Good afternoon. My name is Karen, and I'll be your conference facilitator today. At this time, I would like to welcome everyone to Micron Technology's third quarter 2015 financial release conference call. All lines have been placed on mute to prevent any background noise. After the speaker's remarks, there'll be a question and answer period. If you'd like to ask a question during this time, please press star, then the number one on your telephone keypad. If you would like to withdraw your question, press the pound key. Thank you. It is now my pleasure to turn the floor over to your host, Kipp Beddard. Sir, you may begin your conference.
Thank you, and welcome everyone to Micron Technology's third quarter 2015 financial release conference call. On the call today is Mr. Mark Durcan, CEO and Director, Mark Adams, President, and Ernie Maddock, our newly appointed Chief Financial Officer. This conference call, including audio and slides, is also available on our website at micron.com. In addition, our website has a file containing the quarterly operational and financial information and guidance, non-GAAP information with reconciliation, slides used during the conference call, and a convertible debt and capped call dilution table. If you have not had an opportunity to review the third quarter 2015 financial press release, it is also available on our website at micron.com. Our call will be approximately 60 minutes in length. There will be an audio replay of the call, accessed by dialing 404-537-3406 with the confirmation code of 68369558.
This replay will run through Thursday, July 2nd at 11:30 P.M. Mountain Time. A webcast replay will be available on the company's website until June of 2016. We encourage you to monitor our website at micron.com throughout the quarter for the most current information on the company, including information on various financial conferences that we will be attending. You can also follow us on Twitter at MicronTech. Please note the following safe harbor statement.
During the course of this meeting, we may make projections or other forward-looking statements regarding future events or the future financial performance of the company and the industry. We wish to caution you that such statements are predictions and that actual events or results may differ materially. We refer you to the documents the company files on a consolidated basis from time to time with the Securities and Exchange Commission, specifically the company's most recent Form 10-K and Form 10-Q. These documents contain and identify important factors that could cause the actual results for the company on a consolidated basis to differ materially from those contained in our projections or forward-looking statements. These certain factors can be found in the investor relations section of Micron's website. Although we believe that the expectations reflected in the forward-looking statements are reasonable, we cannot guarantee future results, levels of activity, performance, or achievements.
We are under no duty to update any of the forward-looking statements after the date of the presentation to conform these statements to actual results.
With that, I'd like to now turn the call over to Mr. Mark Durcan. Mark?
Thanks, Kipp. For fiscal Q3 2015, Micron posted total revenue of $3.9 billion within our revenue guidance of $3.8 billion-$4.05 billion. Revenue was sequentially lower as expected in fiscal Q3 due to near-term market headwinds driven primarily by weakness in the PC sector. Net income was $491 million and earnings per share were $0.42. Non-GAAP net income was $620 million, and non-GAAP earnings per share were $0.54. Free cash flow was approximately $600 million based upon operating cash flow and capital expenditures of approximately $1.3 billion and $730 million respectively. The team executed well in the challenging market conditions during Q3 as we supported our customers across a broad range of market segments. Micron is a memory company that produces technologically advanced components, systems, and subsystems for the global marketplace. As such, it's important that regardless of near-term market conditions, we remain focused on the long term.
We are continuing to deploy advanced process technology to enable leading-edge products for our customers and to drive ongoing manufacturing efficiency for Micron. Our customers are looking for a partner in memory system development who is capable of delivering advanced technology systems and solutions that are tuned to their product requirements and who will provide their product and their companies with a leading-edge position in the marketplace. Today's customers are also looking for ways to gain more innovation and efficiency in their own system design as memory is being deployed in more ways than ever previously imagined. From cloud servers to mobile infrastructure and automotive design to industrial and medical devices, Micron Memory is a preferred memory partner and is at the heart of technology product innovation.
Turning to overall market conditions, we expect stabilizing DRAM ASPs across the broader market over time as we manage our product mix and distribute our capacity to a broad set of value-added market segments. Consistent with prior expectations, we are forecasting DRAM industry supply bit growth in the mid-20s in calendar 2015 and in the low to mid 20% in calendar 2016. We currently believe that DRAM demand in calendar 2015 in aggregate will be at or exceed supply. For NAND, the market continues to consume planar NAND at increasing rates, and we look forward to volume production of Micron's differentiated high-performance 3D NAND later this year. We expect NAND industry supply bit growth in the high 30% range this year and in the mid 30% range next year.
We believe supply and demand are in balance this year. Unconstrained demand is above supply in future periods based on currently known capacity plans. We are making good progress with fab expansion in Singapore. We expect the market will demand all of the 3D NAND output we can produce, given the attractive cost and performance of our technology and an elastic storage market we will sell into. Advanced technology deployment requires focused investment and carefully managed transitions. To effectively execute next-generation technology production at scale, we are converting tools and systems today to allow for testing, samples, and product qualification in the future. These activities drive future growth and margin expansion opportunities in our business, although they often limit short-term bit growth and cost reductions. We are continuing to make capital investments today to position ourselves for next-generation production capabilities and for technology leadership.
As always, we will manage our capital to deliver long-term returns and shareholder value. For today's call, I've asked Mark Adams to summarize our operational and BU results. I'd also like to welcome Ernie Maddock, our recently appointed CFO. Ernie possesses outstanding industry experience, has hit the ground running, and we're excited to have him on the team. He will cover Q3 financials and Q4 guidance. I'll return at the end of the prepared remarks with a quick summary and lead in the Q&A. Mark?
Thank you, Mark. I will begin by providing an update on our key technology development initiatives, then review our DRAM and NAND businesses, and conclude my portion of the call with commentary on each of our four business units. On the technology deployment front, we remain focused on three major technology initiatives: completing the ramp of our 25-nanometer DRAM technology, driving scale output of our 20-nanometer DRAM technology, and the launching of our 3D NAND technology. We are currently shipping early production 20-nanometer DRAM and will continue to ramp throughout the remainder of this year. We are expecting bit crossover in the first half of calendar 2016. We are still on track for low-volume production of 3D NAND in the second half of 2015, ramping to a significant percentage of our trade NAND supply in calendar 2016. Now let me discuss our DRAM and NAND businesses.
Let's begin with our DRAM business, which represents roughly 61% of our total revenue in fiscal Q3. We deliver DRAM solutions to a variety of market segments. While PC builds declined well below seasonally slow demand in the first half of the year, we saw relative stability in other end markets and responded to these conditions by adjusting our production mix throughout the quarter. As a percent of DRAM revenue in fiscal Q3, mobile was in the high 20% range, the PC segment was in the low 30% range, the server business was in the low 20% range, and networking, graphics, and AIM comprised the remainder. We continue to move production from DDR3 to DDR4 to meet growing customer demand across our customer base for long-term, more stable business. Moving on to our NAND business.
Trade NAND revenue represented 32% of total revenue in fiscal Q3. Performance was consistent with our expectations. Our trade NAND bit growth was approximately flat due to mix shifts in favor of longer-term design win opportunities for both Micron Managed NAND and MCPs for mobile and SSDs for our storage business. These products represent strong growth opportunities for us going forward. As a percent of trade NAND revenue in fiscal Q3, consumer, which includes cards, USB, and components, was in the mid-40% range. Mobile, including MCPs, was in the low 20% range. SSDs were in the high teens. AIM and other embedded markets combined to roughly mid-teens. We saw a 3% uptick in NAND component pricing in the quarter as we moved more bits to our own SSD and mobile business and reduced the supply available to the transactional channel market.
We continue to better position our MLC portfolio to focus on strategic customers and higher-performance segments. In fiscal Q3, these efforts reduced our MLC shipments into the existing TLC-enabled components channel by approximately 30%. Moving on to our business units. CMBU. For our compute and networking business unit, revenue was $1.5 billion in fiscal Q3, with operating income of $266 million. CMBU was impacted by lower ASPs driven by softness and demand from the PC segment. Consistent with our statements on the last earnings call, we have reduced output targeted at the PC segments in favor of faster-growing, more stable segments. We expect better relative performance for PC builds in the second half of calendar year 2015, along with continued DRAM content growth, resulting in PC DRAM bit demands up slightly for the year.
CMBU enterprise customers continued their transition to DDR4 technology, including 8 gigabit DDR4, to support workloads that require both higher performance and higher density modules. We continue to believe that applications such as in-memory database computing will drive substantial growth. In the networking segment, we see continued LTE deployments in emerging markets, which should represent additional opportunities as we move forward. Cloud server represents a high-growth segment, with analysts projecting 50% bit growth year-over-year. Growth in our graphics business was driven by sales into game consoles and high-performance graphics cards. This segment is transitioning to GDDR5. We also commenced shipments of our first 20-nanometer graphics products. I mentioned our commencing 20-nanometer shipments earlier in my script. These initial products are primarily in support of CMBU, for example, our compute customers in the PC segment.
Revenues in Micron storage business unit were $901 million in fiscal Q3, down 6% sequentially, as we opportunistically shifted more NAND bits to higher margin businesses such as our mobile and embedded business units. Gross margins were up slightly in the quarter as we continue to focus on improving our storage business. Operating margins were slightly negative as we remain focused on investing for sustainable growth in this area. Consistent with our stated strategy to improve Micron storage business, we made good progress on key milestones in our third quarter. We announced availability of our new 16-nanometer TLC planar NAND components in fiscal Q3, and already have several channel customers buying our TLC who will input them into SSDs, consumer drives, memory cards, and other products, offering high-density storage products to market based on this technology.
Our SBU enterprise and data center businesses both grew 45% sequentially, albeit from a lower base. We are pleased with the progress of our collaboration with Seagate, we'll have our first SAS SSD launch resulting from this partnership later in the summer. We have begun early sampling with customers and already have secured two qualifications slots with major OEMs. Finally, we will release our own consumer SSD based on TLC NAND technology in the second half of 2015. We expect to have roughly 50% of our SSDs on TLC by the end of fiscal year 2016. Revenue in MBU, Mobile Business Unit, was $938 million in fiscal Q3, up 10% sequentially. Operating income was $296 million, or 32%, up from 31% in fiscal Q2. The mobile market supply-demand balance remains healthy.
Our mobile business continues to benefit from increasing content growth across the entire range of mobile products. Three to four gigabyte phones announced at Mobile World Congress in March are now hitting the market. Low to mid-price phones targeted at emerging markets are being built with significant memory content, including DRAM specs at 1 gigabyte and above. We are also seeing a pull-in of next-generation 4G LTE chipsets and reference designs that double the content of both DRAM and NAND from 2 gigabytes to 4 gigabytes of DRAM and from 8 and 16 gigabytes NANDs to 16 and 32 gigabyte configurations. Micron continues joint validations for low-power DDR4 across chipset platforms. LP4 adoption is currently limited to the very high end of the market today, but will be adopted more broadly in calendar 2016.
The rapid adoption of eMCPs in the high-growth mid-range market creates a significant opportunity for Micron as eMCPs regard both low-power DRAM and NAND, demonstrating the strength of our portfolio. Micron's Embedded Business Unit posted revenues of $483 million with operating margin of 20%. Sales of our automotive-grade eMMC hit an all-time high. We introduced our auto-grade low-power DDR4 and our high-performance G18 parallel NOR Flash devices in fiscal Q3. These products enable improved performance and power reduction for critical applications in high-temperature, rugged environments. EBU also experienced strong demand in the gaming business. Demand for high-density 45-nanometer NOR Flash solutions are being driven by regulatory change in the Japanese gaming sector, we expect this elevated level of demand to continue. It is also worthy to note that EBU shipments of NAND increased 27% from the prior quarter.
To continue our commentary on fiscal Q3 results and Q4 guidance, I'll turn the call over to Ernie.
Thanks, Mark. It's a pleasure to be joining the Micron team, and I look forward to meeting many of you at our upcoming Analyst Day in August. The third quarter of fiscal 2015 ended on June 4th, and the results include net income of $491 million or $0.42 per share on net sales of $3.9 billion and gross margin of 31%. Aside from our recurring items, non-GAAP adjustments include an additional provision for income taxes relating to a tax rate change in Japan that resulted in a reduction in the value of the deferred tax asset from MMJ operations. Non-GAAP net income for the third quarter was $620 million or $0.54 a share. Let's turn to results by product line, starting with DRAM.
DRAM revenue decreased approximately 13% compared to the second quarter, reflecting approximately a 10% decrease in per bit average selling prices and relatively flat sales volumes. DRAM gross margin was in the upper 30s range as cost per bit decreased approximately 6%. As was previously noted, we adjusted our output mix during the quarter and exited with relatively flat DRAM inventory levels. On the trade NAND side, which includes our growing MCP business, revenue increased approximately 3% in the third quarter, with average selling prices increasing approximately 6%, partially offset by a slight decrease in sales volume. The increase in the trade NAND average selling price was mix-related, as higher ASP units grew relatively faster than the overall average. Trade NAND gross margin improved slightly compared to the prior quarter and was in the low 20% range as the increase in mix-related ASP exceeded the higher related costs.
The company generated operating cash flow of $1.3 billion during the third quarter and ended the quarter with $7.3 billion in cash and marketable investments. During the third quarter, we received $1 billion in proceeds from the issuance of high-yield notes, and we deployed $782 million to repurchase a portion of the outstanding Series C and D convertible notes. Cash expenditures for property, plant, and equipment during the first three quarters were $2.3 billion, and we continue to expect capital expenditures to be within our previously forecasted FY 2015 range of $3.6 billion-$4 billion. SG&A expense for the quarter was below our guided range as a result of lower legal costs, while research and development expense was above our guided range, primarily due to higher volumes of wafers used for development of new products and technologies. Looking to guidance for the fourth quarter.
DRAM gross margins for the fourth quarter, using quarter-to-date ASP and projected mix for the quarter, should be down mid-single digits compared to the third quarter based on bit production flat to slightly up, while average selling prices will be down mid to high single digits and cost per bit up low single digits. Key items affecting our DRAM guidance for the fourth quarter include the continued effects of higher mix of mobile and DDR4 products, which have larger die sizes and therefore produce fewer bits per wafer. Fiscal Q4 quarter-to-date mix-adjusted ASP is below the third quarter average, due primarily to reductions in PC DRAM and cloud server ASPs. Pricing in mobile and specialty DRAM remains relatively stable. We have and will continue to carefully balance output volumes between our end markets.
The trade NAND gross margins for the fourth quarter, using quarter-to-date average ASP and projected mix for the quarter, are expected to be flat compared to the third quarter based on bit production down low to mid-single digits, while ASPs are relatively stable and cost per bit is approximately flat. Key trends for the fourth quarter affecting this guidance are like-for-like pricing down slightly offset by average selling price improvements on changes in mix and continued focus on mobile and Managed NAND products. These products generally have both higher ASPs and costs. On a consolidated basis, we are guiding total revenue for the fourth quarter in the range of $3.45 billion-$3.7 billion.
Looking at other P&L and cash flow results and guidance, SG&A spending in the fourth quarter is expected to be in the range of $180 million, with research and development expense in the range of $400 million. For non-GAAP guidance for fourth quarter, please refer to the dilution table posted along with other materials for this call on our website. The dilution table reflects the anti-dilutive effects of our cap calls at various stock prices. Now, I'll turn it back over to Mark Durcan.
Thanks, Ernie. In summary, I believe Micron executed well in fiscal Q3 as we supported our customers across a broad range of market segments. We expect to continue to similarly navigate changing market conditions in fiscal Q4. Micron is executing on key milestones in technology deployment and product introductions. We believe strongly in the future of the industry and our company. Operator, I think we're ready to begin Q&A.
Thank you. Ladies and gentlemen, if you do have a question at this time, please press star followed by the number one key on your touch-tone telephone. If your question has been answered or if you'd like to remove your line from the queue, please press the pound key. Again, if you do have a question, please press star then one at this time. Our first question comes from the line of Harlan Sur from JPMorgan.
Good afternoon. Thank you for taking my question. DRAM revenue bit shipments were down slightly quarter-on-quarter. Can you just tell us what your bit production growth was? I think you said you didn't build any inventory. Also, I think you previously talked about high single-digit production shipment growth over the next couple of quarters, but the main quarter results and the August quarter guidance certainly doesn't seem to be suggesting this. Is the Micron team still targeting bit shipment growth in the mid-teens range this year?
This is Mark Durcan. Maybe I'll start with the last part of that question. Then Mark Adams can comment a little bit on the mix effects and what happened with DRAM sales bits. For 2015, I think it's still fair to say that had we not had mix adjustments as we moved through the year, we would still be driving to a high-teens number for DRAM bit growth. Obviously, the number for the year is now going to be less than that given the changes we've made as we move through the year. As we think forward over the longer term, clearly we've said in the past, and we continue to reiterate, that we'll be higher than market as we move through calendar 2016.
On the NAND front, we've been less than market in 2015. We're still evaluating exactly what our capital plans look like for 2016. We'll dial those based on market conditions as well as ongoing evaluation of the technology introductions in Singapore relative to 3D NAND and deployment of other advanced technologies there. Mark, if you want to take the DRAM sales bits and the mix effects that drove that, maybe that's the best approach.
Sure. We've said all along that we will obviously take in the market dynamics into play as we think about how we allocate our capacity, given some of the headwinds in the PC market
We redirected some of that capacity to either more strategic long-term markets, such as the DDR4, then some of the PC compute bits, just to more attractive long-term stable markets. A good example of that would be cloud servers. Another example would be our mobile business. We think those are the right things to do, but they do have a lowering effect on the bit growth.
Okay, great. Just my follow-up question on the NAND front. I think the team is starting to make the move to 16-nanometer TLC. You're driving some production of 3D. Obviously, that'll be a bigger part of the mix in the second half of this year and next year. The team is still looking for a decline in production growth in the August quarter. What's your sense on when we should expect bit production to start to inflect higher for your NAND business?
Well, again, that's going to depend on the details of the mix decisions we make. What's been going on here over the last quarter or two is that as we transition to a higher percent of SSD sales, we build WIP. The time the product takes through the line increases, and that creates a dip over a confined period of time as we build that WIP. Mobile is a similar type of effect as some of that output goes out in MCPs as well.
Okay. Thank you.
I think that the longer question is exactly when will we see NAND bit growth. I think you'll see that emerge as we move through the next number of quarters.
Thank you.
Thank you. Our next question comes from the line of Romit Shah from Nomura.
Yes, thank you. Mark, you mentioned that for this year, your expectation is that DRAM demand will exceed supply, but ASPs year to date in DRAM are clearly worse than expectations. How do we reconcile that?
Yeah. What I tried to indicate is that I think in aggregate for the year, the demand is going to exceed the supply, which I think your takeaway from that would be that we believe that demand is going to be higher in the back half of the year, as we said on the last call and as we continue to believe. That as you sum that up over the entire year, you will come to the conclusion that the growth that we continue to see in some of the other markets, mobile, data center, enterprise server, et cetera, and networking, will offset what's been a slow PC period. Even PCs, I think, our view is that by the time we get to the end of the year, notwithstanding the decrease in unit sales, there'll be a small net growth in bits into the PC segment.
I guess what I'm alluding to is that if you look at the ASP trend over the last few periods and your DRAM bit growth relative to peers, I'm just curious, to what extent do you believe that the industry is still being rational?
Well, I think it's fair to say that probably nobody expected the PC segment to be as slow as it's been. I think it's also reasonable to expect that in an industry with a number of high-growth segments, balancing supply and demand is never going to be precise. Ripples in the relative balance of supply and demand, I believe, are inevitable in a market like this. What I would say is that probably supply got a little bit further ahead of demand than many of us anticipated, but that it's not necessarily an indicator of a deliberate strategy by any particular player in the marketplace.
Okay. Thank you very much.
Thank you. Our next question comes from the line of Mehdi Hosseini from SIG.
Yes. Thanks for taking my question. I want to go back to a prior question. In that context, when we go back to when industry consolidated, we all expected that mobile and servers are going to be the more of a secular growth driver. These two segments obviously carry a die penalty. Given the consolidated industry, why is the industry more proactively seeking higher margin? If there is a mix issue in the near term, it seems to me that these headwinds are more structural since the demand drivers are actually in the mobile and server. I wonder if you can help to better understand how the industry is going to overcome this, especially as migration to sub 20-nanometer is going to be relatively more challenging and lengthier. In that context, how are you going to be able to manage margins?
I have a follow-up.
I think the answer to that question is that in an undersupplied market, as we had a number of quarters ago, you tend to get margins that are higher in maybe the least attractive long-term part of the market. If you think about where PC margins were, they were higher than we were experiencing in both mobile and in server, primarily because suppliers viewed those as long-term least attractive, and a market that was maybe less sustainable. Therefore, you see larger price swings as the relative supply and demand balance moves over time.
As you think about what does that imply, that implies that as companies like Micron, and presumably our competitors, over time, readjust their supply to rebalance demand, in different segments and pull supply out of that sector, it creates an ASP headwind for us, because we're moving to products that already had lower margin. It creates, over the short term, a more intense competition for those sockets in the high-growth sectors. That doesn't say anything really about what will happen over the longer time horizon.
Okay. I'm still not sure how you're going to be able to recoup this near-term margin headwind. Is it more of a scaling? I'm still confused at how the near-term margin pressure, you're going to be able to overcome it and be able to manage margins.
Well, I think I believe over time that there is, in some of these other market segments, there is opportunity to differentiate and to innovate and to provide stickier solutions. There's also higher growth and longer qualification cycles. What a company like Micron does is they look at, and I think Mark kind of already alluded to this, but I'll re-say it anyway. We're looking to place our product not necessarily in the short term, to pick off the highest margin point, but to make sure we're placing our product where we're going to have growth and stability, and where over time we can drive higher margins. Sometimes you want to be a little bit proactive and out in front of that.
Got it. Quickly on NAND, right now your plans are to build the pilot line, 3D NAND pilot line in Singapore. Then 48-layer stack samples are coming out. My impression is that you're going to add production capacity in Singapore. Your flash partner, Intel, is also very upbeat about the 3D NAND technology and how they are also going to use that technology. When we look into next year and beyond, how is this relationship going to work? Because the JV fabs are located in the U.S., but you're initially going to ramp the capacity in Singapore. How will the transaction going to play out, and when would you actually going to start adding capacity in the JV locations?
Let me correct a couple of things you said just to make sure we're clear on what we're doing. Then maybe I'll have a couple of comments on the Intel relationship, but I think that's something that's better done by the two of us as opposed to just Micron. First of all, relative to what we are doing today, we have, as you know, a significant installed planar base in Singapore. We have available clean room space there to begin the process of converting some of that planar capacity to 3D NAND. We're under construction of an expansion of that facility so that we can continue that transition of planar NAND to 3D NAND into the back half of 2016 and on beyond then. There is also capacity that Micron and Intel own jointly, in Lehi, Utah.
Today, you're correct, that is all planar NAND capacity. One other thing, I think you said 48-layer. We are currently sampling a 32-layer 3D NAND device, and that is the device that we will initially ramp before moving to a second generation of 3D device, starting in the back half of 2016. Relative to Micron's and Intel's plans for production, there's not a whole lot I can say about that today other than to say, it continues to be a very close partnership. We continue to develop the 3D NAND technology together, and we have existing long-term supply agreements with Intel to support planar NAND, and we anticipate that we will build 3D NAND for them as well through an agreement out of Singapore. That's probably all I can say, absent an agreement with Intel to share more information with you.
Got it. Thanks much.
Thank you. Our next question comes from the line of Kevin Cassidy from Stifel.
Thanks for taking my question. On the transition or moving your capacity away from PC DRAM towards the more strategic DRAM
Where are you with that? Have you stopped the transition? What percentage do you think the split will be in the August quarter and going into the November quarter?
Well, I'm going to stop short of giving you actual numbers on allocation of our capacity for a variety of reasons. I will tell you that the process is in place. You can see some of the growth we've had in certain sectors, as we reported that. It's in process. Mobile overall growth rate from a segment and demand standpoint is up 50%, and PC bits are low single digits, 5% range. As we think about that, the shift we're making has to respond to that, and we're on that path. As we've said, we started it in Q3, and we'll continue to balance that out as the markets warrant.
Maybe just as a follow-up, the visibility you have into mobile, there's some concern in the market that you could overshoot and end up with too much capacity in mobile. Can you talk more about your visibility into the mobile market?
I think we talked a few things. As I mentioned in my script, we've actually seen continued growth in content per device. We see that while some people have noticed that the China market starting to mature somewhat, we're starting to see more growth out of India and emerging markets. In general, our visibility is pretty good with the level of configurations both in the mid-range smartphone market. I think those unit growths are something around 90% unit growth for the year and the bit growth rate. Our visibility supports the moving of our capacity towards that, and I think the results of our MBU further reinforces that.
Great. Thank you.
Thank you. Our next question comes from the line of Daniel Amir from Ladenburg.
Thanks a lot. If we look at the Inotera deal from seven, eight months ago when you talked about it, you commented that at that time, if pricing would be the same, would be significantly accretive. How would we look at the deal right now in terms of the pricing at the moment? Do you think something's going to change there in the next six months? Thanks.
Well, obviously, at lower pricing, the accretion from the new deal, which will begin in the beginning of calendar 2016, will be less than it would have been had pricing been flat. We never assumed pricing would be flat throughout 2016. I think that the net takeaway is that there's less accretion from the new deal than there was on the day we signed it. It's still reasonable to talk in terms of the types of numbers that we talked about when we first introduced the new agreement to you.
It's still accretive at pricing that we're looking at today, or is that not the case?
That's absolutely true. Yes.
Okay. All right. Just to follow up here, in terms of the factors here for your NAND costs going forward, should we be, is it really the TLC mix? Is it more the product mix? What should we be looking at and when? You said that you should start seeing in the next few quarters. What are really the milestones that we're looking at that this will really start impacting?
I think, at the risk of repeating our story on NAND, I think we've already demonstrated substantial growth in the mobile NAND segment, our TLC components shipping in Q3 will be the catalyst for our future growth there. As a follow-on, we have a consumer TLC drive coming out in the Q1 of our fiscal year. When you look at those dynamics coupled with the vertical NAND and enterprise progress we're making, we feel that those are still the pillars to our success in improving our operating performance in the overall NAND business.
Okay, great. Thanks a lot.
Thank you. Our next question comes from the line of Monika Garg from Pacific Crest.
Hi. Thanks for taking my question. Mark, you were talking about move from PC DRAM to mobile DRAM has been headwind to the margins. Are the margins now similar in PC DRAM and mobile DRAM, or is the move still a headwind?
You want to take that, Mark Adams?
Sure. Monika, this is Mark Adams. The margins are getting closer, and we always run the risk of trying to explain that through the past 90 days and what's happened. At the beginning of the quarter, they were certainly further away from each other, and today they're approaching each other.
Okay. Thanks. One question on the NAND side, the SBU operating margins being negative for this quarter or last quarter. You've talked about move to TLC, and you also talked about that pricing is benign in the market. When can we see SBU margins to move higher in the positive territory?
I think you'll see that over time, Monika, don't forget that our NAND business is actually segregated across the BU. We sell NAND not only through the SBU, there's a significant piece, as Mark Adams just mentioned, going out of the mobile space. We also said there's significant growth in the embedded space for NAND. At
There's a counter factor, I guess, that you should always be cognizant of, that we sell to Intel through the NAND BU. Sometimes there are lots of facets to our relationship with Intel that have benefits to us that aren't always about selling to them at the highest ASP.
Got it. Thank you.
Thank you. Our next question comes from the line of Timothy Arcuri from Cowen and Company.
Thanks so much. I actually had two questions. First of all, DRAM costs are being guided up low singles, I get the mix issue that you cited with mobile and DDR4, it still seems a little odd given that Inotera is roughly one-third of the bits, that's still based on market minus. Is there something behind the scenes happening with costs? I had a follow-up. Thanks.
The real answer to that is there's nothing behind the scenes other than what normally happens in our industry on startup. You've got some parallel startup paths that are influencing costs along with the mix effect that you mentioned in mobile. You've got DDR4 ramp costs, and in parallel, you've got 20-nanometer startup costs that, when blended, are providing a little bit of a cost headwind in the short term.
Okay, thanks. Just a quick question on CapEx. To hit your full $3.8 billion midpoint, the CapEx has to be like 1.5, 1.6 just in August, which would be like a $6 billion annual run rate headed into fiscal 2016. How should we think about that? Is that like an anomaly in August, or should we think about that as being close to what the right run rate is as you look into fiscal 2016?
Tim, this is Ernie. As we've said earlier, we've been looking at CapEx very carefully as we respond to market conditions, obviously, we still feel that range in the $4 billion or so range for this year makes sense. I wouldn't necessarily imply or project that forward as a run rate into 2016. We're still in the process of working through our plans.
Thank you. Our next question comes from the line of C.J. Muse from Evercore ISI.
Yeah, good afternoon. Thank you for taking my question. I guess first question was hoping you could dig a little bit deeper on your 20-nanometer ramp, both at Inotera and internally, and here really focused on when you expect to be cost competitive with Samsung and Hynix. Then as a follow-on to that, your plans for 1X and your ability to reduce Samsung's time to market advantage. Would love to hear your thoughts.
Yeah. First of all, I think the best way to characterize the state of our 20-nanometer deployment today is we're running at relatively low volumes, but it's going really very well. We've got low volume, but we're driving fast cycle time through the fab. The run-ahead lots and the yields are progressing fabulously, and we feel pretty good about the way that ramp's going, both at Inotera and in Hiroshima. Of course, as we just said, the costs are higher. It takes a while to work through that, not only because you're driving Q yields up, but also because at low volume, you don't have fully loaded tools, and you have a lot of tools that you're installing that you haven't managed to load yet. That process is going to take a while until we are really driving to significant volume.
Now, the way to think about that is in terms of bit crossover, and we think that we'll have bit crossover 20-nanometer versus the other nodes in the first half of calendar 2016. That's probably about as precise as I would want to be about that. Now, for 1X, we think that as we closed the Elpida transaction, we added significant technology development resources to our team. We've been running a higher R&D spend as we ran more resources deployed on multiple technology nodes, and the 1X node will be the first one that comes out as a result of both those teams working together. Yes, we expect to close that gap relative to Samsung.
It's not going to necessarily be directly on top of it. Remember, what Micron's doing right now is we're converting a lot of capacity from 30-nanometer to 20-nanometer, which is a double step. We're going to get a big bang for that as we move throughout 2016. The step to 16-nanometer is smaller to begin with. In particular, if you've looked at some of the early Samsung samples that are floating around, you'll notice that the architecture changes that they've incorporated don't drive nearly the size of die reduction or the bit density increase that you would see going from even 25-nanometer to 30. We absolutely expect to close the gap with Samsung as we move to 16-nanometer, both in timing and in terms of overall deployed bit density in the market.
That's very helpful. I guess as my follow-up, can you talk about within your outlook for DRAM pricing down mid to high single digits, what you're assuming on the mobile side, and if you can comment at all on your positioning within tier 1 handsets into the back half of the year, would love to hear your thoughts.
I tell you again, we don't get in the business of forecasting pricing too much. Generally what I'll say characteristics about the mobile business is that it doesn't exactly behave like the PC business. That tends to be more of a design win type business. You're designing to a device or platform that has long-lasting life cycles relative to interchangeability of PC modules. In light of that, it behaves a little bit more like a predictable pricing model, where there's a sales price and a cost reduction target over through a longer period, maybe six months, maybe 12 months. The pricing behavior in mobile is somewhat different, and we anticipate that to be similar going forward, and it's reflective of our current mobile performance.
Okay, it's fair to characterize mobile declining less than PC?
I think that's fair.
Okay, great. Thanks so much.
Thank you. Our next question comes from the line of John Pitzer from Credit Suisse.
Yeah, good afternoon, guys. Thanks for letting me ask a question. I guess I want to go back to the question that Tim Arcuri asked earlier, just about DRAM cost per bit going up. I would've thought that given a third of your bits come from Inotera, and that's based upon market pricing in arrears, that the price declines we've seen over the last three months would be a big sort of tailwind to cost for you guys in the August quarter. I guess that's wrong, and I guess I'm trying to understand why that's wrong. More importantly, as you think about the headwinds on the cost side in DRAM in the August quarter, do they begin to reverse in the November quarter? How do we start to think about kind of the cost curve in DRAM going back down instead of up?
Let me take the first part, maybe I'll have you restate the second part. Relative to IMI bits, we're early in the ramp, there's not a lot of upsurge in IMI bits yet, but it is coming. I think the effect will be, as you say, with more volume at any given price, we're going to get a larger dollar discount, not necessarily a larger percentage discount. As we move through the rest of 2015, you do have to contemplate that the discount at IMI is also a function of their cash flow. That stands apart and aside from the effect of market pricing. You also have to contemplate that at lower market pricing, a given fixed percent discount is a smaller dollar discount. All those things kind of go in the hopper, John Pitzer.
I'm not sure exactly how you're doing your calculus, those are all effects. We will, obviously, as the 20-nanometer becomes a more significant piece of their output, hopefully see an increase in revenue and an increase in total dollar discount as that happens.
Mark, maybe just to restate the second half of the question was, beyond the August quarter, if all else being equal, given the ramps that are causing headwinds in the August quarter, would you expect cost per bits in DRAM to start trending down? Or do we really have to wait until you get to that big crossover at 20-nanometer you talked about earlier?
Yeah. I think it's going to depend a lot on mix. Unfortunately, I hate to say that, we've got ongoing DDR4, which drives costs up given the bit density per square centimeter silicon. We've got LPDDR4. We've got mobile DRAM and server DRAM, all of which can drive higher costs. Like for like, fair to say, yeah, costs are going to be coming down as we get a little bit further into this 20-nanometer transition.
For my second question, just to Mark Adams. Mark, just given some of these headwinds in the August quarter, it seems like the P&L is more dependent upon just overall pricing. You made some commentary about PC production being below kind of seasonally weak sell-through for the market, and you expect that to reverse. I'm just kind of curious why, what's the catalyst, and importantly, can you talk a little bit about inventory outside of Micron relative to PC DRAM?
Sure. The comments on the PC business were kind of through our Q3 relative to the performance. We do feel that whether it be Microsoft Windows 10 or Skylake or other Intel roadmap benefits from PC and pent-up demand, we just think it should be moderately better than in the first half, and thus have a better effect on the overall market than in the first half of the year. We think PC should perform better in the second half. Relative to the DRAM side, and given the mix shifting away from PCs, we think that combination leads to more stability in the pricing in the back half of the year. You asked about the channel. It's kind of an interesting dynamic. By and large, the OEMs and major suppliers to OEMs, there's one channel partner is heavily invested in a lot of inventory.
Having said that, a number of the other channel suppliers are pretty low, actually seasonally lower than what we say the three- to four-week normal inventory levels. As these things improve and there's balance, we don't think it takes a lot to get more balance back into the compute segment, which should have a very positive effect on the overall business.
Thank you.
With that, I think we have time for about one more question.
Thank you. Our final question for today comes from the line of Steven Fox from Cross Research.
Thanks. Good afternoon. Hopefully, this isn't too repetitive, but just going back on the DRAM gross margins from a different bent. I understand what you're saying about cost per bits with the mix changing, but also as the mix changes beyond the August quarter, shouldn't we expect some sort of rebound in DRAM margins and to what degree? Are we looking at more of a drag from the 20-nanometer ramp? Within that answer, can you just maybe talk a little bit more about DDR4 and how it's going to ramp on the mobile side? Thanks.
We really want to stay away again from predicting margins for you. I think it's fair to say that assuming market dynamics improve, all else equal, we've got cost leverage.
Fair enough. How about just a little more color around DDR4 as it relates to mobile?
Well.
LPDDR4.
Yeah. I think as I said in my script, it's currently a relatively small percentage of overall bits into the mobile sector. We do see that changing out toward the end of calendar year 2015 and into 2016. Today, Low-Power 4 is in the ultra-high end. We feel obviously we're very well-positioned in those relationships. It's just a small piece of the overall mix today, and we think that will grow, and we've got the technology and are in design qualifications at these OEMs for that type of success in that product category.
Great. Thank you very much.
You bet. With that, we would like to thank everyone for participating on the call today. If you will please bear with me, I need to repeat the safe harbor protection language. During the course of this call, we may have made forward-looking statements regarding the company and the industry. These particular forward-looking statements and all other statements that may have been made on the call that are not historical facts are subject to a number of risks and uncertainties, and actual results may differ materially. For information on the important factors that may cause actual results to differ materially, please refer to our filings with the SEC, including the company's most recent 10-Q and 10-K. Thank you.
Thank you. This concludes Micron Technology's third quarter 2015 financial release conference call. You may now disconnect.