All right. Good morning, everyone. Thank you all for coming. It's always a pleasure to see so many familiar and friendly faces. It's great to see you all again. As for us, I know that many of you lost a good friend last week. I just want to assure you that while we're all going to miss him, at Micron Technology, we got way too many good things going on to dwell on the past. As a company, we're going to continue driving forward. We know that's the best way to honor Steve's legacy. Jumping right in, we've got what we hope is a pretty crisp presentation for you today. We want to cover a number of key topics. We'll lead off with Scott DeBoer, who many of you may not have met, giving an update on our technology roadmap and path forward.
Scott's a longtime Micron guy, actually, 17 years now. Joined us from Iowa State. When I became a COO from being CTO in 2006, I told the internal team, hey, my half-life in terms of a technology leader for you guys is going to be about two years. I'm six years beyond that now. Scott knows about 8x as much about technology as I do. You guys will enjoy hearing from him personally today instead of having to hear from me about stuff I don't know as much about as I either think I do or used to. We've got all the business unit leaders here. They're going to give you an update on our product portfolio and all the market segments we're trying to address to drive value for the company.
We'll start off with Tom Eby, who joined us from Spansion a year, year and a half ago. Tom, with a wealth of knowledge about all the embedded markets. Mario Licciardello, with a long, long history at ST, came to us via the Numonyx acquisition a couple of years ago, a year and a half ago. We'll give you an update on what's going on in the wireless space. We'll have Brian Shirley. I think Brian has actually been at Micron as long as I have, believe it or not. He's younger and better looking. He started as an intern. He was still going to school after I joined Micron out of grad school. Brian, with a long, long history at Micron too, has developed a great DRAM portfolio he wants to tell you about. Finally, on the business unit side, we'll have Glen Hawk up here.
Glen, with a long history at Intel before joining Numonyx, has done just a super job with our NAND solutions group. He is going to update you on what's going on in all the market segments he sells into and all the value-added differentiated products he is building. Mark Adams, our new President, has come up and tied all that together, how we are taking the product to market, how we are making sure we are delivering value not only to our customers but also to Micron Technology by addressing all the right channels and tuning the portfolio for the right customers. Ron Foster will come up.
Ron, with, by the way, a strong, strong history at HP and Applied Materials, other companies, to JDS Uniphase on the financial side, will give you an update on where we sit financially and the strength of our balance sheet and how that enables us some flexibility in these times of opportunity. Finally, I will come up and wrap up with a little bit about how we see the industry and what we think our opportunities are moving forward. Without further ado, I am going to turn it over to Scott DeBoer. Scott.
Thanks, Mark. I'm going to start out with an update on one of our key investments over the last year for our R&D infrastructure. Right about a year ago this time, we started a project that we believe is really a key component to our technology direction. There's supposed to be a video here. There we go. This is from groundbreaking in February last year through what the construction of a fab looks like. It's kind of an interesting perspective. You also get to see the seasons in Boise. I'm watching that. Some of the just pointers around this building, we started last February, as I mentioned before. We installed the first tool into the clean room in December. This was a pretty quick build. A lot of great work went into this.
This building almost increases our clean room space from an R&D point of view by about 50% in Boise. This is a pretty big project for us. It was about a $100 million project. With this technology, or with this building, which is set up for 300 mm and 450 mm, although we're not pushing forward with 450 mm right now, it really sets us up for the future in Boise. This building is connected to the existing R&D fab in Boise and really sets it up as a leading memory technology development center in the world. Overall, we have globally a diverse R&D team with several locations. This is definitely our core. It's our 300 mm center. We do, in this site now, develop the technology roadmap for our NAND, our DRAM, PCM, our NOR, and then all the emerging memory technology.
I'll talk a little bit about the roadmaps on each of those on the slides going forward. First, jumping into our NAND technology position and just a few key points on this slide. There's a lot going on on this slide. We've been in a strong leadership position with our NAND technology over the last several years. The blue dots on this really just show where we're introducing NAND technology nodes. There's a couple of points off of this. One is really around our 128 GB NAND technology, which is the industry's only single chip of that density. Glen's going to talk later about the kinds of things that that enables. I'm going to talk a little bit more about the quality of that and some of the technology behind what's allowed us to make that single 128 GB chip.
One of the other key points on this is really around our it's getting difficult to define the NAND chip by a single number when we put out 20 nm and our competition puts out 19 nm. One of the important things to recognize on technology now is the other dimension that actually kind of defines the size, the cell size. Our technology is 20 x 20, while our competition is 19 x 23 or 24. You wind up actually with a bigger die, but maybe a better marketing number. This technology is enabled by some pretty unique stuff. I'm going to talk about that on the next slide. There are two important points off to the right on this graph also, which are the widely discussed end of scaling on NAND. We think we have technology that actually puts us in a very strong position for planar NAND going forward.
We think we'll continue to be in a leadership position on planar NAND. At some point in the future, the transition to a vertical NAND type technology, which I'll also have a slide on here in a second, will come. Depending on a lot of things on the market mix for Glen's product and other things, we are positioned to make a decision on when we bring planar NAND in or, excuse me, vertical NAND in relative to planar NAND. This is a technology slide. The top of this is really our competition in how they build NAND technology. You can see up to about the 25 nm point or the mid 20s, everyone's NAND technology kind of looked the same from a cell point of view.
For our 20 nm technology, we've actually come up with a very innovative and, I think, revolutionary technology advancement that changes how the NAND cell is built. This inherently gives us an advantage on scaling going forward and also on performance when we look at the kind of applications that Glen needs and the performance required. We think this new technology really plays well into that. On vertical NAND technology, we are aggressively investing and working on this technology. We're strong believers in vertical NAND technology in the future. The timing of this technology really depends on a balance of things across the product mix and some of the things Glen's going to talk about in the future. We do believe we will be positioned with vertical NAND technology in a leadership position again.
This has the potential to really increase the density and reduce the cost per gigabit in the future. There's a lot of challenges around vertical NAND technology. There's a lot of things written about it still. Right now, there's clearly some, even from these pictures, you can tell there's some pretty aggressive aspect ratio and material challenges that are in front of it. It does have the benefit of reducing lithography requirements by some amount. On the DRAM technology side, we've really put a lot of focus into this over the last few years to make sure our DRAM technology is positioned to deliver the kinds of performance that Brian and the DRAM business unit are going to, or Brian's going to talk about later today, that drive his business right now. We're really on a path right now to deliver three shrinks in three years.
Our technology path here is a little bit different than our competition again. Our 30 nm technology, first off, is running in three fabs right now. It's in a strong position with really best-in-class power performance relative to the competition for the server market. The product right now, the 30 nm, is about 15% of our output or our wafer starts right now. Towards the end of the year, the second half at least of this year, we'll be up to about 50% of our wafers running on 30 nm technology. That really depends on the mix and some of the stuff Brian's going to be talking about later. Our path beyond 30 nm is, like I mentioned before, a little bit different. One of the things that is critical to our business is how quickly we can actually have high-quality bits to meet the market mix that we drive.
One decision that we've made going forward is to do a really design-optimized die in between our 30 nm technology and our 20 nm technology. A lot of our competition here have done really kind of not very aggressive shrinks there anyway, but real process technology shrinks that drive, again, a period of time where you have kind of low-quality bits coming out. We believe by going down this intermediate step of not doing a full process conversion but doing a design optimization and getting a good shrink out of it still, we'll be able to go into high-quality markets much quicker with that die, even this year, and then follow it with an aggressive 70%-ish kind of shrink on 20 nm technology. The 20 nm technology right now is running in Boise in the R&D fab. That's our lead product that's running right now.
The 20 nm technology for us is targeted for manufacturing introduction towards the end of this year. It'll be a big technology change and a very small die. We're very confident that through this timeline, we have the right DRAM technology to meet up with the business that we're targeting. Another DRAM-related topic, we've had a lot of publicity and a lot of good feedback on our Hybrid Memory Cube technology. Brian's going to talk in quite a bit of detail about that later, about the excitement there, the products that that drives. On this slide, I just want to talk a little bit about the process technology that goes behind this. The enablers on the process side for the Memory Cube are really things that we've been working on for five or 10 years actually and are really coming into play at this point.
We have a lot of background technology on through- silicon vias, on stress management, on packaging technology, assembly, different kinds of thermal management solutions that have all come together to enable this product. Right now, we have manufacturing capability for the Memory Cube for the TSV technology in place to enable the startup of the ramp. We have the technology rolling out into higher volume manufacturing and that ongoing over the next year. Brian will talk a lot more about the markets and the exact products that this is going towards. Another exciting thing that we've spent a lot of time on over the last year is our phase change technology roadmap. Just over this past year, we've made a tremendous amount of product progress really getting this to a mature state from a yield and a reliability point of view.
This year is really where we're rolling out the technology into manufacturing. Our 45 nm technology is what's rolling out into manufacturing right now. We'll be building up the infrastructure and producing wafers in volume over the second half of this year especially. At the same time, we're now focused on our 20 nm phase change roadmap. Phase change technology actually, in our view, has extendability beyond 20 nm. We're focused currently on the planning stage of the 20 nm. Beyond that, phase change can either be in this kind of a form where it's a planar kind of technology, but phase change also potentially fits into a vertical technology in the future. This is a slide, and I think we've shown different kinds of versions of this slide a couple of different times in the past.
On this summary of emerging memory technology, really, I'm outlining the different kinds of technology that we're looking at. We've covered a pretty large space internally. I'll talk a little bit more about some of our focus projects right now. We think that there's some winners coming up to the surface over the next few years. Really, the winners highlighted on this graph are more around for a specific kind of volume application that would be of interest to Micron . For example, if you look at an MRAM technology, which has an awful lot of red up there, it's more a statement around Micron 's interest in a high-volume application as compared to clearly some people have lower volume MRAM products out in the market now. We're looking at maybe with a little different lens than some people.
DRAM and NAND, we think we have a long runway on both of them right now. The new technology that comes in really comes in with different kinds of specifications relative to DRAM and NAND rather than a pure replacement. Our focus internally right now is across a variety of different kinds of new technology. A couple of them that I'll highlight off of this are CBRAM, which is really a resistive RAM memory technology. There are a lot of different kinds of materials that you can make a resistive RAM to some degree out of. We believe that we've gone through those and come up with a material set that is the best option. We're currently working on this CBRAM technology with Sony in a joint development project. We have people from Sony actually in Boise working with us on that technology right now.
We think that that has a really good path if it fits into the market mix that we want for the company of actually being a product within the next two to three years. Floating body memory, we have a project on that right now. That effectively is a technology that looks like a DRAM but runs without a capacitor, which is one of the scaling limiters on the DRAM. The last one on this chart is FinTorque technology. We did have an announcement this summer about a research collaboration with DSI, which is a Singapore-based research organization. We're collaborating with them. We also have some joint development work going on on FinTorque with IBM that we just recently started. OK, so in summary, I think we're making the right kind of investment.
We're targeting keeping our product base going right now with strong roadmaps on the things that we've already got in volume production. We have some interesting new technology coming up on phase change memory and other emerging kinds of memory. I didn't talk about it in the slide deck, but we are converting our NOR technology over to 300 mm right now for rollout later. Tom is going to talk a little bit more about that later. I think we're really well positioned with the right scope and the right partners on our emerging memory technology and position for the future. OK.
Thank you. Thank you very much. On the NAND flash, you talked about how you have a path forward going vertical. You also said you still had a path on planar. Can you help clarify where do you think in terms of which process geometry do you think NAND in its current form on planar runs out of steam? Also, as a second question, can you elaborate a little bit more on phase change and what benefits that has and which devices you think it'll replace in the future and when? Thank you.
Sure. First, on the NAND roadmap, we're very confident about a mid-teen kind of nanometer technology on NAND being capable. We believe NAND in this context is very much about what product performance you can get rather than whether you can actually build some kind of a product at one of these nodes. When we look at our planar technology roadmap, we feel we have a strong advantage because of this new cell technology. We still look at what kind of performance we can provide Glen for the kinds of applications he wants to go after. It's a two-part judgment on whether we can hit his product space for the kinds of high-performance products more so than just can we build a 15 nm, can we build a 12 nm. We're confident we can build the 15 nm, 16 nm kind of technology.
We're looking hard right now at what happens beyond that based on the technology capability and the business requirements. On phase change, there's actually multiple kinds of applications. Mario and Glen both probably can talk about some of the places that they're looking at bringing PCM in. Certainly, our focus is on wireless and storage with our phase change technology. Both of them probably would be able to elaborate on how they're looking at phase change.
On the vertical technology, how do you start working with the controller? How does that change the controller for NAND flash?
There are challenges for controller technology that are very similar between vertical and planar. I'm going to deflect that question a little bit to Glen to go into more detail on the controller technology. I think it's a very important piece of how you manage the raw performance of the die. We're actually looking at vertical technology as having some pretty big step up in raw performance relative to our end-of-life planar technology. Glen may come back and say it's extremely difficult. I'm going to say that I think actually it's going to give him a little bit of a break.
On the 3D technology, where do you stand in terms of IP? I mean, do you feel that you have all the blocks there? Or would you need to license it? Some of your competitors are developing 3D as well. Do you think you have everything you need?
Yeah, I think we're very confident in our IP position across all the products that we're talking about right now, and don't have any concerns relative to that.
Maybe just to follow up on that. I mean, do you feel because Micron' s done a very good job of not having to pay royalties to anyone for any process technology, by the answer to your last question, you feel confident that Micron will continue not having to pay royalties to anyone else for these technologies?
I think I don't want to go too deep into our relationships with different people or our licensing relationships. We have done a good job of making sure we internally have a strong IP position. I think we're confident in that IP position going forward. It's definitely a focus of our company. If you look at some of the public statistics on IP, we're generally in a pretty good spot there. I'm fairly confident we're in good shape going forward there.
Shift over to DRAM real quick. You mentioned, I think, 50% of wafers by year end would be on 30 nm, if I heard that correctly.
Yeah, sometimes the second half. It actually can be earlier than that. It's very dependent on the mix and the customer demand that Brian comes forward with. The technology is ready for the ramp; it's just a market timing question there.
Is that just Micron fabs, or is that Micron's partners as well?
That's all the wafers that come to Micron from wherever they come from.
Great. Thank you.
Scott, you referenced briefly 450 mm in your presentation. Curious as to when you think the industry makes that jump. At 300 mm, Micron didn't take a leadership position. You kind of took a fast-follow position. Was that a good decision in hindsight? As you think about the 450 mm strategy, are you going to be more a first mover or not?
We're definitely not going to be a first mover on 450 mm. There's momentum from some, obviously from some big players on 450 mm. It didn't work out very well for the first movers on the memory side on 300 mm. I think when we look back on our 300 mm strategy, we were pretty successful with it. I think it's a little bit farther out than what most people would say right now. We'll be closely watching that. I think we have plenty of time to react as the story evolves on 450 mm. We're just not going to be on the front end of it.
Do you think the Hybrid Memory Cube makes it into a product, actual product like a cell phone or something like that?
I think I'll let Brian follow up on that one. He's the guy who's got to make the product.
Could you elaborate the 30 nm shrink you talked about, the DRAM, and kind of what cost per bit advantage you think it can provide?
Sure. That shrink is going to be around 15% cost reduction. The main thing, again, it's a moving story on the cost reduction versus the performance. If you look across the industry right now, people have come out with more aggressive shrinks, but they're selling them into markets that are not favorable. It's really a combination of the shrink plus the capability of the die at the time, which we think is going to provide us an advantage on that time as a bridge to the 20 nm node.
Good morning. I'm going to spend a few minutes to give a quick update on the market opportunities that we see in embedded, how we see those unfolding both from a product/technology as well as a segment perspective. Then I'm going to spend a couple of minutes just talking about our value proposition and how we feel how we win versus actually a couple of different classes of competitors that we go up against in the embedded market. I'm going to walk through a handful of examples as we go through a number of the segments, talk about some specific drivers of some fairly explosive growth in memory as well as kind of the aspects of our value proposition that help us win in those particular examples.
If we take a look at our view of the TAM that the Embedded Solutions Group goes after, this is a 2011 view by product, by segment, and by customer, a bit under $6.5 billion in total. When we look at it from a product or a technology point of view, the smallest segment last year, NAND, is actually the fastest growing. We see it a bit over 15% CAGR between now and 2015, a lot of that driven by managed NAND growth in a number of segments. I'll talk about that. Number two in terms of growth is in DRAM, and that's a bit under 10% growth. Finally, in NOR, it's actually essentially a flat market as we look out over the horizon, although within that, parallel is actually dropping pretty quickly, almost double-digit percentage, while SPI is growing at about 10%.
We actually see a good opportunity with our SPI portfolio, and I'll talk about that as well. I'm going to talk about the segments in the next slide. The final comment I want to make is talking about the customer diversity. Mark will talk about this a bit more later as well. In a number of the segments, it really is an indirect play. We've got thousands of customers that we've got to get to cost effectively. There are a number of things we're doing to optimize the channel and some very specific programs that we've got in place to effectively penetrate customers 101 through 3,000, I call them. Let's try that again. OK, there we go. That looks better. This is taking a look at the same opportunity showing growth out through 2015, breaking it out by segment.
Generally, when you look from the segments at the bottom and you move towards the top, the requirements in terms of product features, in terms of quality, logistics, longevity, and other programs become more demanding as you move up that list. Therefore, they're generally higher value-added segments the further up you go. Overall, total market opportunity, about an 8% growth that we see out through 2015. The bulk of the segments are actually at or above that growth rate. Networking is particularly dominated in terms of what we go after by parallel NOR right now. Given the drop in size of that market, that keeps a little bit of a cap on it. Some of the other technologies that Brian will talk about, though, represent very significant growth opportunities there.
Finally, while we think we have growth opportunities, shared growth opportunities in all of the segments, in particular in industrial, medical, and multi-market from a channel play perspective and in PS and server, where it's really an SPI play, we think we have some good shared growth opportunities as well. Now let's talk about the value proposition. I really look at four things: technology leadership that Scott has just talked about, certainly the manufacturing scale and cost benefits that we have, the breadth of the product portfolio, and the embedded focus. When we look out at some of the broad multi-technology competitors, certainly they have good capabilities in technology and manufacturing and reasonably broad portfolios. They really don't bring the same level of embedded focus from an organizational perspective, from a channel perspective, services like automotive ISOTS. That really is how we compete against that class of competitor.
If you drop down to the larger single technology players, it's similar, although there, again, we have more of an advantage in the product portfolio. When you look at the more boutique players that generally tend to be single technology, tend to outsource some of their manufacturing, they are focused on the needs of the embedded market. However, in technology, in manufacturing scale, the in-house capability that we have, and certainly in the breadth of the portfolio, some very significant advantages. Across a range of competitors out there, that's really how Micron competes in this embedded space. OK, now let's start talking about just a few examples in a number of segments. In automotive, there's actually a lot of exciting growth out there. ADAS actually stands for Advanced Driver Automation and Safety.
This is the video and image capture and processing that allows things like lane drift recognition, collision avoidance, things like that. Displays, all of your analog displays are converting to an LCD, a graphics-driven display. That has good benefits for memory content, car body, backup cameras, and other things. Certainly, the most explosive growth is in the infotainment that's sitting in your dashboard. This is actually representative of a trend we're seeing in a couple of embedded segments where, because manufacturers, in this case, car manufacturers, want to capture the value that would otherwise go to a tablet computer sitting in the passenger seat or maybe in the back seat, they're evolving very quickly and taking a function that used to be single purpose embedded operating system, very modest memory. That's a 1995 old GuideStar for those of you with very good memories on the left.
It's turning into a general purpose connected apps-enabled architecture that has memory needs very much like a high-end tablet. We're seeing DRAM upwards of 2 GB - 4 GB. We're seeing managed NAND, generally EMMC, upwards of 64 GB. Obviously, very beneficial in terms of the memory opportunity and the importance of memory into this system. Certainly, the breadth of portfolio is important. The car guys love buying everything from one shop. I think beyond that, this is where longevity, the very automotive-specific requirements are important, and frankly, where the financial stability that Ron is going to talk about also plays an important role. These people are making decisions they need to live with for 10+ years. They need to make sure the company that's making those commitments is going to be around and owned by the same ownership as they are today.
A second area, which is very, very wide and not nearly as deep as the other segments, we refer to as IMM, Industrial, Medical, and Multi-Market. Just to give you a flavor of the amount of memory that's starting to show up in a lot of a bit more pedestrian products. On the lower left, learning thermostat, a couple of gigabits of NAND, half a gigabit of DRAM. The next two examples, a smart meter and a high-end washing machine that now have as much memory as your mid-range cell phone of 10 years ago. On the right, that happens to be a patient monitor with, again, as much memory as a current pretty enabled feature phone of today. To get to all of these opportunities, we need to have the right channels. We cannot afford to get at all of these directly.
This is where the ongoing optimization that we're doing with the channel, getting out to our reps and distributor networks, and building on the capabilities that came from both legacy Micron as well as Numonyx, I think is a very good opportunity for ongoing share gain here. The other program I think is probably very valuable is our longevity program. This is something we launched about nine months ago. It provides a 10-year fit, form, and function guarantee for a select subset of our memory product and really is aimed not only at the OEMs who have long-life requirements for their products in many of these segments, but also at our chipset partners who need to have who offer the same longevity as well as our distributors who don't want to have to deal with angry customers that are dealing with chains that they don't want to manage.
Finally, like automotive, this segment does help a great deal with optimizing the use of Micron's capital assets. Because of the longevity need, we can take existing capacity, generally highly depreciated capacity, don't have to put much incremental capital into it. We can leverage that with more stable, very high-value returns on those wafers. Again, good for return on Micron's assets. Next one I'll move to is consumer. This is also a case of pressure on a device, in this case, a TV that used to be a single-purpose embedded operating system, very modest memory needs. Again, because of pressure from smartphones and tablets, the move to a connected general-purpose apps-enabled product, again, with very explosive growth, therefore, in the memory demands. Not quite as much NAND in this case, but certainly very similar growth in DRAM.
This is probably the place where being on that leading edge of technology is more important. These do tend to turn over a little bit more quickly than the automotive and the IMM segments. Being at the leading edge is helpful there. Certainly, it's an area where we're seeing very good growth. The other one, I don't use an example, but just to touch on briefly, I mentioned the sort of customized managed NAND for gaming. We are seeing over time that opportunities that historically have been serviced by ROM going forward are being serviced by managed NAND solutions. We see that as a good growth opportunity in the consumer segment as well. The final segment I'll touch on is in personal systems and server. For embedded, this really means PCBIOS.
While the system growth rates are not real large in this area, certainly from a density driven by manageability and security, we see good growth in this. It's actually about a 12% growth rate that we see. This is a good share gain. Prior to our introducing our 65 nm SPI product last year, and we now have the broadest range of those from 8 Mb up to a gigabit, we really didn't have much of a play here. We're starting to see some good traction in design. We expect good growth, again, coming out of share gains in this area. It's one of the examples where not only our current technology, where we've been on 65 nm in NOR broadly for about three years, but also our roadmap we think is going to play out very well. Scott mentioned briefly about leading edge 45 nm NOR.
We're actually in production on that. We'll see first revenue next quarter. We're also well underway in terms of scaling that up to 300 mm. We can drop that into a high-volume fab. We're going to happen to use the Virginia fab. It's about a 70,000 wafer a month fab. When you look at our competitors, those that are building it in-house, if we built nothing but embedded NOR in that fab, it would be almost 200% of the market. Clearly, we're not going to do that. What it means is unless you have a multi-technology fab, you simply can't build 300 mm NOR at the scale that we can. Beyond that, from a technology node, we think we have an advantage. We're dropping it into a substantially depreciated fab.
From a depreciation cost perspective, all of those we think as we go forward with our scale up to 300 mm gives us a very good cost advantage and a good share gain opportunity across a number of segments, certainly including personal systems. To wrap it up, clearly, as I talked about on a segment-by-segment basis, it is a very high-value opportunity for Micron across the vast majority of those segments. It is largely because of longevity, a bit more stable, a little bit more predictable. It's still a memory market. It does extend the useful life of our asset. I've touched on some of the significant growth, both in some cases participating in TAM growth and others taking share that we see. In automotive, I didn't talk about amusement. That's Pachinko. Certainly, the gaming and SPI NOR across a range of strategies.
Of course, we think we're the only ones that have what we see as the four key ingredients for success in this market.
With that, I'd like to focus on Q&A.
Tom, have you seen specific examples where having the NOR portfolio has really drawn in additional purchases of DRAM and NAND that maybe we in the investment community, when you bought Numonyx, wouldn't have originally grasped?
Yeah, I think we've seen it work. We've seen it work both ways. We have situations where perhaps Micron had a very good presence primarily from DRAM. People are on an ongoing basis, again, to a greater extent, the further you get up that segment slide, so greater tendency in automotive, IMM, and networking and in some of the other technologies. We may have a great presence in DRAM, so they really want to consolidate. They're looking to it in NOR and in both SLC and managed NAND. Certainly, on the other side of things, we have some fairly longstanding, I'll call them legacy Numonyx, NOR, and to a lesser extent NAND relationships. They want to leverage what they're seeing. I think I would see that kind of in declining effect, automotive most, IMM fairly significantly, networking some.
You get down where they're more cost-effective, a little bit less of that effect. Certainly, at the top of that segment chart, fairly significant.
Question here, along those same lines. Maybe you can give us a sense for your attach rate or your penetration rate. Every time that you go into an embedded opportunity, what percentage of those times are you coming out with your customer using everything Micron, DRAM, NOR, NAND flash?
I mean, some of that depends on the application. There are significant opportunities, PCBIOS, some of the amusement opportunities, some of the game cartridges, which are just inherently single technology. I would say that at a high level in those top three segments, again, in automotive, IMM, and in networking. To be clear, in networking, Brian Shirley and I coordinate in terms of the DRAM side. I would say that in general, the vast majority of the opportunities there, people are trying to move towards how do we get all of that technology. I'm not saying we're being successful everywhere. Don't take it the wrong way. Certainly, that's an underlying desire that the customer has, that actually some of the chipset partners have, and that, of course, we have.
Please.
Tom, somewhat pursuant to the prior question, can you talk about the presence of MCP in the embedded spaces? When and how might you organize1:47
your supply chain to get more Micron inside of those packages?
Sure. Again, some people only use MCP to apply to different technologies in a stack, things like managed NAND, eMMC, or multi-chip packages. If we talk about what I'll call the multi-technology packages, a lot of that ends up being, I will call it, wireless-like solutions that end up getting repurposed because of particular needs, very often longevity needs. The fact is you need very high volume to justify the return or else the overhead to be developing the package, your risks of inventory. I would say there is some of that. When you move from a very high volume wireless to what are generally lower volumes, that economics doesn't work nearly as often. There's a little of it, but not a tremendous amount.
Yeah. OK.
I think in previous analyst days, embedded was highlighted as an opportunity for phase change. Is that still the case given that some of the markets can support a higher ASP than maybe some consumer-ish markets?
Yeah, no, I mean, if you look out, I think Scott's comments, I probably would tweak it slightly, which is to say that wireless and storage will see those opportunities first just because of generally the slower turnover pace that happens in embedded. When we look at where do we see code storage technology going, I think there's a crossover point at 45 nm where there are probably certain opportunities that may be better addressed by PCM because of performance reasons for the embedded space. Unlike NOR, as Scott showed, PCM's got a roadmap out to 2x nm . When you're looking at the higher density opportunities, clearly that's something that's going to hit in the embedded space. It's just, again, the wheels of the embedded chipset OEM, generally infrastructure, it just turns slower. It's absolutely part of the roadmap.
It's just going to hit wireless and storage first and be significant for us a little bit later. Very good. Thank you. Again, as a reminder, there'll be a Q&A at the end that we can follow up on as well. Thank you.
Good morning. My presentation today will be focused on highlighting the dynamics of the mobile market and trying to show all the trends that that market is undergoing and also highlighting what is our major strategy to be a winner in that market as well. I will also comment a little bit about PCM since I saw there is a kind of interest in that technology. This is the way we consider the market partitioning among the various applications. We go from the very high-end smartphone and tablets down to the low-cost and very low-cost feature phone and the ultra-low-cost terminals.
Of course, I will not go in all the details so that you can read, but just a few general comments that are possible to extract from this very crowded chart are that, number one, we see that there is an extreme variability of the memory requirements for the various applications. We go from, and this variability is both in terms of what technology are required and are used for the various applications, as well as what is the content of memory that are included into the terminal. Of course, this is driven by various considerations, one being clearly what type of positioning you want to achieve and what type of performance you want to obtain from your smartphone or feature phone.
The other one, which is immediately linked to that, is what type of price targets you are setting for your BOM because clearly the choice of the technology may be plus or minus influencing the cost of the BOM. The other consideration is that whatever the technology and all the product requirements that the market application are asking us, all those are within the Micron umbrella. We have all the technology that are needed no matter what application we are considering. We have all the products that are filling those applications. Therefore, in a certain sense, we can say that we are a kind of technology agnostic. We are not, of course, we're telling our views. We're trying to induce our customer to make the right choices also for our advantage. In reality, in the end, we will support the customer with its own choice.
We'll not force the customer to go NAND versus NOR or to go phase change versus NAND or NOR. This is the kind of methodology that we're adopting. We are, as I said, agnostic. We will, in the end, support the customer with whatever they choose. Of course, the requirements are ranging, as I was saying before, from NAND, which is the dominant technology in most of those segments. When I talk about NAND, I am talking about both SLC NAND and MLC NAND, managed NAND, and so on. Of course, in order to be very successful, we can exploit the synergy that we have in the company, thanks also to the cooperation that we are well established within the company with the other views. There are also some mature technologies that are still needed in certain areas.
For instance, the NOR technology, which is clearly in a declining phase, is still needed. There is a segment of applications in 2G, 2.5G that are still requiring NOR with the precaution that in order to keep the cost down, some of the suppliers are switching over from parallel NOR to serial NOR just to save a few cents in those applications where the cost is really a driving factor. In those segments, which is the medium low-end segment, is where we're focusing at least in the initial phase, the phase change technology. We have developed the 45 nm technology that Scott was highlighting before. Within that technology, the first product we have developed is a 1 gigabit memory with a DDR2 interface. The reason for that choice is that the DDR2 interface will simplify the bus access. It will induce some cost advantages at the application level.
There was a question before about what are the key advantages of phase change memory (PCM). Where do we position PCM, at least in wireless? There would be different positioning in other segments. First of all, if we want to compare PCM versus NOR in that area, which was before the domain of NOR, a key advantage is that it is a technology that we can continue to scale down. NOR has reached the limit by physics beyond which you cannot go. Physics is a real limitation for NOR, which is not a limitation for at least a few generations for PCM. Scalability is a key factor. Scalability means also cost along the way because you scale down, you get also saving at the die level and therefore cost advantages. There are many other advantages.
One, for instance, is that if you take NOR, the key limitation of NOR is that while NOR is faster in reading, it is unfortunately very slow in writing. When you write into the memory in the NOR, it takes a lot of time, seconds. PCM has not this limitation. It's fast in reading, but it's also fast in writing. You can overcome the limitation of when you have to write a lot of data, having a fast writing memory is very important. You know, one of the things in NOR is that in every writing cycle, you have first to erase and then write. The erasing cycle is actually what is taking most of the time. You don't have that in the PCM . You just go and write what you need to write in the PCM, so you can do much faster.
There are no charges that are moving in the PCM. While in the NOR, there are holes that are moving from one place to another, in PCM, there are no charges moving. There is just a change of state, but there is nothing that is moving within the memory. There are other advantages. For instance, power dissipation due to that factor is one advantage. There is a long list of advantages that you can take with respect to NOR, but also with respect to NAND in the low-density area, not MLC NAND, because NAND is the reverse. It's fast in writing, but it's slow in reading. You can compensate that again with the PCM. We believe that this is just the start for the phase change technology in the wireless market. We are already thinking about several other applications for the future.
What do we need in the various segments is simple. We need the leading edge technology in NAND, which we have. We need the leading edge technology in mobile RAM, which we have. We need NOR to maintain continuity with the past. We need phase change technology, which is one of our key aspects and access for the future. Overall, the mobile market is an important market, which is contributing in a significant part to the global memory market. We see that the trend is to continue to grow, even if at a pace which is below what was in the past. There is a significant growth. We see there that smartphone and tablets will drive the growth both in terms of units, but also in terms of, and clearly also in terms of dollar.
We see the rest of the market, which is gradually declining, not at the fast speed, but certainly going down and declining. Also, the landscape in terms of who is using our memory solution is changing. We were used to have a few major players in the market which were dominating the market. If you take this smartphone example, for instance, and we go a few years back, we see that we had basically three major manufacturers that were dominating the market. Clearly, there was one which was more dominant than the others. Overall, there were plus or minus three major players in that market. After a few years, what we see is that the situation is much more fragmented. We have many more players in the market. There are some emerging players, while the other ones have been losing share. We don't need to make the names.
You all know who are the names behind that picture. What is also very important is that within this fragmentation, there are new players coming. That is happening also in a geographic region which was not participating in that market, which is Asia-Pacific and more specifically China. There are several large OEMs which are starting to play a significant role in that smartphone market as well. I'm sure Mark Adams in his presentation will pass some messages on how we are matching that trend, increasing our support to OEMs in that geography. The other comment, which is clear evidence from that picture, is that it is very difficult for us to predict in this fragmented market who are the winners or who will be the winners or who will be the losers.
We are somehow obliged to work with a wide range of customers and hoping that some of them, I'm sure, will be the winners and we will grow with them. This is a really challenging situation. Fortunately, we have a strong relationship with all the majority of the players in that market. Clearly, it is an expensive effort to be in the situation of supporting everyone in the market. The end of the story is that the memory consumption is continuing to grow with a very, very high rate. Clearly, as we see in our day-by-day life, the access to internet and the internet traffic is the unique factor that is driving that continuous request for more memory in your terminal, whatever it is, is a smartphone, is a tablet, whatever. Just to conclude, I think wireless markets are the ones driving the growth in mobile applications.
Mobile applications drive more memory, and memory contributes to the global size of the market in that area. There is a dramatic dynamic in the market. The evolution of the market is very fast, but also the landscape in terms of customers is very fast. We need to match with our support and our decision-making those trends. We are at the full range of technology products, as I was mentioning before. Even if among all those technologies, certainly NAND and low-power DRAM are the key ingredients because the majority of the game will be around NAND and low-power DRAM in the future. I will add to that PCM because this will be a really strong contributor for the future. We are already in a strong play in the market. We have a long traditional success story behind us.
We are, with our estimation and based on our shipment to the market, present in more than 650 phones or terminals or whatever you want to call them in the market, which is plus or minus one-third of the total terminals that are circulating in the market today. That's all. Thank you.
Yeah, in the interest of time, since we're running a little bit behind, Mark, we're going to save the Q&A and wireless till the end and kick it off to Brian Shirley on that.
Perfect. OK, good.
Good morning. For those of you that I haven't met, my name is Brian Shirley. I run what we call the DRAM Solutions Group. I want to spend some time with you this morning going over some of the key trends in our segments here. First of all, as we've explained in the past, we really look at life on a segment basis. We've got four primary segments here in DSG. These three on the left, server, networking, storage, graphics, and consumer, these are really what we consider our premium segments. If you combine that with mobile, which Mario talked about, and if you combine that with the AIM segment, which Tom talked about, you're talking about a diversified portfolio of DRAM for Micron that accounts for upwards of 70%, 75% of our revenue.
Just a staggeringly good position, especially when there's some softness in the typical PC markets out there. Really an enviable position here. Going through these trends one by one in the server space, server continues to grow just. Leaps
and bounds in terms of memory consumption. That's really driven by good growth of the number of servers out there, but probably more critically, the amount of memory per server. One of the big trends that we've been keeping an eye on is the launch of what's called the Romley platform by Intel. This is really the successor to the Westmir boards that are the primary server board shipping today. We're excited about Romley specifically because it allows for more memory content per board. I'm happy to report that Romley is here, it's shipping, and in March you will see very significant double-digit percentage volume shipment growth of Romley. This is a biggie for us. That really helps the virtualization side, cloud computing obviously, and we'll speak about that more momentarily.
Networking and storage, you're talking about not so much growth in density, you're talking about growth in numbers of systems, and that's driven by obviously the need for getting all of this cloud and virtualization data out to the actual users. If you talk about the various segments, sub-segments here of the IP space, wireless, and backhaul, just some phenomenal growth in the networking space. Happy to report here Micron is the number one vendor of DRAM into the networking and storage space. Just a good chance for us to ship value-added products that don't have the kind of volatility that you would see in the standard client spaces. Most critically here, by the way, you are seeing just a huge need for increased bandwidth, lower latency memory in these systems. We'll talk more about that coming up.
Graphics and consumer, you know, a little bit of brightness here as well. I'm happy to report the DTV segment seems to be picking up again. There was some inventory growth there over the last six months, and that seems to be clearing. Some good news on the DTV space. Something that we're pretty excited about, the next generation of game consoles are nearing launch here for 2012, 2013. Now that's kind of, as you can understand, an eight to 10-year trend. Happily, as those new game consoles launch, you're talking about increases in memory of almost 10x per box. Really kind of a nonlinear jump in the amount of memory required by these systems. Something that will spur some good growth in the graphics and consumer segment. Moving over to PC, the story's obviously all about ultra-thins.
I think CES, well publicized now, there was a launch of 14 ultra-thin systems out there, and that is expected to grow to something like 70 systems here by the end of the year. We like ultra-thins, by the way. This is really a, call this a new form factor of the notebook space. I'll talk a little bit about what we're seeing on the content side there momentarily, but really a great opportunity to innovate specifically around power and performance. If you tie that against all of these four segments, again, I'd say really a unifying theme. Despite some softness in the PC space, just some huge opportunities to innovate, notably around power, but also around performance and latency here. If you look at a graph of the computing DRAM segments, these are the segments we serve as well as mobile at the top here, broken down in percentages.
Two trends to note. The first is at the bottom, obviously, server bit growth as a percentage of this overall market, growing by leaps and bounds from something just north of 12% in 2009 to almost 30% by the year 2015. Cloud and virtualization are driving a lot of memory per server. The other trend's obvious, and that's ultra-thin. You can see that in essence, the legacy type of notebook systems, that form factor transitioning into the ultra-thin space hasn't been a big factor in 2011, but by the end of 2012, you'll see some significant growth in the ultra-thin market. Talking specifically about servers, cloud computing here, we have the graph for you on the left.
Cloud and virtualization, what's really happening in this space is that to pull off cloud services well and to pull off virtualization well, as well as another trend that we're keeping an eye on called big data, which is finally here now in the market. This is not just a province of government analytical firms. This is actually moving into the commercial space, the ability to analyze big data and the desire to do so for market trends, for fraud, for all kinds of applications, the financial sector. The next generations of servers allow that. That just flat takes a lot of memory. These servers are shipping with more cores per CPU, more CPUs per board, and when that happens, you need a lot more memory per board to keep all of the CPU power fully utilized. Some great trends here.
Specifically to that point, Micron has launched something that we call the cloud product line. We have these products shipping now, and with Romley, you will see this really take off. These are 4 GB , 8 GB , 16 GB, 32 GB, and 64 gigabyte modules, and with 20 nm, the ability to ship 128 GB per module to feed all of that CPU bandwidth. Scott was talking to you a little bit about what we call the Hybrid Memory Cube and some of the technologies that make that happen. One of those technologies is something called true silicon vias. This cloud product line actually uses true silicon vias, and we're shipping those today, by the way, to go and enable some of these higher-end densities.
Hybrid Memory Cube as well, I'll speak to momentarily, but another technology that's really for the very advanced servers, next generation servers, to break through what we call the memory wall. Again, trying to feed the CPU with all of the data it needs in a cost-effective and power-effective way. I'm happy to report on 30 nm, which we're shipping today in high volume. We do have the lowest measured power both in active and standby. This was something, if you talk to the server vendors, there's nothing else that matters these days besides power. That's where the activity is. They'll take all the memory they can get. They just watch this power budget very, very closely. A huge achievement there. On the networking side, really just a phenomenal portfolio here. You can see on the upper left the growth in network traffic really being driven by video.
The number of systems to service all of the client traffic, as well as the demand for video, is spurring the need for advanced memory systems. It's interesting, increasingly in that space, the companies that Micron competes with specifically on high-end memory, it's less the established other memory vendors and more ASIC and embedded DRAM vendors because those are the companies that can deliver the bandwidth. Now they can't deliver the density. Micron Technology, with our logic technology that we've developed through HMC-like methods, as well as RL, some of the advanced connectivity technologies that we have like TSV, we're in a unique position to be able to deliver that kind of bandwidth at high density for these advanced networking systems. As you can see on the lower right, we pioneered what we call RL2, reduced latency DRAM2. RL3 was launched last year. That is shipping now in volume.
Finally, Hybrid Memory Cube. This will be the segment that actually launches Hybrid Memory Cube, specifically in advanced upper-end routers, specifically for the next generation Ethernet protocols. Now, turning a little bit over to the client side, I want to talk about ultra-thin. This is getting a lot of press out there. Frankly, there's some pretty good news. What ultra-thin is at heart is a new notebook form factor that's obviously very, very thin for the obvious form factor reasons. That gives some unique challenges to the battery, some unique challenges to cooling within the system. In a lot of other ways, the clients are demanding exactly notebook-like performances. One of the unique things you get from that new form factor is that if you think about the ultra-thins that are shipping today, there's no way to upgrade memory once those systems ship from the factory.
In other words, that memory is soldered down on the motherboard. That's a bit of a change for the notebook segment. What that means is that as the clients order these notebook systems from the vendors, they are incentivized to go with a higher amount of memory. If you're going to use the ultra-thin for three years, you want to make sure that you're not limping by with two gigabytes, four gigabytes, for instance, as new applications come online. I'd say we have been very, very surprised and very pleased with the content per box, specifically in the ultra-thin segment. There had been some early reports that this was going to be a net drag on content per box. I'd say contrary to that, we're seeing something fairly positive here because of that soldered-down form factor. Moving down the list, again, you need all of the performance. That's good news.
Specifically, as you go down to the third bullet point, which is the need for very, very low power. I'm going to speak more to something we've done uniquely here in this space on the next slide. I'll just tell you that really the end goal here is the Microsoft badge coming up with Windows 8 that an ultra-thin can run for 13 days standby. That's what we're trying to hit. That's a great problem for a memory company to try to innovate around and gives us really some unique opportunities. Video, very, very critical to these systems. You'll see a good push with DDR4 here. Finally, Windows 8 coming in late 2012, volume 2013. Some press out there that it doesn't, you know, the operating system itself doesn't need as much memory. You know, the OS has not been driving DRAM needs for quite some time now.
What drives DRAM needs are the number of applications you're running with the system. Windows 8 does a great job of allowing more applications to run, specifically memory-hungry applications such as social networking profiles in the background, music, video, something that Microsoft is calling the Metro app suite of applications. All in all, we're seeing a pretty good push for all of these reasons for good high-performance memory, very low power consumption, and good content per box. Now, specifically to that point, what we've done here, if you look at this graph in the upper right, this is plotting the power, the relative power versus price or cost-effectiveness of various forms of memory. There's a lot happening in the ultra-thin space as the enablers look to use the lowest power solution, but also make sure that it's a cost-effective solution as well.
What Micron has done here by adding a number of key modes, binnings, screens to our standard product line and doing a few other clever things in the background, we've enabled power savings on our standard DDR3 product that, again, with some of this binning, some of these other modes that we can enable uniquely, we are able to achieve that 13-day standby power. We do this on standard DDR3 memory, which gives a lot of performance advantages and, frankly, is a much more cost-effective way to go. We launched this yesterday. We're very, very pleased with the enabling activity around this. You will see this out in volume here in the next couple of months, starting on 42nm and moving into 30 nm in the second half of the year.
20 nm, as Scott alluded to, this has been a bit of a speed-up for us and is looking very, very good. If 2012 is the year of 30 nm, you can think of 2013 as the year of 30 nm. We actually will have this in production by the end of the year. Through the speed-ups we've done, as well as the die size decreases that this allows, you will see gigabits per wafer increasing on the order of 90%. Beyond the cost improvements, you really see just a quantum benefit in power. As you can tell from every segment I'm talking about, power is everything these days. Power form factor advantages, specifically for DDR4, will allow content per server module up to 128GB and really allows us to get to the next generation of networking and server products.
Finally, I want to close with just a brief note on Hybrid Memory Cube. This is something we launched last year. I think it was about a year ago today. 2011 was a busy year with our prototype. 2012 will be the year of the real product introduction. 2013, you will see this shipping in real networking and server systems. What this is, again, one more time, this is a Micron-designed logic chip sitting on the bottom, four to eight layers of DRAM sitting on top of that, connected together with something on the order of 3,000 to 5,000 of true silicon vias. You connect that to a CPU sitting right next to it. When you do that together, you give a fire hydrant of data at the lowest power per bit you can find anywhere in the industry. It's just groundbreaking what this is doing.
It's been just a great year. Obviously, a lot of enabling work happening both on the technology as well as the design end. We launched a consortium on this last year that, on the developer side, includes IBM, Samsung, Altera, Xilinx, Open-Silicon, a few others. On the adopter side, it includes over 60 adopters. Very, very pleased with the way that Hybrid Memory Cube has gone. In closing, I think really, despite some softness in the PC side out there, Micron is just in a phenomenal position on the DRAM side, driven again by focus on the right segments, the broad portfolio today to enable all of those segments, and some credibility with shipping those in high volume, meaningful piece of our overall revenue. Finally, leadership in these next generation high-value systems. With that, I think we'll just have a chance for some brief questions.
Brian,
yeah.
Brian, I'm curious about your outlook on the supply and pricing side of the industry at this point. Clearly, there's been some challenges there. We've seen stability and increase in spot pricing. As we go forward, what do you think is required to get the division back to a significant profitable level? Are the supply cuts that came off from some of your competitors enough to do that, or will you have to see more consolidation in the industry for that to happen?
Frankly, I think either would do it. We have seen some improvement out there. I'm happy to say that Micron somewhat uniquely did not idle any wafer starts out of the recent downturn. That's a tribute to the mix as well as the diversification that we've done. Having said that, we do see this primarily as a demand phenomenon as opposed to an oversupply phenomenon. There has been some recent rise in spot ASPs, but overall, it's fair to say that they're still at a challenging level for most in the industry. The supply cuts certainly helped. I'd say more than anything, we're pleased with some of the demand signals coming back. Certainly, mitigation of the hard disk drive supply issue from last year seems to be pulling back some of the demand. Pleased on those aspects. In the PC space, it is still challenging out there.
it fair to say that in the back half of the year that things will improve in terms of the overall profitability picture with demand picking up?
It's tough to forecast in this industry, but we're pretty optimistic.
Thank you.
Go ahead.
Hi, yeah, thanks. Maybe a little more detail on that. What is your view on production this year, both for the industry on the DRAM side and for Micron Technology specifically? On the pricing question, you guys have talked about tracking to roughly down, I think, low 20% in terms of your pricing sequentially in your current quarter. Can you give us an update given what spot pricing has done on that?
Thank you. Certainly. You know, on the bit production side, I'd say the collective sense of the industry right now is that we're going to be somewhere on the order of 30% to 35% bit growth. That's what we've messaged before, and I don't think we see any reason to change that necessarily. Micron is in a position right now where, not having idled any starts, having implemented 30 nm, we will exceed that. Also, beneficially, we have a chance to increase that through other wafer supply needs if necessary. In terms of spot pricing, again, really with less of a focus for Micron on the PC space, I think we at this point probably stick with the guidance we gave at the last conference call. It's still challenging out there. There has been a rise up in the spot pricing market.
I think we're probably for the quarter going with the guidance of the conference call, which was mid-teens, possibly 20% down, but some favorable signs out there in general.
Two-part question on the Hybrid Memory Cube. Are you going to be doing the test and packaging in-house? Part two, how are you going to deal with the non-good die between memory and the logic chip?
Two great questions. In the interest of time, I just say that the test and packaging, yes, we will do the majority of that ourselves. The logic die is foundered outside of the company by industry-leading foundries. That requires some combination of collaboration to make sure that you can work with the foundry to get that right. What I'll tell you on the test side, as well as the non-good die side, is that most of the IP that we're pleased about is in that stack specifically to make sure that we can deliver a quality product with all of the means necessary on some pretty clever test algorithms that we've uniquely developed. One more, okay.
I'm going to take two more, if I may.
Sure.
The first one is, this might be the first time in memory that, or in history, that demand has been an issue for the DRAM business. Do you see that as a natural function of the industry maturing and over the next decade, there'll be periods of demand softness, or do you think that's really just a real unique phenomenon to today? I have one more question.
Very good question. I'd say we have looked at the demand issues more as a function of the macroeconomic climate as opposed to what's happening for the overall demand for DRAM on a long-term secular trend. For all the reasons we've outlined here today, it's certainly clear DRAM is not in the 50%, 70% per year growth cycles that it did year over year back in the 1980s and 1990s. That's a good reason for us to be in the NAND business as well from a growth perspective. With some pretty good numbers coming in on the DRAM side because of server, what's happening in brick region, as well as the growth of these new form factors, we're pretty optimistic long-term on the demand side. Certainly, some macroeconomic softness out there right now.
Given what you're hearing from customers, how long before you would see that as a high-volume application?
Certainly. You will see volume of this. I'd say for us, significant volumes in 2013 and growing from there. It's fair to say this is high-end technology. I mean, it has been costly to develop this technology, and it gives a real value-added performance benefit that is valuable to the customer. It also presents some enabling challenges as well to use all that data. That's a great problem to have, but the customers redesign around that much bandwidth. It enables a quantum performance in their systems. It takes a little bit of time, but you will see this in networking systems starting in late 2012 and high volume in 2013. Thank you very much. I'm going to introduce Glen Hawk right now, who leads our NAND Solutions Group.
All right, thanks, Brian. You know, once in a while, a disruptive technology comes along and it changes the world. Once in a great while, that disruptive technology comes along and it creates multiple trends in society that feed off one another. It creates this positive feedback loop, this upward spiral. That upward spiral creates more demand for that underlying technology itself. Of course, you know, I'm talking about NAND technology. I think it's well established by now that NAND technology has been a key enabler for portable computing. I think that's very well established. I think Apple's done a very good job of making that clear. You can even go to their website to check out what they have to say about flash memory and what it's done for some of their products.
What's just starting, however, and I think what a lot of people aren't fully aware of yet, is that NAND flash is equally disruptive to the cloud, to the data center. I'd like to talk a little bit more about what Micron Technology is doing in both of these areas. Let's start with portable computing first. What we see happening here for three very popular devices, notebooks, ultra-thins, and tablets. On the left here, you know, we see overall growth, substantial growth. The more of these devices there are on the planet, billions that are connected to this cloud, the more that creates demand for cloud storage and cloud computing, which creates more interesting applications. People want more devices. That's what we're seeing here with this overall growth on the left.
If you look at the colors of the bars, we do know that notebook shipments are decreasing, but they're more than compensated for by the rising emergence of ultra-thins as well as tablets. What this means for the NAND industry is shown here on the right. You see an even steeper growth rate in terms of the bit consumptions that we're going to see over the next few years. You see the tiny, relatively tiny dark blue bars there at the bottom that correspond to the tablets. Then you see these much larger regions that correspond to the ultra-thins as well as the notebooks. There are a lot more interesting devices connected to that cloud that I'm not showing here. I'm not showing all of the mobile handsets that Mario talked about.
I'm not showing all of the flash that's going to go into the automotive industry that Tom talked about. In fact, maybe we'll have to come up with a new segment and call it ultra big, because I think automobiles are becoming nothing more than big tablets. At least that's the way I would like to look at it. I know at least I'm holding off my next car purchase until I can get some of that stuff in the next vehicle. Micron makes products for all of these segments. One of the ones that we're very proud of that we've had a lot of success in recently is in the client SSD segment. We have been shipping here for about two years now, a little over two years. We've shipped 1.4 million client SSDs to date. We've been gaining share. We're in the low teens today.
We'll continue to gain share throughout this year. We're very pleased with our start here in this new exciting application for NAND flash memory. This is how we did it. It really started with the NAND technology leadership that we have, that Scott talked about earlier. From that intimate knowledge of the NAND technology, we were able to layer on more vertically integrated capabilities to deliver that final end solution. You started over here on the left in the research and development area. Because we have that NAND technology, we were able to drive that into our controller IP, in particular, the part of the controller that interacts very closely with the NAND flash technology. We've also built our capabilities in overall SSD design.
What this does is it gives us the capability to innovate from an end-to-end perspective, from the silicon, the electrons, the transistors, the circuits, the chip, the controllers, the system assembly for SSDs, and the software that goes along with that. In addition to just creating an environment where we could innovate from an R&D perspective, we also own all aspects of the manufacturing flow. I think that's very significant in this space, especially as the volumes get more significant. We do everything from make the wafers to the final SSDs themselves. This gives us a lot of opportunity to optimize. The simplistic way that people start is we make NAND chips and package and sell them and ship them. Then you can take those NAND chips and make them into an SSD.
We're just scratching the surface on our ability to really intimately integrate those two manufacturing flows in a way that non-integrated people really can't. This also gives us a lot of advantages in terms of being able to effectively optimize our supply chain. A lot of the infrastructure that we have here leverages the infrastructure that we have on the DRAM side and the DRAM modules. Finally, over here on the right, it's not just about designing great drives and then being able to manufacture them effectively, but you have to get them to the customer. At Micron, we have good presence in all the important channels. We got off to a fast start with our own e-commerce channel and distribution.
Now what we're doing is we're really leveraging the deep OEM relationships that we have to sell directly to them in the client space as well as the enterprise space, which I'll talk about later. Those relationships are important because a lot of the volumes are, in the end, going to be driven by those large OEMs. To sell to one of those OEMs, it's a two-way street. Not only do we need to talk directly to the OEMs to understand what their needs are so we can optimize our product, but if you think about it, if you're an OEM, who do you want to buy an SSD from? You want to buy it from somebody that makes the NAND flash. That's 80% of the product. It's where a vast majority of the innovation is.
We've been pretty successful at using those OEM relationships to continue our growth in this market. Again, it all comes back to that silicon leadership. I'd like to go back in time a little bit and compare what's happened or what is happening in the SSD area to what happened in the hard drive industry. If you go back a number of years, what you see is that hard drives were around for quite some time before you saw significant growth. What I'm showing here is the number of suppliers, the number of entrants into this market. As this became a very interesting product, one of the things that catalyzed that was the advent of the IBM PC.
It was sort of that combination between the availability of the hard drives and the ramp of the PCs that really drove a lot of people to go off and start up businesses that made these hard disk drives. Over time, however, there was some consolidation in that industry, as I'm sure you're aware. If you look at some of the drivers for that consolidation, a lot of them were related to the media itself. Whether it was thin film technology earlier to magnetic recording heads to vertical or perpendicular recording, one of the differentiating factors between those that were consolidated versus those that did the consolidating was their intimacy with that underlying media. If we look at what's happening in the SSD industry today, it looks very similar, at least on the front end.
What we're showing here is that the real inflection point was created by the combination of the availability of SSDs and flash for portable computing. What you've seen here is a rapid ascent in the number of entrants into this market. That's, I think, if you notice, the peak's higher and it's about equally steep as what we saw in hard disk drives. I think one of the differences here in SSDs that has resulted in this larger peak is that a few years ago, it was relatively straightforward for some of these companies to dip into the slipstream of NAND components that were finding their way into more commoditized segments. Pull that NAND out, apply some relatively straightforward error correction techniques and some other controller IP to create a very cool product that looked a lot like a hard disk drive. That was an SSD.
I give a lot of credit to a lot of small startup companies that really got the flywheel going for us in this area. A lot of those companies are great customers and they will be for a long time. What we're seeing now, though, at the far left side here is that we are seeing it start to level up. There were a number of events last year that signaled that consolidation is beginning. We would expect that just as we did in the hard disk drive industry. If you look at some of the things that are causing that consolidation, I think one of the bigger drivers right now is we're right on the cusp of the situation where the media is becoming much more difficult to deal with than it was in the past, just as we saw in the hard drive industry.
We've talked a lot about the challenges associated with NAND scaling. Also, as the products ramp to higher volumes, cost becomes more of an issue. Of course, if we're going to sell to any of our customers, it's going to be at a profit. By definition, that's going to be an advantage for us in the long haul. As the volumes increase, another thing that becomes very important is the SSD quality and reliability, and even the service and the support that goes along with that. Because we're vertically integrated and we have all the pieces, we know that we can do a great job in this. As you guys know, Micron is a consolidator, not a consolidatee, and we expect that to be the case in this space as well. It all begins with that core competency in NAND silicon technology.
Here's the marketing version of the technically accurate graph Scott had shown earlier. One of the key proof points we had over the last few months was the introduction of our 20 nm NAND technology, which we are shipping in volumes today. From a product application perspective, this represents a very significant breakthrough for us. This 128 Gb monolithic device that we offer fits, you know, literally on your fingertip. Each one of these we can grind down to where they're about the thickness of a sheet of paper with conventional packaging technology, not true silicon vias. We can put eight of these into a package and you've got a terabit on your fingertip. It's just an amazing amount of storage. We are going to continue to drive that technology leadership. We haven't really talked about the hard stuff yet, which is enterprise.
What I'm showing here on the graph is the enterprise SSD industry growth in terms of gigabytes through the traditional SAS or SATA hard disk drive interfaces. One thing that I've had some questions on in the past is, hey, these are pretty low numbers compared to the other graphs that you've shown. Why are you spending so much time in this area? The reason is that this curve goes up and to the right, and we think there's a little bit of an exponential twist to that. One of the things that's happening is that as enterprise customers, as cloud computing data center experts get familiar with the technology, they're finding different ways to apply this in the data center and change the way the data center works. What I'm showing here is an example of one of our latest offerings in this space, our P400E.
One of the things that we detected very early on here was that a very good place for us to start in this segment was for boot and login drives at an affordable price point. We were able to take our 25 nm MLC NAND flash, do some work to it to get it to where it had the quality and reliability expected of the enterprise segment. We're off to a great start here. This graph goes up and to the right. The reason we're excited about that is we're absolutely convinced that the use of NAND flash in the cloud has just started. What I had shown on the prior graph is shown here in the lower left bar that I'm labeling storage here with the SATA and SAS SSDs that are used in the storage array or on the server.
This really started in the second half of the last decade. That's really where we saw the emergence of this technology. We think that's going to be a great solution going forward as far as we can see. As the data center experts get more familiar with flash in the cloud, now what we're seeing is a lot of excitement around using flash in CAS applications and specifically PCIe cards that are on or in, rather, the server. That's really just started taking off in the first half of this decade. We're very excited about that as well. Ultimately, where this is leading is at some point out in the future, this is really heading towards something called storage class memory, where we're able to put the flash in very close proximity to the CPU. There's going to be a lot of heavy lifting required to pull this off.
You can bet that Micron i s going to be first in offering products for storage class memory as well. To help us pull us up the value chain, so to speak, and really deliver the best possible products for the data center, we recently announced the acquisition of Virtensys. Virtensys was a company that used our PCIe flash cards, our P320 in particular, to create an appliance, if you will, an IO virtualization appliance. A lot of excitement in this area. We acquired this company. While I can't talk externally yet about a lot of the plans that we have in this space, Ed Doller, our Vice President of Architecture R&D, will do that later this year. What I can tell you, and I think what's very obvious, is that the acquisition of this team adds another dimension to Micron Technology's ability to innovate in this space.
What we see going forward is that it's very important that we understand what the cloud customers want and to help build our relationships in that area. This brings some more capability in-house that allows them to have more effective communications with them in this area and to innovate more in this space. Overall, I think that it's well established that flash is disruptive to the cloud. You don't have to hear that from me. Earlier this week, on Monday, EMC unveiled their Project Lightning, a product that we now know is named VF Cache. I love this slide that Pat Gelsinger showed, which, in a nutshell, kind of sums up those three tiers, if you will, of flash penetration into the cloud. On the left, it's shown the kind of performance you can expect from a hard disk drive.
Then there's the array flash or those SAS and SATA drives that I talked about. On the far right, as you see, what's very exciting about the deployment of flash in a PCIe card form factor on the server itself is just a tremendous amount of performance improvement. You can see here, EMC has shown exactly how significant that is to them. I think this is great. I think this is a really great proof point. In addition to them, EMC is saying a lot of things about flash. I was at this launch, by the way, and I'll tell you, as I sat there, I couldn't help but think, wow, what I'm seeing here looks a lot like what I saw Apple do to consumer products.
It really looks like EMC is heading down a path here where they're using flash to disrupt the data center and to make it very easy for their customers to get the most out of flash. They were nice enough to actually show that and announce that they were using our P320 SSD, our PCIe card for this VF Cache application. Rather than say a lot about that myself, we'll go ahead and roll a video here and let you hear it from them directly.
EFD, clearly the number one storage provider for enterprise data center use case today. Our first partnership is with Micron. We have Glen and Greg here from Micron at the Bay from setting up the T320 card, the prime launch vehicle for the VFCache technology set. Micron Technology has truly been a great enabler for VFCache. We're very excited with the performance capabilities that the T320 card brings forward. If you look at specs for that, a 300 GB drive, the SLT technology, the best in class performance capability. We also have consistently delivered the best product in the category, and unquestionably VFCache is that product today. You talk about the Micron role. First, you know, Micron, we are truly delighted with our partnership with Micron. We have worked very closely with Micron as our preferred hardware partner for bringing this forward.
Working for software stack, drivers, all those types of things, details of hardware that they're doing. It simply is a better hardware card today. I showed you some of the specs, the way the control functions are, the way it utilizes hardware-based control capabilities, select load on the CPU. Just the number of performance characteristics, the Gen 2, right by eight characteristics, the way the controller's done. It's just the best in class product today. We're very excited to be working with them. We expect that we will always be taking the best hardware technology. We'll be working closely with Micron to hopefully have them always be the best.
All right, I apologize for the quality of the audio there. I promise I didn't capture that on my cell phone. I think between the time that they announced this on Monday and given our challenging week getting this to you here by Friday, we couldn't quite get the quality that really reflects the significance of what Pat had to say there. I would invite all of you and all of you on the webcast to just go to the emc.com website. It's very prominently displayed. You can go click on that yourself and get a high-quality version of that. You can also listen to a lot of the other great things Pat had to say about the partnership that we have with EMC and innovating in this space and for the use of flash overall in the data center. Thank you. I'll let...
What's the impact of the long drive?
What do you see as the impact of Ivy Bridge on SSDs, both from a sort of an interface technology point of view and just from an uptake point of view?
Yeah, I think that's a good example of what the future holds for us as, you know, at the platform level, some things are being done to optimize the use of flash, whether it's drives or more flash that's located in greater proximity. I think that's actually a good example of some of the innovation that we see going forward as flash becomes more ubiquitous in this space.
Thank you. On a couple of the same questions as for Brian, in terms of pricing, do you have any update in terms of the trends at your earnings call? You said down 15%, pricing may have been a little bit better. Also, on production, what do you expect global industry production to be in NAND and Micron's old production? A third bonus question for you in particular is, given some of the upside that you saw on NAND flash because of the hard drive constraints, is there any way to quantify that at all in terms of how much that was and if there's any risk that when drives come back, that could kind of peel back a little bit?
Yeah, great question. I'll try to answer those together, maybe working backwards. Overall, the hard disk drive shortage that was seen received a lot of attention, there's no doubt. Customers came to us immediately and asked, you know, can you help us out? We did everything that we could to ramp faster and to get more drives out to them. We did see an uptick there. However, keep in mind, this market was on fire to begin with. You add this on top of it, and it was very difficult for us or any other drive suppliers to really fulfill all the needs that they had. I think that it is really hard to kind of tease out a specific number that we can attribute directly to that event versus the baseline growth that we're just seeing anywhere.
I think that, on a related matter, what's probably a little bit more, you know, and this is sort of a short-term situation. I think what's a little bit more interesting is that, and longer lasting, is that the hard disk drive industry has been pretty clear about signaling that costs are flattening and that their equivalent scaling isn't going to be going quite as fast in the future as it was in the past. I think that does tip the balance a little bit more in the favor of SSDs going forward as that cost per bit, cost over bit point sort of pushes us out from an entity perspective, if you will.
Does that mean...
In terms of, you could, very similar answer to Brian from, I guess, the pricing perspective is, you know, a lot of our flash goes into very diversified segments. It's very different results by segment. For our flash that's used either in our own SSDs or that we sell to customers that use for SSDs, we do some special things to those devices. It's been a very good environment for us. It's just been very strong, been holding up very nicely there. I think the margins that we've reported reflect that. In the more commoditized channels, there is some softening. Most of our focus is on these diversified channels and production. No update to our guidance. I'll let Mark address some of our plans from a capacity perspective.
Can you comment when, what's the decision making in terms of using your own controller for SSD on the client side versus enterprise side? A lot of the innovation in the SSD space is actually around the controller and the architecture there.
Yeah.
What does it lead you to use your own decision? Are you agnostic? I mean, you'll just use the best performance controller out there, or is there more of a clearer strategy here going forward?
Yeah, absolutely a clear strategy. I think that we will always use the best controller possible for our product, period. If you look at the way that we've started, we've used the two extremes to date. We have used external controllers, and we've also used internal controllers. I think the P320 product that I just shown is a very good example of what I would call an internal controller. We do use foundries, of course, and we have to partner with others. We don't fabricate the wafers themselves, but the vast majority of that very hardware-intensive controller is coming from Micron. We're pretty proud of what we've been able to accomplish with that. I think that what you'll see from us going forward is a greater use of our internal controller capability.
We will definitely be focusing that on the very top end of the controllers, which are those that are required for the advanced enterprise SSDs. We are developing it in a modular fashion so that we can, with good reuse, rapidly proliferate that into a variety of other segments. That said, we know going forward that we're not going to be able to do every internal controller that we want because the proliferation in this space is huge. We will continue to still partner with some external controller companies, but it will be a balance.
Two questions. First, specifically on the Ultrabooks, where do you think your market share is likely to be by the end of this year? The second question is on the cost dynamics, how do you see yourself in the client side in terms of against the competition? Thank you.
On ultra-thins, I think that I don't have a good share number for you off the top of my head, to be honest with you. I can get that for you later. It is a focus for us. We think that's a very good market. There are some exciting new opportunities to deploy flash there, not just in an SSD form factor, but cache. I think that's pretty exciting. From a share perspective, I don't have that number off the top of my head. I'm sorry, the second part of your question was?
Do you hear the cost comparison?
Oh, okay. Cost comparison, I guess it depends on we have to define who the competition is. You know, there's a lot of people in this space. I do think from a cost perspective, we do, we make the flash, and that's the majority of the bond. You know, we have an advantage there. Against the tier one players, I think you'd have to talk to them about their cost structure, but we're very confident in our technology leadership and our manufacturing capability. We'll be right there with them. Yeah, great question. Yes, I do think that there's a place for that. If you look at the market from our perspective, it's not the most interesting place for us to deploy our technology. We have three bit per cell products. We do manufacture a small amount of them.
In order to optimize our overall portfolio, it's just not something at the current time that we're going after in a big way. If market conditions change, we can easily shift that way. I think that going, you know, Scott said something earlier that I'll challenge him on, but also kind of used to answer your question, which was, you know, Scott talked about going forward with NAND flash, how some of the technology innovations are going to, what did you say, make it easier for the controller or something like that? I kind of have a different perspective, and I think it relates to the three bit per cell question in that, you know, once we develop some of these very advanced, not just error correction capabilities, but digital signal processing techniques, that gives us a capability that we're never going to let go of.
If we get a better technology from Scott, we're simply going to use those tools and tricks to develop a more highly performing or higher reliability product. I think, you know, three level cell with NAND flash is kind of at that same point, is that it's hard enough to make single bit per cell or multi-level cell flash on our 20 nm node today. Again, going forward, you know, it's three bit per cell just becomes harder and harder. We have it, we have it in our arsenal. It's just, it's not the major market focus for us right now.
Could you just comment on what the percentage mix, I guess, wafer outs are like today for your 20 nm product? I guess the other question I have is, as you just mentioned, it's becoming more and more important as a competitor in the SSD market as it relates to firmware and software. How big is the team, and is this an area that you plan to continue to expand going forward?
Okay, yeah. With regards to the 20 nm volumes, we are in low volume production. We'll see this significant ramp occur the second half of next year. I think that the Micron that you're seeing today is different than you had seen a few years ago where a smaller percentage of our focus are in those channels, those markets that ramp quickly with the latest flash. With regards to the software firmware question, absolutely a key learning for us in this space over the last few years. I think that's one of the things that we learned very quickly, even on the client SSD ramp, is that there's so much IP and innovation that's possible in the firmware within the controllers and the software that goes with the drives.
In fact, that's a key differentiator in my opinion that we've seen as we've moved from being primarily shipping through our e-commerce channels into the direct OEM engagements. Even though these products are drop-in replacements in many cases for hard disk drives, for our customers to get the most out of the flash drive, they need help from us to make some changes, some tweaks to that firmware and that software. We have been building our capabilities substantially in that area. I don't want to give you numbers, obviously, on exactly how many people we have there, but we've essentially been hiring people almost as fast as we can in that space.
All right, thanks.
We're just going to take a short break now, about 15 minutes. There are refreshments outside the door to your left there as you go out. Like I said, 15 minutes or so, roughly 9:20 A.M. back in.
I don't know why I'm scared because I've been here before. Every feeling, every word, I've been mad at it all. You never know if you never try to forgive your past and simply be. If you can hear me, say it to let me be your one and only. The pills are on, I get me the jive. The pills are. I'm the one who care. Whenever I'm alone with you, you make me feel like I am cold and dim. Whenever I'm alone with you, you make me feel like I am cold again. Whenever I'm alone with you, you make me feel like I am young again.
Whenever I'm alone with you, you make me feel like I am fun again. However far away, I will always love you. However long I stay, I will always love you. Whatever words I say, I will always love you. I will always love you. Whenever I'm alone with you, you make me feel like I am free again. Whenever I'm alone with you, you make me feel like I am sleeping again. However far away, I will always love you. However long I can take, I will always love you. Whatever words I say, I will always love you. I will always love you. However far away, I will always love you. However long I stay, I will always love you. Whatever words I say, I will always love you. I'll always love you. I'll always love you. 'Cause I love you. I told you you.
Why are you so shy? Ain't like you to hold back or hide from life. I hate to turn up out of the blue uninvited, but I couldn't stay away. I couldn't fight it. I had hoped you'd see my face and that you'd be reminded that for me, it is no worry. Never mind, I'll find someone like you with nothing but the best for you. Don't forget me. I'll fix. I'll remember you safe. Sometimes it lasts for love, but sometimes it hurts instead.
Okay, everybody, three-minute warning here. Try to get everybody in here and get things going again.
No heather, dark eyes on me. Yesterday was those time of our lives. We were born and raised in a summer haze, bound by being surprised of our glory days. I hate to turn up out of the blue uninvited, but I couldn't stay away. I couldn't fight it. I had hoped you'd see my face and that you'd be reminded that for me, it is no worry. Never mind, I'll find someone like you with nothing but the best for you. Don't forget me. I'll fix. I'll remember you safe. Sometimes it lasts in love, but sometimes it hurts instead. Nothing compares on worries or cares, regrets and mistakes, and memories made. Who would have known how bitter your sweet tears will taste? Never mind, I'll find someone like you with nothing but the best for you. Don't forget me. I'll fix. I'll remember you safe.
Sometimes it lasts in love, but sometimes it hurts instead. Never mind, I'll find someone.
Okay, everybody, we're going to get started with Mark Adams up here next.
Good morning. Thank you again all for coming today. We really look forward to the opportunity to spend some time with you and wanted to give you an update on really our go-to-market around what you just heard from the BU. I don't want to oversensationalize anything, but we are really in an unprecedented time in terms of our industry, in terms of how we participate in the industry. When you start at a macro level and you consider the continual uncertainty in global economics, if you look at Europe as an example, there's a lot to be played out. If you look in technology, and as I sat this morning and I listened to Tom and Mario and Brian and Glen talk about trends in technology, 2011 was a year that it all started to play out as we were talking.
When you look at cloud computing and IT architectures, when you look at social media and those types of things, it really started to actually occur, even down to what Glen talked about in terms of solid state drive and the enterprise and storage. It started to actually produce really tangible results and the growth started to happen. You look at something like the supply chain. In March of last year, the Japan earthquake and the tsunami had a huge impact on the front half of the year. Early fall, obviously, the situation in Thailand had a dramatic impact on the hard drive industry. That had ripple-down effects to the rest of the markets. If you put all that together, it's been a very dynamic time for the industry.
We look at it and are trying to take an opportunity of those sets of conditions and how we participate in it. A good case study for that is really the wireless market. If you go back four or five years, I think Mario had a slide that talked about the customer share. Embedded in the customer share of who owns what in terms of the overall wireless market are a lot of different messages, one of which is the evolution of the smartphone and what people are actually doing with the smartphone. Even beyond that, a lot of companies struggle with where do they put tablet computing. Is that a wireless device or is that a notebook? You can see a lot of convergence happening. You can see hyper growth in terms of the raw units.
How it impacts Micron is our ability to really integrate solutions to serve that type of growth. As you'll see in further slides, the dynamics of what we're doing and the BUs talked about is more solutions-oriented than where it might have been five to ten years ago in the memory industry. A good data point to say, okay, so that sounds great. What does that really mean to your company? Again, you heard it today when you heard Brian talk about segments and you heard Glen talk about segments. The segments from server and storage to networking to enterprise, you can see the shift over time. When you look at the bottom of the chart and you look at the personal systems devices, you'll see how that has become less of a mix when you look at the overall portfolio of where we sell our products.
On the other side of it, we've been asking for the last year to two on when is SSDs, when is that going to be something that we can report to and be material to your business? You can start to see that growth. In between there, you see the premium segment. You see server, you see networking. I'm sure you'll see us continue to report out on the success of Micron in the enterprise. I think as you heard, and I'm reinforcing in my presentation, the BUs talked a lot about industry segments and applications. Our focus as we go to market is to really drive unique solutions, custom solutions into those market applications because that's where we believe the value add is for our products and technologies.
This looks back six years, this slide as it relates to the overall technology mix of the company, DRAM versus NAND versus other. I would suggest if you just went back one more year to 2007, this slide would look dramatically different, even further concentrated towards DRAM as we were just starting our manufacturing relationship with Intel at the time. In light of that, you can see the shift to where we are today. As many of you know, we reported on our last earnings call that NAND outpaced DRAM in terms of revenue for the first time in the company's history. We believe underlying these opportunities in new application segments, we believe the portfolio that we have to go to market with is really second to none in the memory industry. We think that's going to help us drive to greater levels of performance going forward.
When you look at the picture, I started talking about the macro trends and the environment in which we compete. If you look at the picture for us, starting with the left-hand side, a broad portfolio of technologies and products, a diversifying segment of applications that we sell to. The implication for us as a go-to-market organization is really that our customer base is shifting. We have to take advantage of a more diversified product portfolio and who we market to. Tom Eby on ESG this morning in his presentation discussed the nature of his channel, versus what you might have looked back 10 years ago and seen a very heavily computing-driven market focus for Micron . We continue to measure ourselves as we go forward to how well we're diversifying our customer portfolio to take advantage of the new technologies that we're bringing to market.
As many of you know, we've talked in the past about the different channels that we have access to, and we view that as a competitive advantage. It's nice to talk about what's there. I just want to take a few slides to tell you internally how we measure our execution, our operating around the opportunities we're seeking. One of the ways we look at it is if you take a look at the top customers versus the rest of the market, your top 20 customers versus the other segments that we sell to in each of the technologies. You can see in DRAM, NAND, we've been able to further penetrate new customer sets to grow the non-top 20 customer base, to diversify our customers. That's usually where we find more value-add customer opportunities. We're going to continue to measure this forward.
Not to say that our top customers aren't important because the next slide I'll talk to that. It's a different corporation and a different competency as an organization that we need to develop and grow and continue to grow from an industry that, again, 10 years ago was primarily selling to PC manufacturers and server companies. As you think about the shifts in the segments, the applications, it also has pretty big ramifications in terms of the customers and the opportunities. On that first slide, when you talk about social media, we're now selling products to the likes of Google and Facebook directly, extending more customers and growing that customer base. The types of customers we change continue to grow and will continue to drive beyond what are usually known as our top 20 customers.
I am cautious because having said that, in our top customer base, we've had a very big focus on how we penetrate with our portfolio, the strength of our portfolio. There was a question this morning in Tom's presentation about how effectively you are marketing your products across the portfolio. About a year ago, this slide would have had a lot more empty boxes in it. I would hesitate to say that where there are empty boxes, it's not by lack of our ability to market to those customers. It's just potentially that they don't participate in those markets. We are very happy with the representation of our portfolio at our major customers, and we are going to continue to leverage that as memory becomes more solutions-oriented. I couldn't help but think this morning when I heard numerous times more than just about the silicon.
I heard about our roadmaps in the silicon, but I heard controllers, I heard firmware, and I heard about enablement of other products. That should show you the shift that's going on with our customer relationships. Beyond our channel through the major customers and our distribution and reseller channels, we've talked often about our captive channel. A really good example of some success that we've had in building our channel is through the Crucial branded SSDs that we go to market with. I would hesitate to say that it probably has been some time that you've seen Micron under our own brand obtain 20% share in any market segment. We've had tremendous success. As a matter of fact, to date, more than half of those drives that Glen referenced were selling under the Crucial brand through our own Micron captive channel.
We will continue to leverage that channel where it makes sense to grow our business. When you couple differentiated technology and new application segments with a broader set of customers, our goal and the end result is to add more value to the company. On the go-to-market side, that shows up more on the ASP side. When I think about that, we continue to measure ourselves consistently against our competition on how we're doing on ASPs versus our competition. We think that we will continue to grow this distance as we add value and really expand our markets into new additional customer types and segments. In summary, a lot of opportunity, new customers for us across all the BU, you know, again, in networking and cloud and enterprise and storage, developing new strategic engagements. When I say new customers, it's not someone small who's buying low volume.
It's the changing dynamics of customers that I talked about earlier in the presentation. We have a very unique product portfolio, again, unrivaled by anyone in our industry. We have the ability to take more products through our captive channel. We want to evaluate that and balance against the rest of our go-to-market. Ultimately, we want to measure ourselves on how we take Micron's capacity and add value to it and make sure we get it to the right customers in the right segments. Before I hand it over to Ron, I realized one of the benefits of this new title that Micron bestowed on me is that you hammered Hawk and Shirley on pricing. Do you have any other questions?
Sorry, I thought I didn't have anything.
This is almost another question. You showed the slide of the premium price that you're getting for both NAND and DRAM. More importantly, Mark, I think you said that you think that that spread is going to widen in the future. That's potentially a really important comment. Would you please expand on that? It was a pretty quick slide you passed by.
Sure. I thought it's a lot. When you look at the opportunities that I represent in the presentation today and that you heard earlier through the BUs, we believe there's more opportunity there to capture. Glen referenced the announcements this week with the partnership with EMC in the enterprise space. Quite frankly, the enterprise portion of SSDs hasn't started yet. It's very early in the game. There's a number of other segments and technologies that we're developing for those segments that are yet to play out. As I said, I really believe that the fact that we've actually demonstrated a propensity to be very strong in these specialty segments, understand requirements, and we have these new technologies come to market, we're optimistic that we're going to be able to continue this trend and grow on the trend.
Thanks. Mark, maybe just expanding on that, with enterprise SSDs, it takes an entirely different support staff or a sales team. Can you talk more about how that grows, how the selling into the enterprise is different than selling to PC manufacturers?
Yeah, sure. I think when you look at the PC manufacturer criteria, you know, certainly quality and price come up there pretty high on the list. There's a certain level of technology engagement, but nothing near to the engagement at the CIO's office or the IT office in a company that's banking on your technology to run their state of storage. The level of engagement is certainly more strategic. I think the early development and qualification process is longer, and the metrics around your performance, around quality and tests are dramatically more important in the enterprise. It's not to say that they're not important in the client, but the level of performance testing and endurance is just dramatically higher in the enterprise. We need to continue to invest to be able to support that post-sale as well.
I think we are able to find the opportunities, and I think you're highlighting a point that we talk about a lot internally, that it's not the same thing as selling a DRAM module or a micro SD card to a retailer. It's a lot different model. We're investing, just like Glen talked about, in terms of the value add around the NAND. We're investing in terms of how we service those customers in a different model.
I just elaborate on the same question. Can you talk about your market share in the enterprise SSD space, and what do you think the next couple of years it's attainable to?
You know, I don't have a good market share number for you in enterprise SSD. We did have our first material quarter that we announced in the December quarter in enterprise SSD. The announcement this week, which we've been working on for some time in development, will certainly help us going forward, but we don't, we haven't released an enterprise share today.
Thanks. You talked about selling into Google and Amazon. Those are different customers than the traditional guys you sell into. I'm just curious whether some of the advantages of Micron shine in particular as you sell to those high-touch customers versus the more traditional server base.
I think absolutely. I think it's in talking to those types of customers, the idea of them building out data centers is actually a pretty strategic weapon for them because that's their business. There are certain aspects of that that they're still ramping up in terms of their own internal capabilities and competencies. From that perspective, I think we bring an experience of engagement around the technology and the solution they're looking for that puts us in a unique role. The other side of that is that role continues to be a little more strategic, again, than client-based devices when you're talking about servers and data center applications for both DRAM and NAND. Architecturally speaking, we get in early on in that process and help them discuss where they're going as opposed to filling out a spec on a commodity part.
Okay.
Thank you again for coming. I'm going to hand it over to Ron.
Good morning and thank you all for coming. I'm the brief interlude between Mark and Mark here, so I'll be brief. I just wanted to cover a couple of topics. One is a number of our competitors have just announced results in the last few days, and I wanted to give you a couple of competitive comparisons referencing our results of our last quarter, last first fiscal quarter in comparison. I got a couple of updates on our two key manufacturing and technology joint ventures that I'll go through at the end. First of all, looking at cash, one of the key measures of our industry being very capital intensive, this is an EBITDA measure, so it's more meaningful when you look at it on a quarterly basis.
Micron is in the blue on each one of these slides, and you can see that we performed quite well in our latest fiscal quarter compared to the latest reporting earnings of some of our competitors. Notably, as you know, we play in a number of different markets, as the BU presenters have already pointed out. Sorry for that. We've got the comparison to Sandisk on here, which is a pure NAND player. We are very comparable in terms of overall results. As you know, the DRAM market's been a little more challenged of late, and the classic pure play DRAM players there on the right, Elpida and Nanya, for example, are struggling with their cash generation capabilities. I'd also point out that Inotera is our partnership. I'll comment more on it in a minute.
They are generating positive EBITDA and cash from operations as they're starting up on Micron's technology. If you look at the operating cash flow over time, it's important to note that Micron has stayed in positive territory throughout this challenging multi-year cycle, going back to 2006, 2007 timeframe on this graph. We're one of the only pure play memory manufacturers. In fact, I believe the only pure play memory manufacturer that actually stayed positive in every quarter throughout the last few years and continues to generate positive cash. If you look at the capital structure, I just wanted to point out a couple of slides on this. Debt to capital ratio is pretty important in our business given the capital intensity. This shows Micron running below the pure play memory competitors in terms of debt to total capitalization. Our target range has typically been in the 20% - 25% range.
We're actually a little bit below that at this point in terms of our performance. In a somewhat challenging market cycle, that's a pretty good place to be. If you look at another measure of debt, this is showing again Micron in blue over time from 2004 through 2011, the ratio of cash to short-term debt. We've got the industry averages for the memory players in the yellow. This shows that over a number of years, Micron's continued to perform well on the ratio of cash on the balance sheet to short-term debt obligations. We've got ourselves in a stronger position in the recent couple of years as the industry has gotten more challenged in terms of their short-term debt obligations and overall debt structure. Now I'll just turn to comment briefly on a couple of our very important manufacturing and technology joint ventures.
IM Flash Technologies—I think it's chopped off there. That should say NAND and not AND. At any rate, this is our NAND venture with Intel. It consists of two pieces. One is the IM Flash Technologies in our Utah operation where we own 51% and Intel is a 49% owner. The output share in this joint venture arrangement goes with the ownership, so we take about 51% of that output. We have a separate cost-sharing agreement on R&D where we share 50/50 the DRAM R&D with our partner through time. IM Flash Singapore at the bottom there is our new activity that we started up. Several have commented about it. We now own about 82% share. We have been investing the lion's share of the capital into the startup of this operation.
It is doing extremely well, moving through the technology nodes and progressing up its yield curves as well and full output capabilities. The actual output share catches up to the ownership on a lag basis, generally about an eight-month lag. Last quarter, we reported that we had about 65% of the output from Singapore IM Flash. This quarter will be about 78%, and obviously, it'll be just about at the 82% ownership level when we get into the next quarter. Again, the R&D cost sharing on this is 50/50 with our partner, Intel. If you look at our DRAM joint venture with our Inotera foundry partner and Nanya, our development partner, I wanted to update you on a couple of items here. First quarter, 2012, we owned about 30% ownership, as did our partner Nanya, owned 30% of the Inotera shares. The remaining shares are owned by the public.
Our output share is split 50/50 historically, and we split the R&D cost sharing 50/50. Going forward here in the third quarter, I'll just update you on some of the financial moves that occurred. You may have heard that we loaned $133 million to Inotera at the end of the calendar year to enable them to purchase the next advanced technology node equipment along with their own personal financing and debt financing that Inotera engages in. Secondly, Micron plans to subscribe to the upcoming equity offering that'll be in the March-April timeframe, be approximately $170 million. Once we acquire those shares, Inotera will repay their loan to us and we'll take the equity share. It depends partly upon the stock price and partly upon the shares issued in that offering because it's prospective. As I mentioned, it will happen in the March-April timeframe.
We'll probably own greater than 40%, 40% - 45% range probably of the ownership. Click the next bullet there. We do expect to receive higher volumes from the advanced technology nodes that are enabled by our investments. Our ownership share is likely to go up and we will be receiving more of the advanced technology node that's enabled by our investment in Inotera. Summing our total activities together in terms of overall volume, this is a graph that shows DRAM on the left and NAND on the right from fiscal year 2009 to fiscal year 2012 to give you a picture of what we've been doing in terms of DRAM gigabits on the left.
The Micron wholly owned is in the dark blue, and it shows the overlay of the Inotera contribution, which is pretty significant in fiscal year 2012 as they go through their ramp yield improvements and technology node migrations from the 42 nm to 30 nm node that they're now piloting and buying equipment for. On the NAND side, on the right, same graph, it shows the two pieces of our joint venture relationship IMFT, which as you recall, we own 51% and Intel owns 49%, and IMFS, where we're approaching 82% of the ownership of that output. In combination with the ramp, the yield improvement at IMFS, and the technology node migration, and the shifting of share from Intel share to Micron share, you can see the volume growth in trade NAND bits. This shows trade NAND.
We think it's more meaningful to look at this measurement because these are the bits that we sell in the market at market prices. The Intel volumes, since we consolidate the joint venture, actually are in our revenue number, but we sell them at cost. When we report margins, for example, operating margins in Glen Hawk's organization, NSG, about half that volume is Intel volume today. Obviously, it's shifting over time, and we sell that at cost. The market volumes we sell with normal market pricing and market margins. We will have more leverage opportunity financially as we ramp that volume at IMFS. I'll summarize the company's in a real strong financial position.
We've got a good balanced portfolio you've heard a lot about, and the balance of our technologies, the balance of our products, the premium pricing that we experienced because of our mix of products, and the way we roll through our technology nodes to keep the best profitability in the total structure, not necessarily the lowest cost, but the best profit equation as we move through time. The company's executed extremely well under the leadership of the businesses. We are strategically positioned to emerge stronger through industry downturns. As you saw, our balance sheet in terms of debt to capital structure, we're continuing to generate good cash flow. We have financial and operating leverage in the business and with our partners, a very, very low cost of capital and low capital cost structure as we go forward.
I'll open it up for any questions and then turn it over to Mark.
Yeah, Ron, if you go back several years ago when you were predominantly a PC DRAM company, this company was hard enough to model for us self-taught analysts. You look at what you guys have done with diversity.
It's hard for you to model.
This is going to be my question. If you look at the diversity of the revenue stream, either the different memory technologies or the end market you're serving or the customers, mix has become a much more significant driver of profitability for the company. How do you on a quarterly basis or yearly basis try to figure out and match the proper way for starts to the right end markets to maximize ROI? What's the process you go through internally? If you look over the last three or four quarters, how successful do you think you've been at doing that?
In terms of our success at doing it, I think it speaks for itself with some of the numbers. We've generated great cash flow. You saw in Mark's presentation the premium we get against the market averages in terms of pricing. Sometimes there's concern that, for example, Micron Technology won't be the first mover with the highest volume under the next technology node, but as already described in a couple of BU presentations, we do that for a purpose. We do that for an economic purpose, and that's to get the best margin generation, not necessarily the lowest cost. That can be a very different answer to your question. Internally, we use the BUs a lot and allocation of wafers among the BUs and then BU decision-making within those BUs to decide how to optimize that mix.
They drive a lot of that in conjunction with our operating organizations and obviously a lot of financial analysis that goes into that decision-making. I think it's a great process, again, evidenced by the results we get. The other piece of the challenge I think you were alluding to is that mix is becoming more significant in understanding what's going on with our business. We've been talking a lot internally in terms of managing our own business and also how we handle external communication to help you understand what's going on prospectively because it just puts another variable in the equation. We'll be working more on that and trying to give you some guideposts going forward. It's pretty exciting that we have that problem. We're victims of our own success in that case.
In terms of the NAND opportunity, it obviously is probably the fastest growing part of your market compared to DRAM. How do you exactly look at capacity expansion going forward? Do you need Intel in terms of expanding into the second phase of Singapore? Would you do it on your own? Clearly, looking at the growth drivers we've seen today, you will need a lot more capacity going forward. That is probably a decision that you might need to make in the next year. If you can give any more clarity on that, it's great.
Sure. One part I'm going to leave to Mark Durkin, who's going to be talking a little bit about capacity growth and supply in the market. He's got some good slides on that, NAND and DRAM, to show what that picture looks like. A short part of the story is that there is some extra capacity needed for NAND, but current projections at least don't show huge growth requirements in terms of NAND capacity growth. Obviously, we can pick a chunk of volume up every year with just technology node migrations. In NAND, we still do need, as an industry, some additional capacity. In terms of your question about how to handle IMFS, we've got the lion's share of the investment already done. The fab is fully ramped to the first step of their 65,000 or so output level.
We've done that, as you saw, with 82% of that capital investment coming from us. We've got a great debt to capital ratio despite the fact that we poured a lot of our capital in the last fiscal year into IMFS. The timing was great. Obviously, the market's in good shape. The bits were needed. You heard Glen's comments about the channel opportunities we have in SSDs, etc. It's a great place to be. We'll be continuing with our partner to figure out moving forward what scale capacity we need. As has been commented in prior meetings, Micron maybe has a different requirement set for our part of the venture. We are a volume memory manufacturer, and we need volume, and we're looking for scale to service all of our businesses.
Intel has some perhaps other objectives that they can comment on in terms of how much volume they actually need to further their platform aspirations, etc.
Thank you. Two questions. One, can you give us any update on your CapEx plan for this year? Two, what is your firm's appetite for acquisitions, whether it be distressed competitors elsewhere in the world or new technology areas, just your general view on acquisitions? Thank you.
I'm sorry. The first question was just an update on the CapEx, right? We have reported on our last earnings call, reiterated that our CapEx budget's about $2 billion for the fiscal year, and there's no new news there. We're on track for that kind of number for the fiscal year. It's still in that range. In terms of appetite for new acquisitions and that sort of thing, I'll leave it to Mark to comment on in terms of those options going forward. In general, I think the company's in very good shape structurally where we are. We're doing great organically with our organic operations and how our core business is performing. Obviously, at difficult market times like this, some of our competitors are in distressed situations. As has already been commented, we're always in the game to look for good opportunities.
Mark will talk more about that in terms of what we've done historically and the position of the company going forward.
Ron, what's your tolerance if you were to participate in some consolidation to have your debt ratio go above the 25% that you said was kind of your long-term thought process? Secondarily, relative to Inotera, is there ever a time in your mind where it would be appropriate to buy shares in the open market?
In terms of the debt ratio question, we've set sort of strategically through cycles to be in the 20% - 25% range. We felt that was the comfortable range for the business. Obviously, we reevaluate that over time. As I mentioned, part of our strategy as a company is always to get in a position to buy less expensive assets. We've been incredibly successful at that. Mark's going to, I'm not going to steal his thunder. He's got some comments about that coming here. We would evaluate based upon the facts and circumstances of the time if we need to adjust that goal for either strategically or for particular activities. That's certainly the range that we think is the right range for now to be operating in. In terms of Inotera, your question was about, oh, yeah. I already mentioned that we are going to be taking an equity share.
That would be a private placement equity offering. We've got about 40% of the shares that are in the open market today. We have no specific plans to do that. Don't necessarily have any aspirations to be owning or consolidating Inotera. We think the structure we've got in place is a pretty good arrangement right now and very capital efficient for us because Inotera is a separate entity, can go to the public markets. Our partner is very involved and very helpful in terms of accessing the public markets, both notably in terms of the debt markets on their behalf and on our behalf. It's a great structure.
Just on the Inotera conversion, if you will, of the debt to equity stake, how did you price that versus, you know, potentially what else is available in the market, including purchasing your own shares?
You're talking about the purchase of the equity that we're dealing with right now or looking at right now. How do you price it? There's an interesting set of rules in Taiwanese regulations in terms of pricing. Part of it's just formulaic. You do a one-day, five-day, and 30-day backward average, and you can pick any one of those three. That's by Taiwanese law regulations. Because it's a private placement, there'd be some discount off of that. That all has to be negotiated and worked out and will be part of the process here in the next month or so. How does it compare to the other use of capital, the $170 million I mentioned in terms of investing in Inotera? How does it compare to other opportunities? Inotera is highly valuable to us, as you saw from the graph. It's a significant part of our gigabit output.
It's a low-cost operation, very valuable to the economics of our business. It's a high priority for us. In terms of other options, we'd look at them on their merits as they become available. Obviously, as I already mentioned earlier, we are always in the market looking for lower-cost assets to acquire as we continue to grow our business.
Okay. One quick question. You know, it's kind of hard given the pricing environment, but I just wanted to find out that your NAND has crossed over DRAM revenues last year. How do you visualize going into 2012 how this mix is going to be given, of course, we don't have that much visibility into the pricing? Where does Micron really want the mix to be, really?
Honestly, if you're talking about DRAM to NAND mix, yeah, I don't think we have a specific target in mind. What I'd refer to, and again, Mark will show some data on this, is that we've got different growth characteristics in NAND versus DRAM as far as we can see it today. As a result, we're probably going to proportionally end up with more NAND over time in general just because of the mix of the markets. That does depend upon whether those market forecasts are correct. We're in the game of memory, and we'll move where the opportunities are. That's the way it looks right now. All right. Thank you all.
You on now? Okay. All right. Sorry about that. Now I can speak in a normal voice. Apparently, the slides aren't working. I'm just going to ask the guys in the back to roll them as I go through this presentation. Before I jump in, let me say this. You should have got a flavor by now that for Micron Technology going forward, our business is really all about delivering not only value-added products, but probably more importantly, memory systems. We understand that to deliver memory systems takes a lot of different assets. It takes a lot of different operational characteristics, and it takes different investment from an R&D perspective. We get all that. That's where the company is going, and we will continue to drive in that direction as a way of adding value-added products. I think that is part of where the industry is going.
Actually, it's a very exciting time for us because when we think about memory five years from now, it's going to look very different than memory five years ago. It's really going to be a much more mature industry, whether there's consolidation or not, in terms of the value add that Micron can bring to the customers and the decommoditization, I think, of the space that we'll see moving forward. If you can roll the slide, we didn't forget that we're a manufacturing company. For Micron to be successful, we got to continue to be focused on being a world-class manufacturing operation and deploying advanced technology into very efficient manufacturing operations.
We have a lot of sight of that, and you can be assured that even though we're trying to drive high-margin products and high-value-add solutions, we haven't forgotten where we came from and what our roots are, which is world-class technology and operations to drive cost-effective solutions. I just want to focus on a few of these manufacturing assets deployed around the world today. A couple of 300, we now have at Micron , we now have five, if you include Inotera and the IM Flash Technologies joint venture. We now have five 300 mm fabs, many of which are quite large and are at full scale. I want to focus on a couple of them and then talk a little bit about the backend because I think as we start talking about systems going forward, the backend is a differentiating capability as well.
If you can roll the slide, Ron commented on what's going on at IMFS relative to ownership and the bit growth we'll see in terms of Micron trade bits as we move through 2012. I can't tell you enough about how seamlessly this manufacturing operation ramped up. From day one, it was really close to best-in-class yields, cycle time, and line yield manufacturing efficiency. It's really been a fantastic operation for us. It's run well ahead of our projections relative to cost per bit coming out of the facility. As we look to the future, we did plateau here right around the end of the year at what we thought was the first effective operating point at around 70,000 wafers per month. We're currently holding there, although I think as we move through the next quarter or two, we'll tweak that up a bit.
We'll tweak that up into the sort of the 75,000 range as we fine-tune tool mix and operations to the evolving technology mix. You'll see a little bit of improvement there while we wait to figure out when we pull the trigger on the next increment. There's still a lot of headroom in Singapore to take the capacity on up into the 125,000 per month range. That'll happen, but we want to make sure we understand, A, what's going on in the marketplace. As you guys are pointing out, there's a lot of interest in consolidation and what's going to happen there in terms of other companies in the memory space. We want to make sure we understand all those parameters before we start paying $1.00 on the dollar for new capital. If we roll the slide and just click through that, let's go back to Inotera.
Ron talked about Inotera as well. You guys saw the significant bit growth that we'll be driving at Inotera in 2012. Not only, by the way, from the successful ramp of the 42 nm node that happened last quarter, but also from the 30 nm node that's happening this quarter, but also from some pretty significant improvements that have happened over the last year in terms of the tool matching between Inotera at the 30 nm node with the other Micron fabs and what's going on in R&D, as well as a lot of operational improvements as we really started to finally get our hands around what some of the issues had been there historically relative to running the manufacturing operation. All of that is really at this point, you know, in the past.
That facility is now operating really at a level that's very, very similar to what we see in all the other Micron fabs around the world. We're quite excited, as we would have to be before we'd continue to make new capital investments in the fab. We feel like there's a pretty good story going on there now. This relationship continues to be a significant piece of our future. You might question, you know, how big do you want to be in DRAM? We heard that question as well. I think the story for us is, as Ron said, we don't have a specific market target for DRAM versus NAND. As we said over and over again, we think of this as a memory business. We want to optimize the memory business. That means we also want to optimize our physical assets. That'll move through time. Yes, NAND's growing faster.
There's going to be a lot of high-value opportunities there. We're going to make those investments. DRAM is also an area that's changing rapidly in terms of the value-add opportunities through things like Hybrid Memory Cube and RL3 DRAM, etc. We've got choices to make. We're going to do that at a detailed level, looking at all the variables that are on the table at the time. If you can roll to the next one, Xi'an, China has been a big story for us over the last year. You guys probably don't normally focus that much on the backend. A lot of our competitors outsource a lot of their backend. Micron has always done all its tests inside. Not only have we done that, we've also done a lot of work through the years optimizing our test operation in harmony with our DRAM products.
Today for Micron, as we've talked about through the years, we've developed our own burn-in ovens and smart burn-in ovens and things of that nature. We also do a lot of internal tester development today. A big focus for the year has been in ramping the second phase of our test operation in Xi'an so that it now represents, or by the end of this year, will represent about 50% of our test capacity. A significant growth is going on in Xi'an, China for Micron's test operations. Obviously, it's a very cost-effective operation for us, as well as a great accomplishment by some of the backend guys to develop the underlying technology that can support things like the more complicated system-level solutions that are more and more a part of our business.
As you start selling systems instead of components, as I think it was Glen pointed out, you start getting below the system level and then optimizing the entire package. There's a lot of things you can do. The backend has certainly been an interesting area for us this year and moving forward. If we roll forward, I like this slide. Probably should have included NOR and PCM , but in terms of capacity, it's such a small slice in terms of wafer capacity. We do look at memory as an integrated memory business. We don't look at it as a DRAM business and a NAND business. What this slide will show you is that we're in a pretty benign environment and have been for a while in terms of supply growth. Yes, it's been more of a demand story here recently.
The growth you'll see starting in the second half of the year, you'll start to see some NAND supply growth as some of the competitors bring on supply, but in a relatively muted fashion and not one that we think is highly disruptive to the business overall. You can see that, as we pointed out already, NAND will continue to be a bigger chunk of the memory business in aggregate. Any strategy that Micron Technology might have relative to capacity additions or consolidation or anything else it does to rearrange its manufacturing operations, it's going to have to contemplate that picture in aggregate. That's what we're doing. If you roll to the next slide, we said there's going to be wafers coming online in the NAND space.
Again, here you can see what we think for 2012 is now, I think, probably a little bit less than what many industry prognosticators are saying. There have been a number of push-outs in terms of ramps by some of our competitors. We've also said, by the way, that we're going to grow maybe a little bit more than this. You never know exactly how these things are going to play out. There's an error bell around this. Generally speaking, we think this is a pretty benign environment for NAND growth. We think that bodes well as long as we don't have some sort of macroeconomic issue. It bodes well for the NAND business. By the way, I go back just a sec here, I want to make sure I explain.
By the way, the colors here, we showed this chart before, but just to make sure everyone's on the same page. The blue line is the overall bit growth. We have broken that out in terms of how much of it is coming from capacity, how much of it is coming from technology migration, and then what is going on in just equipment utilization that is already installed. You can see that there is some supply growth in terms of wafers, but it is not dramatic as we move into 2012. Go to the next slide. This is the DRAM situation. This has been updated based on our most up-to-date knowledge. At the moment, we think it is in the mid to low 30% for DRAM bit growth. You can see ongoing underutilization in the DRAM business, obviously driven by what has been a relatively weak market.
We think even though there has been an uptick in the DRAM business here in terms of ASPs, certainly on the spot market side recently, it is by no means a great DRAM business today. That is why you see no new supply. In fact, so.
and really just everyone optimizing their existing operations to drive the most efficient manufacturing environment they can. If you go to the next slide, I think this is a good slide to kind of give you a sense of where we are and why the market is the way it is. This is just cash cost. Cash cost for the industry on average on the light blue line, and then where we think the range for the various competitors are. Depending on whether you're at the top of the bar or the bottom of the bar, you've got capacity offline if you think it's going to be sustainable in that range.
Over there in the bottom right corner, you can see that sure, we've had a little uptick in DRAM ASPs here in the first quarter, but we're still well below the average cash cost for the industry as a whole. There are certainly a number of competitors who have a cash cost that exceeds where the DRAMs in the spot market are going, what they're going for today. Again, it's an improving environment for DRAM, and our view is we're probably through the worst, that the path forward is better, but by no means a robust environment. That's something that we're cognizant of as we go through our decision-making. Now, I will reiterate something that I think it was maybe Brian said earlier on, which is, you know, Micron didn't take any supply off throughout this cycle.
You can think of us as one of the guys that has the lower cash cost scenario here relative to some of the other guys on this chart. If we roll. Obviously, I think first and foremost, and there have been some questions going to this issue already, is, so what's Micron going to do relative to consolidation? You know, it's a tough environment, particularly in the DRAM space. Is there going to be consolidation? What's Micron's view as to whether it would participate or not participate? And sort of what's the philosophy going forward for Micron? That's a difficult question to answer because at the end of the day, for Micron, it's no different than it's always been, which is, Micron's interested in doing what's best for Micron and Micron's shareholders.
Unless you know all the details around the circumstance, I can't tell you, you know, is Micron a consolidator or is Micron not a consolidator. What I can tell you is, we've got a pretty good track record, and that's what this graph shows. We'll do it for a number of different reasons. We'll do it for capacity, if we can get good capacity at a good price, as in the TI circumstance or the Toshiba circumstance. We'll do it for market access like Lexar. We'll do it for products and customers like we did with Numonyx. We'll do all of those things at different times based on what's best for Micron . When you think about where do we sit today and what's going on specifically in Taiwan and Japan, et cetera, I can't tell you what the outcome is.
I can tell you that Micron will be in there figuring out what's going on and trying to understand if there's something we can do that's beneficial for Micron . That's no different than it's ever been, and it won't be any different going forward because, you know, the market dynamics for Micron and the way I think about the business is very similar to the way Micron thought about the business for a long, long time. I think that's probably a good place to stop and answer as many questions as we can get to here. Before we do that, you know, let's just go back through. Micron 's in a great technology position. We've got all the various technologies, and there's really only one other company today that can support that level of activity to build the right package for the future.
We've got, I think, the best portfolio in the business in terms of its breadth, not only in terms of technologies but also in terms of differentiated products. We've got a great balance sheet, and we're really in a great position to embrace whatever opportunities might be out there in the marketplace.
Hi, thanks. Thanks for holding the analyst today given the circumstances. I have two questions, one on corporate governance, I guess you'd call it, and the second is on production. Before Steve's accident, there was some turbulence. A couple of board members were leaving. You announced that you were going to be leaving. Was there anything happening, or can you share with us your thought process and what was going on at that time and what switched you over to come back full-time as opposed to interim? My second question is, the production guidance that you guys have been talking about on the DRAM side, the 30%-35%, your competitors have talked about it too, and all of you are saying you're going to outgrow it.
Is there any kind of risk there, or is there something we're missing in our calculations such that you can actually stick a 30%-35% growth rate this year?
Yeah, thanks. Okay, to the first question relative to turbulence, the two longstanding board members, Jim Bagley and Terry Aoki, who left the board, really did so because they were 70, and they had reached that point where Micron actually has a policy that that's when board members are going to retire. We're sorry to lose them both. They're both great longstanding board members, and I have personally appreciated all the advice they've given me through the years, but that was pre-planned, and that was going to happen. Relative to my own thoughts and my own decision to come back, there's nothing interim about my view of my role here as CEO of Micron today. Yes, I was going to take some time and really actually spend it with my family, and that was all. There was no disagreement with Steve over any policy issue.
There was no concern over the direction of the company or anything of that nature. One thing you guys should know about me is I can't do anything half-assed. I am not as good as Steve at some sports, but I'm a pretty hard-nosed competitor, and I hate to lose too. As long as I'm CEO of this company, which is for as far as I can foresee and beyond, I'm going to be giving 100% to this company and driving it forward to make it as successful as it can possibly be. The second question about capacity, yes, sometimes when you put all the various pieces of the puzzle together, you get something that doesn't add up to one. I don't know where the error is.
Our guys do a pretty good job historically of modeling where the capacity is going to come and what the technology migrations are going to be. We're just giving you the best answer we possibly can. Who's going to outgrow that, whether it's Samsung or Elpida or Hynix or Micron, or who's going to undergrow what they're saying? I can't tell you, but we know what we're going to do, and we're telling you the truth as best we know. We have our own models relative to what the other guys are going to do, and we're telling you the truth as best we know. It could turn out to be different, but that's our best view of what's going to happen this year. Yeah. Oh, sorry.
Probably just more of a hypothetical question. Suppose you were able to buy a DRAM asset at the right price. Do you think it's possible that you can turn that DRAM capacity to NAND capacity given the more demand in the NAND market?
Yeah, I think this is hypothetical. Hypothetical is always very dangerous, but I'll go ahead and answer it. A lot of capacity in the memory business is fungible. We've done that before. On two occasions, we bought trench technology assets and converted them to stack DRAM capacity. I've talked to you guys in years past about what's the fungibility of NAND to DRAM, and it's a little different today than it used to be given some of the lithography changes that have taken place in the NAND arena relative to the DRAM arena. Generally speaking, there's a pretty significant overlap in NAND and DRAM assets. That's always a strategy that any consolidator or any acquirer of capacity could think about.
Mark, we talked a lot about consolidation. What about using cash around repairing the balance sheet and/or buying back stock? What's your view? How do you rank that? My second question, going back to my inability to model the company, there were some of us that thought that the diversification of the revenue stream would allow you guys to go through a downturn without losing money. You've done a much better job than historical, but that didn't play out. I'm kind of curious, given the fixed cost nature of the business, PC DRAM is only 11% of revenue, but what does it represent as a percent of wafer starts? Is that a better way for us to think about it? As you think about diversification at a wafer start level, where are you relative to where you want to be?
Okay, great. There's a lot of questions in there. I'll try and hit them all, and ask me again if I miss a few of them. First of all, relative to PC DRAM as a percent of wafer starts versus revenue, it is more. It is more for a reason, and that's because you have to have a certain amount of PC DRAM in order, first of all, to have access to the high-value piece that you want. You also have to have a certain amount of that in order to ramp new technology nodes because you don't just turn on a new technology node and start shipping it into the highest quality server opportunity. Brian, do you know what the number is in terms of PC DRAM wafer starts? I don't. It's probably 20%.
That's the right kind of figure to think about. I think to Mark's point, probably the right way to think about it is that as we look at the server market right now, we're actually constrained on ships there for supply reasons as opposed to demand reasons. That percentage of wafer starts, we're always trying to get our server percentage of ships up. From that overall scale perspective, we're where we want to be.
Yeah. The next question was, what about the balance sheet and what's your thought process around impairing the balance sheet or whatever versus stock buybacks or buying down debt versus consolidation? I think Ron addressed it in a similar way to what I would say, which is long-term, we think 20%-25% is the right spot to be. If there's a good enough opportunity, we'll do something different. Certainly, in the short-term, we're prepared to do that if that's necessary. There are lots of different kinds of debt too. There's high-interest debt, there's low-interest debt, there's debt that you can count on, et cetera, and all that goes into the mix. In terms of our interest in going out and diluting the company in some significant way right now, we're going to be very careful about that.
We're also going to be very careful about putting the company in a position where we're not confident we can deal with any additional debt. That's probably all the specificity I can give you on a question like that, given all the variables that exist in the marketplace. We're certainly interested in doing some of that at the appropriate time, but that's always got to be weighed with what are the other opportunities. I think we indicated we'd do some of that debt or stock buyback last year, and we're still interested in doing that versus going out and spending a lot of CapEx, but we always got to trade it off as, okay, what's the return on equity going to be, right?
Yeah. Last year, you announced the Hybrid Memory Cube, and you're rolling that out into the DRAM market, and it's exciting new technology. Do you see a roadmap to take that technology into the NAND space? If so, you know, is there a timeline?
Sure. I think this is where memory in general is going. There's no timeline right now for that. However, there's certainly no reason why, as these memory systems become more solution-oriented, you wouldn't also include significant amounts of storage in a system like that. It's not happening anytime soon, but it's the same trend you see with the flash cache, et cetera, et cetera. By the way, sorry, I forgot to answer the other question around modeling. Let me loop back to that before I take the next one. We have raging debates about this all the time internally, but when we understand the issue you guys have presented with relative to Micron today, we would like to make it easier, but we also deal with the constraint of we don't want to make it easier for our competitors to understand exactly what we're doing.
We do have ongoing discussion about what can we do to fine-tune the information we give you, particularly relative to mix, in order to give you a better opportunity to model without creating a situation where our competitors can take advantage of the incremental or additional information we're giving them that we don't get back.
Mark, a number of opportunities were presented over the course of this morning. As you look out over the next two to three years, what do you believe is the single biggest challenge that's going to be facing the company to capture the opportunity?
I think somebody asked the question earlier on, which is, you guys understand, they didn't say it quite this bluntly, but you guys understand that these high-value opportunities take more resources and that you have to invest in the customer support, you have to invest in the software and firmware, and you've got to take some of your, what would have been R&D dollars going into hard iron, sorry, what would have been capital dollars going into hard iron and manufacturing fab and take some piece of that and apply it to making sure you're really executing on the whole system solution. We understand that. You still have to have the foundation. We're still a manufacturing company, and we're not going to lose that.
We do understand that we need to make those investments in a careful way to support the value-added solutions we're going after and in a way that is prudent given the rate at which those businesses are growing. Getting that balance right is always a trick given how volatile our business is because these aren't things that you turn on and off very easily, or you just waste money. We're trying to get that right given the volatility of our business.
Great. Thanks. I'd also like to express my appreciation for doing the meeting and best success to you and to Mark Adams and the group.
Thank you, Bill. It's actually a pleasure to come out here and do this. It helps get us all back in the saddle and moving forward. Thank you guys for coming.
As you transition NOR to 300 mm production, what are your plans for the freed-up 200 mm capacity that you might have?
Yeah. The question is, what's the plan for our 200 mm capacity as we take NOR and move it into 300 mm? We have, over the long haul, an issue with all our 200 mm capacity, whether it's NOR or imaging or whatever else that we have to deal with. For NOR at the moment, it will be relatively slow because we're not doing it with the legacy businesses. We're not doing it with the 90 nm or the 65 nm, It's all targeted at 45 nm and high end. Some of that's new business, incremental business. We plan to hold on to the stuff that's in the rest of the embedded space that's pre-value add. The place where we have probably more issue is in the wireless piece of the business where that business is clearly slowing in terms of its uptake of NOR bits and declining.
That'll be an issue that we've got to deal with as we move through time. We're looking at a number of different opportunities, new products. You could imagine that we might do something similar to what we did in Japan. That's an option. It's also an option to find some additional products that are non-memory that could socket into some of that capacity. There's also, over time, the possibility that some of that capacity goes away.
Question, one question on execution, one question on kind of product mix in the second half. First question is, Mark, if you just give us an update on the 30 nm DRAM transition, yields, cycle times, how are all the metrics kind of trending relative to your targets?
Yeah. Yields and cycle time are very good. They're ahead of, and I'm not going to tell you what they are, but they're ahead of what we had in our plan and what we anticipated in both Virginia and Inotera. We're really just starting in Singapore. We're very happy with the way that's all going.
Great. Thank you for that. On your 20 nm NAND, can you give us a sense about maybe second half of this year, what percentage of your output's going to be 20 nm?
I think it's going to depend a lot on where Glen takes us. We're pretty flexible there. We have more capacity installed today than we're actually running. It's not because the technology is not doing well from a yield or cycling perspective. It's because we're optimizing margins, not bits. I can't give you an answer. It'll kind of depend on where the market takes us.
Two questions, Mark. First one, can you tell us, given your past experience with Inotera, why are you continuing to invest in there? Rather, why can't you expand in your DRAM factory in Singapore? Secondly, you haven't talked about any updates from the LED side. Can you tell us where you are?
Sorry, from which side?
LED, the LED side. Thanks.
Okay. All right. First of all, relative to Inotera, it's pretty simple. That site is functioning very well now. I just told you the 30 nm is doing very well, in fact, on a par with what's going on in Virginia. Overall, operationally, the wafers that are running for Micron are coming out with good cycle time and good yields on the 42 nm node. Overall, that operation is running much more smoothly than it did last year. We're comfortable with the way the operation's running. Now, why would we take capital and put it there? It's pretty simple. It's the most capital-efficient way to get all those extra bits that Ron showed you. There's a lot of incremental bits we can get out of Inotera by moving that capacity to 30 nm and taking a disproportionate share of that 30 nm capacity in Inotera. It just makes financial sense for us.
Sorry, the LEDs, yeah. We're really trying to think about that business as a solid-state lighting business as opposed to an LED business. The LED business is pretty tough. I think our vision relative to that is the same as we see memory, which is, we don't want to sell LED components. We want to sell solid-state lighting systems. That's how we think we'll make money over the long haul. Having said all that, we're doing a lot of technology development on LEDs themselves because we think that underpins our ability to build differentiated and value-added systems. We like our technology position in terms of the differentiated technology we've developed. I don't want to say too much about what that is.
Having said that, we're not to the point yet where we're looking at saying, hey, we're going to commit a large amount of capital to this because the business is soft and we got a little bit more work to do. They're hitting their milestones, but it's not something we're pouring a lot of capital into today.
Hi, Mark. Can you talk a little bit about how you tie your CapEx plan to the technology transitions? We're going to see a lot of pretty significant technology changes. If you look at NAND, you've talked about your three-dimensional. What impact is that going to have on the way we think about your CapEx plan? The 450mm transition is out there, whether it's a couple of years, three, four. How is that going to impact your CapEx plans? If you could link the technology and CapEx with us, I think it would be helpful.
This is something that, I think historically we've tried to be kind of clever about. I can point you to a few things. For instance, when we ramped 300 mm DRAM, we did it on copper metalization. We did that probably five years ahead of the rest, so we wouldn't have to retool. When we did our NAND ramp, we started out with half the manufacturing capacity, pitch doubled even at 50 nm because we didn't want to buy a bunch of emergent scanners that didn't have the registration capability to take us forward. We knew we would need that double patterning technology in order to get down below 40. We put it in place early and wrung the bugs out and then had it there as we made the 34 nm node transition. That's part of what drove that technology leadership that Scott showed you earlier on today.
I think we've established a track record that we can think through these things and time them appropriately. Having said that, specifically relative to vertical NAND, that is coming. We showed you that we think our existing capacity is extendable from 20 nm on down. We're also cognizant that there's an investment coming in vertical NAND. To the extent that drives significant efficiency relative to planar NAND capacity, we're trying to weigh that and make sure we don't make the investment at the wrong time in additional planar capacity that then turns out to be somewhat wasted in terms of its scalability going forward. We're cognizant of those things and we're thinking them through. The detail on what all goes in the hopper and what the output of that is, I think is probably a three-week conversation and not something I want to share publicly.
There was a second piece, sorry. Yes, it's my confidence that we're going to be able to deliver. Scott, frankly, these days, but at the end of the day, he's got to convince me too that we have a technology that's going to be manufacturable in a timeframe that gives us the confidence to go out and make those big bets. I think one point I should tie in here is, okay, we've got a fab in Singapore that's two-thirds ramped. It's at a nice volume operating range that gives us pretty good capital efficiency, but we could improve that if we took the volume up a little bit more. We're not doing that. We're not doing that for a couple of reasons.
One is we want to make sure that we make the right decision as to what kind of capital we put in there relative to the technology question you're talking about. The other is we want to make sure we make the right decision relative to doing that versus buying used equipment from somebody else, so to speak, in a consolidation play. Finally, you know, we want to make sure that we really understand the market. All those things go in there in terms of that decision-making process. If your question is, what's the algorithm? You know, I can't share that. There's no formula I write down, but I can tell you we play out all those scenarios and then do what we think is the best. Now, relative to 450 mm, I think Scott commented on this. This is not a big issue for us.
If we were out building a major new greenfield fab today, we would have to consider that. Even so, frankly, it's pretty easy today to go out and build a new 300mm fab if you need greenfield capacity and not worry about it becoming useless due to a 450mm transition in 2018 or 2019. I think that's a better question for two years from now. I will admit, by the way, that we were probably a little late going to 300mm. One thing you can count on us as a management team to do is try not to make repeat mistakes of the past. We'll be cognizant of that and pay close attention to it. Okay. Oh, is there one more? Okay, let's take one more and then we'll wrap it up.
You know, from Glen Hawk's presentation, there's a big focus on SSD drives, right? You'll move into that. Now, that's a very different business than moving commodity, you know, bits with a NAND or DRAM. What are the changes you need to make to succeed in that? That's the first question. Secondly, could you comment about your biggest competitor and how, you know, Samsung and their increasing market share and what that means for Micron Technology and how you should position yourself?
Okay, good. Relative to the first question, we've already done a big piece of it, which is we reorganized the company and brought in a lot more management talent. That's happened over the last couple of years. We've also done a lot of addition of personnel resources in order to facilitate not only the development side with firmware and software. We didn't put up a picture of the R&D center in Shanghai, but we're adding resources there as well as other places domestically in Boise and other places in the United States in order to make sure we have the development resources. We're adding resources on the sales side in order to make sure we have a go-to-market plan to support the customers. Those are kind of the big pieces there. Glen commented a little bit on, sort of beyond that, what's the next level beyond SSDs.
It's making sure you're also building the infrastructure in order to build appliances. The Virtensys acquisition, I think, is an example of that where you see us going out and selectively picking off capability and resources that we can use to continue moving higher up the value chain. You'll see us do more of that as we find the right opportunities. Sorry, the other piece was, oh, Samsung. How are we going to deal with the fact that they're continuing to grow market share? I think we're continuing to grow market share. When you look at the overall memory space, Micron' s been effective doing that. Some of it's been done organically. Some of it's been done through acquisition. Those are still two strategies that are right in the middle of our operating parameters. By and large, we just got to be smarter, right?
This is not a new problem for us. This is a problem that's been going on for a couple of decades. We are continuing, I think, to do all the right things from a product portfolio perspective, as well as an asset acquisition perspective, as well as a technology perspective to stay in the fight. My vision obviously is not to stay in the fight, but to eventually be the number one. We got a lot of stuff in our toolkit, and we're going to use all of them at the appropriate moment. Maybe that's a good place to wrap up. Thanks again, everyone, for coming.