, we allow products around the world to get better and we really feel that way. So thanks, Dave and congratulations to his team for all he's done last year. So let me pivot my presentation and sort of where Steve started and sort of echoing some of the things you heard Phil say back in the demos, you heard Steve say up here. Seagate is a company who thinks long term. We really do, whether it's our people, process, technology, whether it's internal investments, external investments.
We like to look at it as the long game, as we said, because we think within the day fundamentally owning that IP, that technology, those capabilities allows us to scale and enable more and more products and more and more solutions and answers for ever growing complex problems for our customers. So if you take a look at this chart, it has a few financial vectors and you clearly see the theme from fiscal 2020 to fiscal 2015 that we've been talking about and there's a reason for that. A couple you heard clearly you see this revenue shifting from the client to the cloud. And we were here in fiscal 2020 and that was viewed as a big risk. But as you go through and look at this chart where you have the revenue, the EBITDA percent as revenue and the non GAAP gross margin, some of the things you see there are the story tells a lot of stories fiscal 20 10 2011 pre consolidation.
So as Steve talked folks that weren't vertically integrated they could pick up lower costs and made the industry more challenged. We consolidated, but then we get hit right with the floods and we have a major supply disruption. It felt good economically, did feel good a lot for our customers, didn't feel good for a lot of folks that needed these products. So within a year, we quickly ramped up the industry and got the supply chain back and going. And everyone thought when that happened things would get out of control because the client whether there's a tablet coming, whether there'd be price erosion or whatever the concern was that we wouldn't be able to manage it.
But as you can see that as I said the client was disappearing. Even with that, we're able to hold the strong gross margins and the strong EBITDA because we're able to manage the supply chain, manage the technology and deliver it in a very efficient way. So as you take a look at that, as that product has shifted from a client going away to whether it's SSD, phone, whatever the device or even enterprise where folks thought, well, the enterprise is going to get started cannibalized. If you go back to the beginning of the time of this chart HDD revenue was about 32,000,000,000 dollars You fast forward, it's about $32,000,000,000 So even with all the lost the client units and the add in SSD, SSD, we've had expansion. We've had expansion.
So this is a growing market. This is not a cannibalization market. And I think a lot of folks like to think as Steve said, it's not a zero sum game. You don't take one off and the other one just pops up. You grow this ecosystem, the architecture.
So we're very convicted about that. And if you look forward internally in our own company's plans, we see our revenue growth on top of that and we'll come to that. But this has delivered value like we delivered the margin through the market stabilization number 1 and the product portfolio. As much as the you might be down in units. Everyone in this room many of you were in love with a TAM.
It was a very key metric. But we like to think of that not in spindles anymore. I mean that's the world we used to think of the world, but it's really in exabytes. We ship exabytes. And as Steve talked earlier, it's how we package them is more the complicated, where do we deliver them.
But we can do that pretty quick, but we got to figure out what the customers need in line our supply chain to that. So it's the underlying technology that allows that. And then EBITDA management is through the portfolio mix. Even with this the as Steve's talked about, there's new products. Surveillance several years ago wasn't as large as it is.
That's growing. We're going to see new products evolve whether it's the client, whether it's the cloud, whether it's gaming, whether it's surveillance. There'll be more and more expansion of the portfolio and cost containment.
Part of
this, the EBITDA is we have to manage the cost and that's sort of looking inside out. At certain times, we make investments to go after the revenue. And as Steve talked about earlier, we don't see that revenue or that return coming to levels we think we'll pull back. And so as we look at this, we look at all aspects because we 1, we design our own products. And 2, as Steve said, we are a manufacturing company that has a wide distribution base of the products.
Harold, thanks for setting us up earlier. I didn't know you're going to do my homework for me, but I appreciate it because the numbers I looked at say we're pretty darn close. But if
you take a look
at the performance metrics, what's really important about this chart is really the first two lines. Where Seagate really looks to return get a higher ROIC is really invest in its own business in our manufacturing, in our design capabilities, whether it's through people, process, tools and invest in the portfolio. So if you take a look at this period of time, we've done over $10,000,000,000 in investing in ourselves in what we believe the products offer to drive value. So that's where we start our thesis. That's exactly where we start our thesis.
And that thesis yielded about $15,000,000,000 in operating cash flow over that period of time. So a significant thesis that you have to invest in yourself first. Sometimes you go externally as we've done. I didn't put the chart of $1,000,000,000 on M and A. But at times you go in get a more robust portfolio by doing that.
But most of that investment that you can see of that $10,000,000,000 much more internally. And that's yielded a strong ROIC over 30% over this period of time. And what do you do with when you're all done with it? Well, you either hold it and do M and A or you return it to your shareholders. So the shareholders own the company and we return significant amount of that through dividends and stock repurchases.
And in fact, Steve had a chart up there how many shares we took down. If you take a look at that period, it's still under $30 average share price. So we think we've been good stewards of that as well of not only returning capital, but returning capital at a good return. A lot of times it's criticized where we tend to buy high and when we're flushed with cash. But as Steve said, we've looked at it many times.
We have many different metrics. In fact, we had metrics that I hope never say again, but we had a $382 issue of many of you suffered through with us on how fast we could buy the shares. So I think we've looked at it from many, many vectors and we tried to return value. And like I said, under $30 average price, we think we've done fairly good with your assets. And therefore, when you look at this chart, what the number one investment has been in ourselves.
Like I said, whether that's through the factories, through the design centers, through the people, through the portfolio, we think we've been good stewards and we'll continue to expand the portfolio where we see need to. If you looked at this backroom 5 years ago, it would have been pretty much HDD. It is expanding. Steve talked about a lot of these architecture changes. There's really tectonic shifts.
This is much as the world went from this the PC into the office is now it's going into the cloud, these hybrids, all these different architectures happening that 1 year at a component and device level and you own your fundamental manufacturing and your IP, it gives you great opportunity to move up when you want and co travel with your OEMs or establish new channel partners. So we see great opportunity in the future. And that probably the number one thesis will be again just investing primarily in ourselves. So turning a little closer to next year. I know a lot of your folks are we've talked a lot about this with you since the July call.
And really when we look at fiscal 2016 is a year to return earnings growth. We know the world we live in whether it's secular changes, whether it's the currency dynamics, whether it's the macro, whether it's the strong or not, we see that world. So we're not looking next year to really manage earnings growth for revenue. If it happened, Dave's team, whether the manufacturing of the portfolio there will be able to respond to it. But we're taking a view this year that revenue is going to be down this year given all these challenges, but we're going to manage EPS through a stronger mix.
And that mix is going to come through. Why is that happening? Well, this tremendous data growth, exponential data growth, which is going to manifest itself into a lot of cloud drives. In fact, even our client drives, Steve talked about earlier, the client drives are going up in capacity per drive. So all segments in our product line are going up our capacity per drive.
But what's really going to fuel the model, which we've been saying for the last several years is the cloud. As more and more people deploy high capacity drives, which require rich content and heads of media process test time, there's more value there. So we're going to even though this year is going to be a challenge on the revenue, we can manage the gross margin as the year transcends. Now we believe that. Steve laid out assumptions if things get worse on the macro front or the spending front, we'll adapt and adjust.
But we believe that where we see these high capacity drives, high performance drives accelerating through the year should be able to allow us to offset that revenue with margin expansion. Also in our adjacent businesses, whether it's our ES businesses, we like our portfolio this year. You heard Phil talk about that. That should be revenue accretive this year and should also be relative to 15% earnings accretive. That goes the same with our systems business.
So we feel very comfortable with those businesses that we're focused on relative to 15 will be stronger both the revenue and their earnings. Cost containment, we like to say it's all top line growth. But like I said, we have to live in the world we live in and we have to adapt. And so we're looking at it from the inside out. We're going to focus on operationalizing some of those investments we made for the last couple of years, whether it's through IT systems, whether it's portfolio.
We think it's now time to sort of harvest those savings through the OpEx line, drive efficiency through the factories, drive efficiencies through our business processes and take that down. We talked about in the call about that OpEx being significantly under $500,000,000 a quarter by the second half of the year. We have good line of sight of that. Very confident that things happen. And as Steve said, we're probably not going to really hurt the portfolio.
That was one of the things we had an edict is to protect the revenue, protect the good business and not really hurt ourselves strategically for long term. So we feel fairly comfortable that we can execute that and we'll redeploy and we'll repivot as time goes on, whether it's more opportunities to extend the channel, whether it's more opportunities to extend the portfolio, we're always open to invest that. But this year is one of cost containment to help that earnings growth. And then continued share redemptions. Last year fiscal 2015, we did 19,000,000 shares.
By the time we did the call in July, we had done 20,000,000 in the 1st month. So we're not shy buying back the shares when we see there's value. We know how to run the business, what cash levels we need. We know certain things on the working capital model we want to make tighter again. I think we would like to get that get more cash out of the business through our cash conversion cycle.
We have plans to do that, but we feel fairly confident that we could we know the levels to run this business from a day to day business and long term give us opportunities. And then we'll continue a strong capital allocation policy, the chart before. That's where it really drives a lot of our thinking. Invest in your business if it makes sense, which we've done. And then also when it's all said and done with that excess cash flow, what do you do with it?
We'll manage that. And then the other part of that capital allocation is the investment grade debt. We're very proud of how that. We really went after that investment grade because it's philosophically how we run the business. And therefore, we want to put that structure around us because we feel that that's a prudent way to run it in good times and bad times, probably more important bad times or strain times.
So we feel fairly comfortable where our position is. And everyone on the management team, Steve, makes sure that we're all versed in what our capital allocation policy is because within a day, it will land a limited resources. We have to be very, very thoughtful. And I think these charts show we've been fairly, fairly prudent. So resilient financial model.
We say resilient because as Steve talked about in fiscal 2010, we had all these risk factors in front of us. All these things are going going to come and haunt us. And yes, some of them we have to go through. But I could say for the most part, we've managed that business. You saw the earnings whether it's a gross margin, whether it's EBITDA been very, very strong.
And we've used that cash like I said into the business or to deploy either through our stock buyback and dividend. Revenue, 2% to 5%. I know a lot of folks in this room are probably saying how do you really going to have 2% to 5% on that. We realize that fiscal 2016 is going to be a challenge on that. But we do believe there is a stabilization of the client.
We do believe there's an accelerating growth of the cloud and high performance, high capacity drives that help drive that revenue as well as our adjacencies. We're just getting going in some of their ES products launching. Phil talked about our enterprise SaaS SSD. There's clearly a market there that we see revenue. And on the cloud based, this is the systems based, we see opportunities.
So we feel fairly comfortable in that long term. So will you see that every year, every quarter? No. But will you see that over time? Like that's why I said we extend these trajectories.
When Steve had his first chart, he showed we had 4% revenue growth over that time. We do believe long term, we can sustain that. Non GAAP gross margin, probably not in position really to change that from the model today. But since we introduced this model several years ago, I could say that we've checked the box every quarter. We've always said we won't may not get there every quarter.
We may exceed the high end from time to time. We may broach the low end if things the macro or something wasn't working. But the point is we'll adjust to get back to it. This is a sign of health. We will just move the business back there whether it's managing supply, whether it's managing the portfolio, whether it's managing the economics that are transacted out there.
Non GAAP operating expenses, 13% to 15%. We were running a little hotter than that. We asked everyone a year ago that we were going to make investments in operating expenses that we saw some revenue growth and we didn't see it come to fruition the way we want, we'd cut back. You'll see that come back in line this year. That 13% to 15% will be well within our range this year.
And so we'll manage that and that's going to be a key to the earnings growth. And then capital expenditures 6% to 8%. If you take out some of the things we did optionality to upgrade our design centers over the last few years, which I think was the right strategic thing because these now are in place and there'll be assets good for 20 to 30 years, gave us great capability for the long run. Even within that, we ran under the 6% to 8%. If you took those facilities out, we'd probably be in the 3% to 5%.
That's probably an issue on the other end if the industry keeps on underinvesting because I'd say we are underinvesting because of where we see the world at today. Yet we see long term it's going to be a need to make those investments. It's going to be a breakpoint somewhere down the line. We'll just have to manage that when it comes. But some period of downtime, we may have to step up the CapEx to go after that revenue.
But we'll probably be back on our heels and not make those investments forward looking. We'll sort of probably wait till that business comes to us. But we got to be mindful of that. We don't want to be too far back on that if there is a pop in some of the applications that Dave and Steve talked about. So the capital deployment assumptions, a pop in some of the applications that Dave and Steve talked about.
So the capital deployment assumptions, nothing new really on this chart from last year other than last year we had done a dividend raise and now we're at 2.16%. We talked about last year keeping that dividend raise around 10%, a minimum 10% for the next couple of years through 2017. I can say here where we see our cash flow and where we see the fundamental basis of our business, it's probably pretty sound. Now that's a per share because sometimes you'll get actually higher because it's not a dividend pool because as we retire these shares, in some ways, it's self funding. But we look at it as a per share.
So we'd expect to have some sort of increase on that, and we feel fairly convicted on that. Share redemptions value based. If we see there's a dislocation in the market and there's opportunity, we will be willing and active buyers to support the stock because of our long term beliefs and like Steve threw out a few metrics that we look at. But we look at a lot And we also try to look forward to where the world is projected to go in. We're not going to be right on that, but we want to take some.
We do look from the outside in on that a lot to say where the world's heading and try to make some best guess. We reach out to a lot of you. We reach out to a lot of our investment banker economists to figure out where things are going. So we try to really do our homework and be very thoughtful on that because like I said, we do want to be prudent with that because it's optionality what we do with that cash and more buying stock we're feeling that's a pretty good bet. Debt retirement, that has been opportunity dependent.
We've been really good at getting a lot of the high cost debt off. This upcoming May, we have an opportunity to take that last high cost debt off, which we probably would do. But we've been looking at we still run the strategy of looking at our debt profile three-dimensional, whether it's spacing between the tranches, whether it's the magnitude of the whole debt and then what the cost is. We think we've managed that pretty well with good spacing, good cost debt service significantly lower than it was at fiscal 2010 even with more leverage on the company. In M and A strategic investments, we look at a lot of things every year.
Obviously, don't close many. Steve said we are very conservative. We would like to make sure that we see a line of sight where that investment is going to make money for not only for the company or is going to add some strategic element of IP, manufacturing capability, people capabilities, customer expansion. That's how we go about that. And so we're very prudent and very somewhat conservative on that, but we think we've done pretty good track record on that as well.
So with that, I think I'm going to ask everyone back up on stage and we'll go through some Q and A here.
Pick?
Yes, we won't need this.
Put Phil in the middle and make him answer all the
Rich?
Hi. So I want to ask 2 questions. One is on visibility. Obviously, we have had the PC OEMs over shipping laptops into the channel. Certainly, disk drives on the other hand, it's been what Steve 5, 6 years since we've had excess loose drive inventory in the channel, right?
So the impact though the TAM and your shipment ability has been noticeable in the model. So at the same time then how do you feel about your visibility today in the business as you look out over the next 6 to 12 months? And how much is that factoring in? How you're resizing the cost structure, so that we can be resilient? And then I have a follow-up.
I don't know that our visibility has changed a whole lot just because the only thing we're directly visible on is what's in our hubs or what's in distribution. And both of those items have been managed, I would say, very well probably near all time lows. And what's in house, I mean, I think one of the things, big differentiator between Seagate and the competition is just how lean our inventory is on the balance sheet, a third less, which we think is reflection of our operational excellence, but we also think it's a smart thing in terms of how we're using our dollars and what it means in terms of responding to any decrease or increase in demand. I think that not necessarily that the visibility has changed much with the PC OEMs as much as their public disclosure of dealing with inventory gives us some comfort that they've started to address or have addressed the issues that they've seen, whether it's in Europe or Asia. I think in the U.
S, the inventory probably wasn't that bad for them. I think in particular the currency issues caused problems in Europe and of course with the kind of relative strength in the euro, I think a lot of that stuff starting to move now. And I think in addition to that, I think those companies actually depending on who they were and where they were exposed have kind of started the tactically issue and they're probably halfway to 100% done I think with adjusting the inventory. So we feel better that while maybe the last quarter or 2 was depressed by a few 1000000 units of adjusting to inventory that you start to trend back towards a higher unit TAM that's reflective of maybe a more kind of consistent PC demand, but we'll see.
And then Phil, just wanted to talk a little bit about your bigger picture vision to the extent you can comment on where the systems business should be for a company like a Seagate. And last year obviously this time we talked about a different kind
of revenue
number. But even just percentage of business, what do you think would be considered a success for the systems business 2 or 3 years out?
Well, I think that really this is a journey for us. I think that if we look at what happens, I think that more and more of our customers are coming to us every day asking, all right, can you deliver more solutions? And I think there's going to be more opportunity for us. So I do expect that we will continue to grow at a faster rate than the overall core business simply as we start to grow up that value stack. So I mean, I look at it come in and I want to build a meaningful profitable business that's contributing to the company.
And that's going to take some time. That doesn't happen overnight. And Seagate is very good at playing the long game. And that's what we're doing right now. So we're acquiring capabilities, customers, IP and going to continue to work with our traditional customers and frankly new customers to continue to build that.
And so that's going to take some time, but we're putting the pieces in together one step at a time to get there.
I think that it's a sub-one billion dollars business today, but growing well. And then as you can see from the products, the positioning of the products is actually quite strong right now, especially in, for example, the PCI performance accelerator card business is growing very well from us. The product is particularly competitive. It's great performance. It's not super high revenue, but sticky revenue.
When you deploy those cards, it's not like they call up and say, send it to me in the mail. You've got a lot of engineering that works with the IT departments to actually adjust the workloads to the cards. The new SSD that we just announced is certainly competitive. And the opportunity in general in terms of the exposure to flash in the data centers is high and we like our positioning. The challenge would be, yes, but you don't control the media cost.
And I think in the value add segments, I'm not sure that's completely relevant, at least not today. And the reason I say that is because there's a lot of value add that goes way beyond the media. It has to do with how you manage the media and how you manage the media both at the device level, the subsystem and the system level against the workload and that's what we're particularly good at. And now we actually control the ASIC strategy across all three of those areas as well as the ability to optimize each device or individual devices. And what you trade for that is we have a controller technology that's flash agnostic.
And if you're a flash company, the problem you have is you do have a cost advantage if you're leading. And there's one leader and that leader kind of rotates. And probably the biggest mistake Seagate made 3 or 4 years ago is when we tied up so strongly with Samsung on the belief that they would obviously be the leader. It turns out in that next cycle, they weren't the leader. And we had basically tied all of our technology to them and we were just uncompetitive to in that instance it was either Toshiba or Micron depending on kind of which 6 month segment you were in.
So I actually feel that for the next several years maybe it's a 3 to 7 year statement that being flash agnostic is actually a nice position to be in, because as long as we have good strategic relationships with Best could either be a performance number or it could be a cost number or best could either be a performance number or it could be a cost number or a combination versus the flash provider has to use their flash. And with 5 players by definition 4 of them aren't competitive. So it's an interesting kind of transition. So for us in the meantime, we just say how do we make sure that we have access to the best flash and forge relationships with the leaders like we've done with Micron and that we'll pursue with others because like I said earlier, we're going to need a lot of flash. So it's an interesting thing.
But I think the answer to your question is that business that's sub $1,000,000,000 today should be a multi $1,000,000,000 business in a 5 year timeframe. So I mean it becomes significant to Seagate overall. I'll let the mic person control.
Hey, thanks a lot back here. It's Ananda Baruah at Breen Capital. Steve, thanks. In the context of the new opportunities that you guys talked about today for growth. What's your view on what zettabyte growth rates look like exabyte growth rates look like on a normalized basis?
And what might be some of the mechanisms if any that need to align to begin to catalyze some of these new opportunities? And I have a quick follow-up. Thanks.
I'll let Dave let me just do a little macro answer and then I know Dave said a lot about this. I think the thing about zettabyte growth is zettabyte growth in flight and there's zettabyte growth being stored. And I think those are 2 different numbers. Obviously, the second one is a subset of the first one. But in flight, it's getting massive because of 4 ks video going to 8 ks video.
And as long as again, as long as people can manage the user experience of latency, I don't see a slowdown in that. The potential for value whether or not it's entertainment value or whether or not it's true value in terms of analysis for retailer security or for financial services. There's lots of reasons why analyzing that data is important. So I think in terms of the zettabyte growth as it relates to what's going to be stored, we're still of the opinion that even given compression and all the things that people are trying to do to rationalize what can be stored, the industry is probably a full industry short by 2020. And that's with assumptions that say then there's a question what's the architecture and we've probably been conservative I.
E. We've been generous to what could go in Tier 0. We continue to say 3% to 5%. Reality is Tier 0 is actually 1%, 0.5% something like that. It's not anywhere near because again the cost of that as a storage media.
And most of that flash system gets sucked up into the on the CPU side we think. So I think we have a big challenge still as an industry. That's why I still believe there's shortages everywhere and whether or not it's capital or R and D related. So the real issue will be who gets that storage. And I think that's an interesting question.
Dave, if you have any
The only thing I'd add is that I made reference to this in my talk that the drives are being used more and more. So the average drive that we have in the field today mission critical, business critical, we shipped a terabyte 4 years ago what we terabyte per drive that's in some of those spaces. And what we see is about 50 terabytes of workload median right now in the field and it's growing. So if you think about the amount of data that's going back down into back and forth to the device, it's becoming larger. So the drives are not becoming colder.
They're becoming warmer. But there will be applications so the tails start to spread out. Some of those medium drives will be worked even harder. And then there may be cold applications as well. We may see just even tiers within the drive themselves.
So translating that into zettabytes or exabytes is really, really tough math I think.
I got it. That's really helpful. And I guess the shortage comment, Steve, would suggest that margins would have another opportunity
to continue
to go up longer term as well?
I mean, I think the capital model is not consistent with delivering 5 to 6 zettabytes by 2020. And 5 to 6 zettabytes is a fraction of what's in flight. It's a fraction with what most industry experts say will be needed for storage. I think EMC still talks about 30 zettabytes. China still talks about needing 30 zettabytes just for their surveillance systems inside of China.
So yes, we're a long way away. And the question is when do you have the margin profile to increase your capital budgets? We wouldn't increase the capital budgets without the margin. Let me move to some new people. Here.
Thanks. Steve Fox with Cross Research. Could you maybe expand on your near term expectations for the adjacencies this year? What exactly is going to be leading to growth? Where are you getting maybe some synergies?
And what levers could you pull to maybe accelerate it over the next 12 months? And then I had a quick follow-up.
Adjacencies meaning the products Bill has? Yeah.
I think the if we look some of the new areas we have, we just announced kind of an agreement with HP and Spectrum Scale on the high performance computing side. Really most of our high performance computing space has been in the government space today. So expanding that into enterprise, we talked about oil and gas. Our flash products, particularly in the PCIA accelerator sides are going fast. And on the SaaS side, we frankly are in we have de minimis revenue there from a SaaS SSD side.
So those are some of the elements there. Certainly, I would look to the pending acquisition of Dot Hill to continue to expand our customer footprint there and the opportunity to get expand our product portfolio that we sell into our customers from that side as well. So those are the products that I would look to drive the growth as we look next year over this year.
And then just a quick clarification on the dividend. So the 10% goal, it seems like that could be easily achieved if you were to grow your earnings by 10%. Is that a minimum goal? Could the dividend be increased more than 10% over the next quarter?
Well, we've tended to do more than 10%. Historically, we've done more. But for those of you who love to do cap rates, you can put in 10%. Yeah.
Thanks for taking the question. 2 here. 1st, could you talk about the visibility from the cloud customers, especially given that more and more data is moving there and the test times is increasing. What do you think can be done to improve the visibility of project demand better?
Visibility from a demand perspective? Yes. Again, it's fairly concentrated in terms of 20 big companies and within those 20 companies they kind of buy in very bulky patterns. We have pretty good dialogues with them. And because your dialogues are typically at the BU unit head or at the CIO head or the infrastructure head versus the procurement side.
It's probably better visibility than you would get when you were talking to procurement person at an OEM that felt that they had plenty of access to devices. I think the advantage of the cloud customers they understand the specificity of the device and the complexity of the device. So they kind of get it now that they can't be wrong by 1,000,000 units of upside in a quarter because there's no way to flex that. But the OEMs, of course, grew up in a habit of flexing 1,000,000 units a week and we would respond because it was mostly client. So I think in that sense you have a better dialogue.
That being said, even those companies I think are they surprise themselves in terms of either being wrong about demand on the upside or the downside for the need to store or compute. They're probably not as wrong on their ability to deploy. I think they've got rigorous schedules of how they take down time to deployment and they usually aren't too wrong on that. I think they get forced sometimes into higher utilization rates based on financial metrics, at which point they just turn on quality of service. And if it's a retail application, you can get away with that, right?
Because if it takes 6 seconds to load a video instead of 2, while your user may not like it, you probably don't care all that much as the company as long as you kind of ultimately make them happy with a faster search result. So I don't know. It's not perfect. It's probably more genuine and transparent than with the OEMs in part because the products are more specific. And that's again, we talked about this last year before.
The thing about the cloud providers is which again is counter to the one big cloud in the sky thing is all those architectures are really different by those 15 or 20 companies as a function of the workloads and therefore so are the requests that they have for us in terms of what those devices look like. And since they're trying to achieve a total cost of ownership, marginally they don't they're not as interested can they save $5 on a disk drive because there's 2 bidders as much as what does it mean in terms of their overall cost of ownership. So I think it's a better customer for us long term. I think the real issue is when does the next 100 customers come into place that are buying
Then just as a follow-up with the Dortil acquisition, do you still think OpEx could be significantly lower than $500,000,000 per quarter by second half?
That would be a goal, but we're not going to choke that asset to because that asset today is already accretive. So we wouldn't look to disrepair.
There's not a ton of OpEx. Right. Yes. I don't think there's I don't think that's going to be an issue.
Hello. Is it okay, okay. Somewhat of a follow-up on the previous question, 2 part question. So regarding the hyperscalers, I think in the June quarter, it was demand was a little lighter than you expected and you thought that it would come back in the December quarter. So again, maybe a question around the visibility there.
And you have some high profile hyperscalers that have kind of got found some new new found CapEx discipline, I think shareholder enforced. And then also about the related to that and the product mix, I think you've talked about you need the 4s and 6s to grow to 6s and 8s this year to meet that mix that's going to help you with your plan. So I'm sure some of those disproportionately are going to the hyperscalers. So can you talk about how you're off that question is how are you off this year? What started you off this year on transitioning clients to the 6s and 8s?
And do you think the hyperscalers have you have visibility on them coming back in the December quarter?
Well, I'll let Dave talk about 4s to 6s to 8s and then I'll take the second one, which you go first.
So there may be maybe if I understood your question right, so there may be a little hesitancy people going from 4s to 6s to 8s really quickly. I think they do a TCO calculation and look and say if the 8 is coming right here, do I move to a 6 or do I move to a 4 off of a 2? And it's complex because in some cases they're doing new build. But in a lot of cases they're replenishing what they already have. And so those are and there's failure rates on the old drive like Steve talked about earlier.
There's failure rates on the new drives. They have to wait for maturity on the new drives. So it's a fairly complex TCO calculation. But we think that over time that's so compelling to get to those higher capacities that they will go hard to the A's. What is it in TCO?
Well, you think about it, you have to install the back infrastructure, whether it's power you're paying for or host plus adapters that you have to pay for network connectivity you have to pay for to manage that slot all the real estate. That's all fixed cost relatively. And then do you would you rather put in 2 terabytes or 4 terabytes or 8 terabytes? So that's the way the calculation goes. But underneath the controllables that we have, there's a bunch of other metrics as well that may factor into their decision as to when to go.
Oh, yes. And then they'll continue to take the highest capacity drive that they can. Then it's just a question of where they are versus if you're on Tuesday and you go to 6, that's a triple. If you're on 4s, you go to 6 versus waiting for 8, right? A third versus a half.
So it's 1st versus a double. So the second your first question was this whole better discipline because of different management, part of that's marketing, I think. I think these companies are always managing their capital deployment rates, utilization rates and their forecast of demand. And maybe there's discipline on top of that at the margin. But I don't think that's what really changed some of the big purchasers.
I think their view of the world starting I'm just trying to think of one meeting that we had. When was that meeting? December. Was it oh, it was in December. Wow.
Yes, I think probably so probably by spring, people started having a different view of 2015 than they did in January. And that's why I keep saying this year reminds me so much of 2010 where you kind of went into the year thinking it was great and then by May it was like, this is going to be a hard year again. Like we all forget how great 2014 was. And so I just think that that's probably what adjusted more. I don't think it has anything to do with data growth.
It has to do with again, how do you manage your budgets temporarily and because so many of those companies can just change quality of service slightly to get through quarters, but then it comes. And this happened to us 2 years ago when it was flatter than we thought. We said, but it will accelerate and then boom it accelerated for 3 quarters in a row. And it seems to kind of be 9 month cycles, which is why we have confidence that in the December, March, June timeframe, we're likely to see an uptick. We wouldn't start getting signals on that until probably another month or so, but we'll see.
Yes, sure. Whatever, either way.
Thanks. Two questions for me. Amit Arianani, RBC Capital. Steve, you talked about how being flash agnostic is a good place to be. Would that suggest that you would never be interested or look to acquire a flash asset in its entirety?
And there's been some market talk that you may be interested in buying something in the flash side, so.
Yes. Well, we wouldn't comment on that specific rumor, of course, Notwithstanding at 21 PE stock that's probably behind in cost and behind in technology doesn't seem to be a great target. But that was not a comment about our interest. We have no comment. I don't know.
I mean, I think there is it's not a bad strategic argument to say that horizontal consolidation makes sense in any industry. And particularly in technology when value is attriting from the integrator to the either the core technology owner or to the service provider. And the integrators, even if they decided the right play was to become a core technology player, they can't do it. They can't do it 1, because they can't reacquire the technology and 2, because Wall Street won't let them. So they're forced to go up.
And in going up, they shed whatever remaining technology they have and they seed territory that quote is more at the technology component level. So if you're a technology component company, you have two choices. You can say I can kind of permission is being granted for me to move up slightly, take it, which is what we're doing. But it also says that if I have more area under the curve and do that, I want even better. So there's it makes sense kind of, but then there's all sorts of more interesting transactional and strategic and technical issues that I think have to be covered.
So it's not the craziest thing in the world. And if that happens, it won't shock me. For us, we feel that given the cycle we're in right now and that there's a lot of advantage to doing what we're doing with our own portfolio. And that that's probably on a risk reward basis the best thing for us to do to basically grow revenue at accretive margins and continue to increase the importance of us as a technology provider given the shift to component devices being overlaid with kind of open software. So that's where we're at right now.
Got it. And then we should do one more here and then we got to wrap it up. Sorry. There is a gentleman here. Yes.
Thanks. Wamsi Mohan, Bank of America Merrill Lynch. Steve, you mentioned earlier that if client does not take off, you can consider reducing footprint. However you choose to define client, how and when would you think about how you would what metrics you would look at to determine that client is not taking off? And then I have a follow-up.
Well, the metrics I think would be would basically be units and capacity per unit. And the time I think is 18 to 36 months. I think for the next 18 to 24 months, we're not going to expect that there's some resurgence of client other than maybe modest resurgence based on the existing class of clients. But I think you're talking about when does that when does the new client arrive. I think you have to give it time to breathe, especially if you believe it's 3 d printing or robotics as a part to a more fully functionally existing client.
I think we're on the front end of that and we want to wait. I mean my worst fear would be we'd cut all that stuff and then people are selling 100,000,000 3 d printers with 4 terabytes and we can't service it or then 6 terabytes. So I think for us it's more how do those markets develop? Are we in dialogue with the leaders? Yes, we are.
Does it seem like they're getting traction? Do we see the value add? And if we are then and if we see none of that happening at some point, if the rest of the model is being challenged then you just have to decide you're going to make cuts and rebuild later.
Thanks. As a follow-up, could I just have a follow-up?
Yes. Kate, we'll allow him a follow-up.
Thank you. Thanks, Kate. So architecturally as enterprises move more to the cloud, how should we think about the multiple copies of data that are sitting within enterprise environments? Do you think that changes? And in cloud architectures, part of the premise is just more efficiency that can be gained in those architectures.
So as data moves to the cloud, do you think that the capacity utilization in the cloud will be a deflationary factor for exabyte growth? Thanks.
Well, yes, but when you kind of still include all that stuff, you still get to the big delta. If you don't have that, then the delta is ridiculously massive. So you can include that in compression. I mean, reducing numbers of copies, reducing the amount of data you have to solve in terms of like digitizing analog data, compressing whatever it is, quote more efficiency, less copies per virtual desktop. That all has to happen and we're still kind of an industry short.
And do I believe it's going to happen? I don't know. I mean, the reality is it doesn't matter if it's hybrid cloud or a big cloud. The number of copies that they keep per data set is a lot, because they can't lose data either. The question isn't so much how many copies they keep, it's basically how accessible is it.
And if it's less important like it's a video of your kid at a birthday party, then yes, maybe they don't keep the copy as close to response as it does if it's a financial transaction, if you happen to put a financial transaction in the cloud. So I think most of those elements people have thought about in terms of still what needs to be stored and we're just we're still a long way away from making the math work. But are those systems inherently more efficient? I don't know that they're that more efficient than like what Morgan Stanley or JPMorgan can be able to do. I mean they have really smart people in their IT departments too.
So I think they're all we'll keep multiple copies. We'll have data centers all over the world. And the question really is what's the response time you want as a user and that's going to be a function of the application. If it's a really important application, you want no latency. And if it's not, you can have more latency and that just means how far away you put it.
It doesn't mean I think it influences number of copies. So I think most stuff is still copied 3 or 4 times. And I don't see a big change to that. The leverage you get if it's a virtual desktop in a corporation, you don't have it on every desktop, right? And then the question is how do you manage the bandwidth.
Okay, good. Thanks everyone. Appreciate it. And I guess we'll see you this time next year.
Oh, and
we have in the back we have more demo stuff and then we have treats receptions. Yeah. Great. Thanks. Thank you.