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 fourth quarter 2017 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 would like to ask a question during this time, please press star, then the number 1 on your telephone keypad. If you would like to withdraw your question, please press the pound key. Thank you. It is now my pleasure to turn the floor over to your host, Shanye Hudson. You may begin your conference.
Thank you, Karen, and welcome to Micron Technology's fourth fiscal quarter 2017 financial conference call. On the call with me today are Sanjay Mehrotra, President and CEO, and Ernie Maddock, Chief Financial Officer. Today's call will be approximately 60 minutes in length. This call, including audio and slides, is also being webcast from our investor relations website at investors.micron.com. In addition, our website contains the earnings press release, which was filed a short while ago. Today's discussion of financial results will be presented on a non-GAAP financial basis unless otherwise specified. Comparison to prior year non-GAAP financial results excludes stock-based compensation and the amortization of acquisition-related intangibles. A reconciliation of GAAP to non-GAAP financial measures may be found on our website, along with a convertible debt and capped call dilution table.
As a reminder, the prepared remarks from this call and webcast replay will be available on our website later today. 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. As a reminder, the matters we will be discussing today include forward-looking statements. These forward-looking statements are subject to risks and uncertainties that may cause actual results to differ materially from statements made today. We refer you to the documents we file with the SEC, specifically our most recent Form 10-K and Form 10-Q, for a discussion of risks that may affect our future results. 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 today's date to conform these statements to actual results. With that, I'll now turn the call over to you, Sanjay.
Thank you, Shanye. Good afternoon, everyone. Our fourth quarter results accentuate an unprecedented year for the company. I thank the Micron global team for maintaining intense focus on our key priorities in delivering outstanding results. Our fourth quarter revenue was $6.14 billion, with record gross margin, operating income, and free cash flow. Full-year revenue, profitability, and free cash flow also set company records. Our results were driven by favorable industry fundamentals and solid execution in deploying our next-generation lower-cost technologies and diversifying our product portfolio toward a richer mix of differentiated high-value solutions. We are excited about future opportunities as customers increasingly recognize the strategic value of our memory and storage solutions across a range of high-growth markets. Now I will share details from each of our business units, followed by our perspectives on industry dynamics and an outline of our corporate strategy.
In our Compute and Networking Business Unit, we saw robust growth in Q4 revenue and profitability compared with the prior year. Our results were driven by strong demand in cloud and graphics, complemented by a healthy pricing environment. Revenue growth from these two segments significantly exceeded overall CNBU growth, which more than doubled compared with the year-ago quarter. Cloud sales are supported by increasing DRAM content per server, which is up nearly 50% versus a year ago. In graphics, we continue to leverage our industry-leading GDDR5 and GDDR5X performance to address strong demand, primarily from gaming. The business unit is also benefiting from the initial ramp of our first-generation 1x 8-Gigabit DDR4 product, which was sold primarily into the client and cloud segments.
In fiscal Q1, we anticipate continued growth of our 1x portfolio, coincident with the ramp of our second generation 1x 8-Gigabit DDR4 and GDDR5 products, both of which have already been validated at certain partners and customers. We also received initial customer qualifications on our TSV stacked DDR4 products, enabling modules with up to 128 GB and the highest speeds supported on industry-standard server platforms. These products address the growing demand for analytics and in-memory databases in both the enterprise and cloud segments. Fourth quarter revenues in our Mobile Business Unit were driven by a favorable pricing environment and significant growth in our eMCP business. Due to strong execution, sales from our mobile NAND and eMCP solutions nearly doubled year-over-year. We believe that increased DRAM and flash capacities in flagship smartphones will continue, due in part to new applications such as augmented reality in mobile devices.
Our roadmap of new LP DRAM, discrete managed NAND, and eMCP offerings position us well to address these market requirements. In fiscal Q4, we achieved our first 1x LPDRAM qualification at a major mobile OEM and have several others underway. Also, our technology capabilities in 1x LPDRAM Package on Package products allow us to offer cost-effective, high-capacity mobile solutions ranging from 3 GB to 8 GB. We expect volume shipments of these new products in fiscal 2018, following successful customer qualifications. During the fourth quarter, we also qualified our first 3D TLC eMCP and eMMC solutions at a major chipset vendor and now have dozens of high-density products in qualification with several OEMs. We expect production shipments to start later in 2017. Our 64-layer 3D TLC UFS products will also start OEM qualifications later in 2017, enabling us to participate in the mobile market's highest density designs.
The storage business unit recorded a revenue increase of 71% in Q4 compared with the prior year quarter, supported by strong demand for our SSD product portfolio. Late in the fourth quarter, we identified and corrected a flash component issue on select TLC 3D NAND products. We paused shipments of affected products as we worked to implement a solution to the issue, which appeared only under a narrow set of performance conditions. As a result, our SSD revenue declined sequentially during the quarter. Shipments have now restarted, and we expect to resume solid sequential SSD revenue growth in Q1. We continued to garner positive momentum with our SSD products across a broad range of customers. Our flagship SATA 5100 SSD has been qualified at enterprise server OEMs, cloud service providers, and Fortune 500 companies.
Demand for our client SSDs is also strong, with Micron shipping solutions to most leading PC OEMs. We see healthy demand trends for SSDs moving forward. Client SSD attach rates continue to increase, and although storage density growth has slowed temporarily due to a tight pricing environment, we foresee longer-term demands for higher density SSDs. We made substantial progress in growing our relationships and our business with cloud and hyperscale customers in fiscal 2017. Cloud data center customers are seeking innovative tailored to their workloads. Micron's unique capabilities and expertise in DRAM, 3D NAND, and emerging memory technologies make us a compelling partner for these customers. Our embedded business unit delivered strong performance, growing revenue 39% for the full year. We strengthened our leadership position in automotive in fiscal 2017, with growth driven by increasing connectivity and electronics content in vehicles. Automotive applications continue to require leading-edge performance.
As a result, we have seen significant ramp of our 20 nanometer DDR and LPDDR technologies this quarter and began sampling automotive grade 1X DRAM to meet these needs. The growth in edge analytics in both industrial and consumer connected home applications led to record quarterly revenues in both segments. We saw strong growth through the year of our NAND and LPDDR MCP products, driven by form factor and performance needs in applications like machine-to-machine communications, surveillance, drones, and home automation. Turning to Micron's technology progress, our 1X DRAM and our 64-layer NAND production rollout is proceeding on plan, and we expect to achieve mature yields in both technologies before the end of calendar 2017. We are pleased with our 1Y DRAM technology progress and are focused on the late stages of technology and product development.
Our third generation 3D NAND development is also proceeding well, with production expected to commence later in 2018. This latest generation technology continues to utilize Micron's industry-leading CMOS under the array architecture, which yields smaller die sizes. We have made significant progress in our technology development and volume ramp execution. We see meaningful opportunities to further shorten the cadence of new technology node introductions, accelerate new technologies into volume production, upgrade our fab infrastructure, and expand our captive assembly operations. Through successful execution, we expect to narrow our technology cost gap and optimize bit output growth in both DRAM and NAND with a disciplined focus on profitable growth. Our fiscal year 2018 CapEx plan targets achieving these objectives through technology migrations with no new wafer capacity. Ernie will discuss our CapEx plans in further detail later in the call.
Our ability to successfully execute our technology transition plans will be a key enabler of our cost reduction and supply bit growth capability in the foreseeable future. Moving on to the demand and supply fundamentals, we expect the industry to remain moderately undersupplied for the rest of 2017 for both DRAM and NAND. We see DRAM industry supply bit growth of about 20% in calendar 2017 and expect it to grow at relatively similar levels in calendar 2018. The DRAM industry supply-demand balance is expected to stay healthy throughout calendar 2018, driven in part by ongoing strength in data center and cloud computing trends. We expect Micron's fiscal 2018 DRAM bit output growth to be slightly below the industry growth rate.
Our bit growth is supported by our 1x DRAM ramp, which represented mid-teens % of our DRAM bit output in Q4 and will grow throughout the next several quarters to achieve bit output crossover as we exit calendar year 2018. We expect industry NAND bit supply growth to finish calendar 2017 in the high 30% range. At these levels, supply remains below demand, which has created a constrained environment. As the industry continues to transition to 64-layer 3D NAND, we estimate industry bit supply growth in calendar 2018 will approach the 50% range, which should better satisfy the current unfulfilled demand. We expect that Micron's ongoing transition to 64-layer 3D NAND in fiscal 2018 will result in bit output growth that is somewhat higher than the industry range.
In fiscal Q4, 64-layer NAND represented mid-teens % of our trade NAND bit output, and we expect to achieve bit output crossover during the second half of our fiscal 2018. The dynamic industry transition to 3D NAND is taking place in the context of a NAND market that has consistently exhibited demand elasticity. We expect this behavior to continue for the foreseeable future as higher density SSD solutions increasingly displace HDDs in client computing, cloud data centers, and enterprise environments, as average capacities continue to grow with more performance-sensitive, storage-hungry devices and applications in mobile and other end markets. These trends support our view that NAND demand drivers will remain healthy into 2018. As I begin my first new fiscal year as CEO, I would like to outline our strategic priorities.
First, we are focused on driving our cost competitiveness to best-in-class levels, primarily by accelerating the percentage of our output on leading-edge technology in both DRAM and NAND. Second, we will drive execution excellence, delivering solutions to customers quickly, predictably, and in line with their product launch windows. Third, we will accelerate our transition to high-value solutions. We intend to lead the industry in deploying disruptive memory and storage solutions. Fourth, we will leverage the full breadth of our capabilities to develop deeper collaboration and partnerships with marquee customers, maximizing our value in the market. Finally, we are strengthening our focus on our teams, investing in the best talent, and driving a winning culture. We believe our diligent emphasis on the speed and urgency with which we execute these strategic priorities will have a transformative effect on our market competitiveness and financial performance.
I look forward to sharing the results of our progress with you in the year ahead. I'll now turn it over to Ernie, who will walk through the specifics of our financial performance this quarter.
Thank you, Sanjay, and thanks for joining the call today. Our solid operational execution and favorable market dynamics resulted in excellent financial performance in our fourth fiscal quarter and the full year. For the full fiscal year, we achieved record revenue of $20.3 billion, up 64% from the prior year. We narrowed the technology cost gap with our competitors by successfully executing our technology migration plans. Our efforts resulted in strong levels of DRAM and NAND bit output growth for fiscal year 2017, enhancing our competitiveness while maintaining our wafer output. I'll now provide some further details on Q4 results, starting with a breakdown by technology and business unit. DRAM represented 66% of our Q4 revenue, with the following segmentation.
Server held steady at approximately 30%, mobile was 20%, specialty DRAM, which includes networking, graphics, automotive, and other embedded technologies, remained in the mid-20% range, and PC was also in the mid-20% range. Our trade NAND revenue represented 30% of total revenue. The three segments comprised of SSDs, mobile, which includes managed NAND and discrete solutions, and MCPs, and automotive, industrial, and other embedded applications, each represented approximately 20% of our quarterly trade NAND revenue. The remaining 40% of our trade NAND was made up primarily of component sales to partners and customers. Turning to performance by business unit, the Compute and Networking Business Unit reported record FQ4 revenue of $2.8 billion, more than doubling on a year-over-year basis. The growth was supported by strong demand from cloud customers who are architecting data centers and computing capabilities that enable them to execute their specific strategies.
Our successful conversion to 20 nanometer DRAM production and the initial ramp of 1X DRAM output boosted Q4 CNBU operating income to $1.6 billion, or 56% of revenue, up 31% compared with Q3. The mobile business unit delivered its highest-ever revenue quarter at $1.2 billion, up 76% from the year ago quarter. Mobile operating income also set a record at $364 million, or 31% of revenue. Our results are due in part to the positive progress we've made in qualifying our mobile NAND solutions, and we expect this momentum to continue in fiscal 2018. The storage business unit reported fiscal Q4 revenue of $1.3 billion, up 71% on a year-over-year basis. Revenue and operating income were slightly lower quarter-over-quarter due to the NAND component issue that Sanjay discussed earlier.
With this issue behind us, we are focused on leveraging the progress we've made in penetrating the SSD markets over the past year. We estimate that our global SSD market share nearly doubled in fiscal 2017, enabling record fiscal year revenue for the storage business unit. Our operating income for the fourth quarter was $250 million, or 19% of revenue, compared with a loss of $57 million in the same period last year. The embedded business unit also achieved record performance in fiscal Q4, with revenue of $827 million, up 18% sequentially and 61% on a year-over-year basis. Growth in the quarter was driven by solid sales for consumer applications, which include home automation. Automotive also remains a key revenue driver for this business, as a shift towards smarter and connected cars is driving increased memory content.
These trends bode well for Micron as we continue to capture new design wins and strengthen our leadership position. Operating income for EBU was $348 million in Q4, or 42% of revenue, more than doubling year-over-year. Moving on to overall company results, revenue for the fourth fiscal quarter was $6.1 billion, up 10% from last quarter and 91% on a year-over-year basis. Sales of server and SSD solutions each were more than three times higher than year-ago levels, reflecting our focus toward a higher value add revenue mix. Non-GAAP gross margin was 51%, up three percentage points from Q3, and 33 points from the fourth quarter of fiscal 2016. The improvement reflects the successful adoption of products based on our advanced technologies, combined with a healthy industry pricing environment. Non-GAAP net income was $2.4 billion, or $2.02 per share.
For the full fiscal year, we achieved non-GAAP net income of $5.6 billion, or $4.96 per share, compared with $273 million, or $0.26 per share for fiscal 2016. Turning to results by product line, DRAM revenue more than doubled on a year-over-year basis and increased 13% sequentially. The sequential results reflect a 5% increase in bit shipments. DRAM non-GAAP gross margins for the fourth quarter was 59%, up 39 percentage points from year-ago levels, and up five points sequentially, benefiting from an 8% increase in ASPs. We are seeing the benefits of execution on our technology migration plans and the continued strong market environment. Trade NAND revenue increased 81% on a year-over-year basis. Sales were 8% higher quarter-over-quarter, supported by demand for mobile and embedded segments, and a 5% increase in ASPs.
Trade NAND non-GAAP gross margin for the quarter was 40%, up 24 percentage points from a year-ago, and down one percentage point sequentially as a result of the NAND-related issues mentioned earlier. NAND bit shipments increased sequentially by 3% during the quarter. Our fiscal year results reflect strong adoption of our 3D TLC NAND products in a strong market environment. As we consider the ongoing progress of the business, as well as the competitive environment, we plan on making a few changes to our disclosures, beginning with FQ1 2018. Specifically, we will be eliminating the presentation of changes in cost per bit and market segmentation detail for each technology. The evolution of our business to higher value add solutions, which often have higher BOM costs and higher margin opportunities, makes cost per bit comparison a less reliable indicator of our progress.
Relative to the market segmentation by technology, we will continue to provide qualitative color through our business unit reporting. Non-GAAP operating expenses for the quarter were $601 million, essentially flat from the prior quarter. The company generated operating cash flow of $3.2 billion in fiscal Q4, compared to $896 million in the year ago period. During the quarter, we deployed $1.5 billion for capital expenditures, net of partner contributions, and $5.1 billion for the full fiscal year. DRAM investments were between 40%-45% of the full-year spend, and non-volatile memory was around 30%. Free cash flow for the quarter was $1.7 billion, and for the full year was $3.3 billion, compared to negative $2.7 billion in fiscal 2016. In fiscal 2017, we deployed $1.6 billion or approximately 50% of our free cash flow for de-levering activities. The result of these activities represents approximately $0.07 of annualized EPS accretion.
We ended the fourth quarter with cash, marketable investments, and restricted cash of approximately $6.2 billion. Turning to more near-term matters, as we have been discussing for some time, our two key priorities for cash flow are accelerating the company's cost competitiveness and improving our financial foundation through reducing leverage. Our fiscal 2018 plan enables both priorities. We currently expect our fiscal 2018 CapEx, net of partner contributions, to be in the range of $7.5 billion ±5%. Our investments will be focused on technology transition and product enablement. Generally, we expect that between 35%-45% of CapEx will be deployed for DRAM, 35%-45% for non-volatile memory, and the remainder for technology and product enablement. There are no wafer capacity additions planned for fiscal 2018. We will continue to target our free cash flow generation for the opportunistic retirement of debt.
We see the opportunity to reach our interim target of $8 billion-$9 billion of gross debt during fiscal 2018. These actions, together with the progress that we've made in fiscal 2017, will drive annualized EPS improvement of between $0.18-$0.23. We also see the potential to be net cash positive as we exit fiscal 2018. Moving on to guidance for fiscal Q1 2018. On a non-GAAP basis, we expect the following: Revenue in the range of $6.1 billion-$6.5 billion. Gross margin in the range of 50%-54%. Operating expenses between $575 million-$625 million. Operating income ranging between $2.65 billion-$2.85 billion. EPS ranging between $2.09-$2.23 per share based on 1,191 million diluted shares. With that, I'll turn it back to Sanjay.
Thank you, Ernie. As part of our strategic planning process, Micron developed a new vision statement that embodies how we see the opportunities in front of us. As we close out one year and look to the next, I would like to now share it with you. Our vision, transforming how the world uses information to enrich life, captures the tremendous potential Micron possesses. New technologies like artificial intelligence will change the world in ways we can barely imagine today. Fast data access and high-performance data analytics will be at the heart of those transformations, making memory and storage core to the data-centric world that is taking shape in front of our eyes.
I believe our strategy to tighten our focus, accelerate our technology and product development, and strengthen our presence in critical markets will make Micron an increasingly prominent player in the industry as these revolutionary new technologies take shape. Fiscal 2017 was a record year for us, but I'm confident that the best is yet to come for Micron. We will now open for questions.
Thank you. Ladies and gentlemen, if you have a question at this time, please press star followed by the number 1 key on your telephone keypad. If your question has been answered or if you'd like to remove your line from the queue, you may press the pound key. Our first question comes from the line of John Pitzer with Credit Suisse. Please go ahead, sir.
Yeah, good afternoon, guys. Can you hear me?
Yes.
Yeah. Ernie, Sanjay, congratulations on the strong results. I guess, Ernie, my first question just with the CapEx number of $7.5 billion. That doesn't surprise me as much as your sort of guidance around bit growth for DRAM and NAND for the fiscal year, just given the significant jump in CapEx. I'm wondering if you could just help me understand to what extent is this being driven by a back-end loaded expectation around CapEx, and how do we think about the bit growth exiting next fiscal year? I guess at what point do you think you begin to start to outgrow bits relative to the industry for both DRAM and NAND?
Yeah. Multi-part question. Let me try and break it down in an easily understandable way. First, as you pointed out, there is a time lag between the time CapEx is invested and the time that the bit growth is realized. Certainly, if you look at what we did in 2017, that was actually a function of the investments that we made in 2016. It's also important to remember that the bit growth that we were able to realize from the Inotera investments were not part of Micron's CapEx plan then. If you had Micron spend and Inotera spend, you're probably in the upper $6 billion range, and that's a good point of comparison as we now look at 2018.
In 2018, we would expect the majority of the bits that are the results of the CapEx to come on toward the latter part of the fiscal year. You heard Sanjay in his prepared remarks say that we would be at bit crossover on an output basis for our 64-layer 3D NAND around the second or third fiscal quarter of the year, but it was going to be the exit part of the calendar year for DRAM before we hit bit output crossover on 1x DRAM. That gives you some sense of the timing and the impacts of timing on those investments. I would also tell you that our objective over a multi-year period is to grow at about industry levels. Certainly, we think this investment plan, over the course of a multi-year horizon, will allow us to do that.
Obviously, being above in 2017 fiscal, as we've just reported, DRAM at the low end, little above on NAND. I think looking at snapshot comparisons across our fiscal year is interesting, but really important is the statement that we intend to grow aligned with industry over the course of these multi-year periods. I think that's probably the best broad perspective I can provide. Happy to address any specific question.
Of course, our focus also would be on high-value solutions so that our revenue share outperforms our bit share.
That's helpful, guys. Sanjay, this is my follow-on question. I think clearly the most significant concern in the investment community is the sustainability of the current environment. I'd love to get your thoughts on sustainability. I guess specifically, to what extent in an environment like this can you move the business from more sort of a transactional hand-to-mouth business to something that might have more backlog and visibility with your customers, especially as you try to move the mix towards more higher value end solutions?
In terms of the market environment, we are certainly excited about the demand requirements for DRAM as well as for NAND. As I mentioned in my prepared remarks, big growth driver for DRAM, certainly that's outpacing the average growth of the industry, is in the area of server and cloud. Here we are making great penetration with the hyperscale customers in terms of driving the growth of the DRAM business. We remain very bullish about the DRAM market environment through 2017. We think it'll be undersupplied, given the demand trends, we think we'll have healthy demand supply balance in DRAM throughout 2018 timeframe as well.
In terms of NAND, as is well known, that average capacities are increasing, certainly in mobile devices, but even more importantly, SSDs are displacing HDDs at a rapid pace, with attach rate continuing to be projected to be going up over the course of next several quarters. Of course, there's a strong value proposition for SSDs in the cloud and hyperscale data center environment as well, given all the trends of artificial intelligence, machine learning, all of this is driving big data analytics. All these trends related to artificial intelligence, big growth in data, customers wanting to offer differentiated value to their end customers, all of this is driving need for memory and storage solutions. Overall, we remain pretty bullish about the demand trends. Look at DRAM as well as NAND. Even in autonomous vehicles, the demand requirements for flash, data is being generated.
Much data is being generated by autonomous vehicles that it requires fast processing, both within the vehicle as well as on the cloud. I think demand trends for the foreseeable future continue to be strong, and that bodes well for our industry. In terms of your question regarding customers and some of the customers wanting to engage in longer-term requirements, yes, that is absolutely happening, and we do consider that based on various customers. It depends on the nature of the customer's requirements. Really cannot get into the details of that here in this call. Certainly, our business includes customers that are more transactional in nature, that have business more on a monthly transaction basis, some that are more on a quarterly basis, and certainly, certain customers that are also involved in longer-term trends. I think customers are just seeing increasing value of memory and storage.
DRAM and flash is becoming strategic to our customers. Our customers are seeing Micron as being uniquely positioned with having a strong portfolio of DRAM as well as flash, being the only company in the Western Hemisphere with those capabilities, and that is definitely making us an attractive and valued partner to our customers.
Thank you, guys. Congratulations again.
Thanks. Thank you.
Thank you. Our next question comes from the line of Rajvindra Gill with Needham & Company.
Yes, thank you, and I echo the congratulations. A question on the NAND bit growth. The bit growth is increasing to 50% from the high 30s this year. A lot of this obviously being driven by 3D NAND. I wanted to talk about, in terms of how you're seeing the supply-demand balance going into next year in the NAND environment, and your thoughts around pricing as we start to have more supply, and the impact also to cost per bit.
In terms of going into 2018, we see healthy industry demand and supply balance for NAND. You are right to note that the bit growth is going up because of the technology transition in the industry to the 64-layer technology. When we look at the demand trends, those demand trends continue to be strong, as I just pointed out, related to SSDs as well as increasing average capacities of Flash in mobile devices and all kinds of other devices. Demand continues to be strong. We see healthy trends in that regard in 2018 timeframe. Regarding pricing, we don't specifically, for competitive reasons, provide comments on pricing on the call.
We would just like to point out that we believe that a healthy industry environment is one where price decline is less than or equal to cost declines, and we are certainly focused on aggressively reducing our product costs with realizing successfully our technology production ramp of 64 layers.
Thanks. Just my follow-up, Sanjay, one of the positive trends that has been happening at Micron is the diversification of the end markets for DRAM. You can just compare this cycle, say, to previous cycles, maybe not just too long ago, maybe only about three years ago, where PCs were a higher percentage of DRAM, and now they are mid-20% range today. Can you talk about that phenomenon, and you mentioned it in your prepared remarks, but the diversification of the end markets in DRAM to specialty, particularly in graphics and automotive and server, how that is affecting the business model, how that is affecting the customer relationships and engagements going forward. Thank you.
Certainly the diversification of the end markets for DRAM absolutely bodes well for the future health of our business. We have enjoyed the benefits of that. Calendar year 2017, as you noted, is a great example. You just pointed out the mix of our DRAM business between PC, what used to be just about PC and mobile, is now very much about PC, mobile, server, automotive and multitude of markets. Micron has really great presence with a whole slew of customers and channels. This really bodes well, plays very well to the strength of Micron. It has really, for a long time, enjoyed a diversified set of global customers and great presence in channels, and that is coming into full play as the demand requirements for DRAM continue to grow nicely and into multiple mega markets.
Thank you.
Thank you. Our next question comes from the line of Mark Newman with Bernstein.
Yeah. Hi, thanks for taking my question. Congrats again on the great results. My question's about the technology migration. You talked about 64-layer, second crossover, I believe second half FY 2018. Can you talk about what's next on NAND flash 96-layer or whatever it is, and what the timing is? Will Micron look at transitioning to a charge trap flash alternative at 96-layer or perhaps one after that? Then similarly for DRAM, what is the plan for 1y and timing for EUV?
With respect to NAND, in terms of our third-generation 3D NAND, we are not yet disclosing the number of layers in that technology. As I said in my prepared remarks, we are continuing to make good progress with that, we plan to be introducing that technology in the calendar 2018 timeframe, and continuing to deploy CMOS under array technology that continues to provide Micron a die-size leadership position, which is hugely attractive cost point of view. With respect to DRAM in terms of 1y node, we will be introducing that node also in calendar 2018 timeframe. Beyond that, we are not providing any specific details for our technology at 10 related to competitive basis. Your question regarding floating gate, we have a strong roadmap of future technologies related to floating gate.
Just a quick follow-up, if I may. When do you think you will narrow the gap or catch up potentially with Samsung in both the 3D NAND and DRAM? You've alluded before to closing the gap and narrowing the gap to zero, hopefully eventually. I don't think it's happened yet. Is there an update on your thinking of when that could be?
As we have indicated before, that in terms of technology cost position as well as the technology node readiness, in recent times, over the course of last couple of years, Micron has lagged the competitors in terms of getting advanced technology ready and at par with them, and deploying those technologies into volume production. However, in recent times, Micron has made very good progress in this area, we are getting the benefit of that as we are ramping those technologies into production. I've said before that these kind of undertakings, driving accelerated deployment of new technology nodes into volume production and continuing to narrow the gap on the cost front is a multi-year phenomena, we have made very good progress in this regard. I fully expect us to continue to make good progress in fiscal year 2018 as well.
We are, of course, very much focused on continuing to accelerate the timeline of our future technologies into production well-positioned to ramp those technologies into volume production as well. Along with this, of course, remain very much focused on driving a greater mix of high-value solutions, both in DRAM and NAND as well. These are really two very important pillars of our strategy, driving cost competitiveness and driving greater mix of high-value solutions. These things don't happen overnight. They will continue to be strong growth opportunities for us going forward over the course of next few years.
Thanks so much.
Thank you. Our next question comes from the line of Steven Fox with Cross Research.
Thanks. Good afternoon. Two quick questions from me. First of all, you mentioned that as demand, your bit growth accelerates for NAND next year, that you expect to capture some of the pent-up demand that the NAND market has seen. I guess I was curious if you can talk about what gives you confidence that pent-up demand hasn't actually turned into demand destruction this year, why it'll still be there. Secondly, if I look at the CapEx breakdown that you provided, it looks like about $1.5 billion or so is dedicated towards product enablement, and I was just curious how that number compares to maybe a year ago and if there's anything you'd say it's most focused on. Thanks.
I will let Ernie comment on the CapEx. On the demand side, as I pointed out earlier, it's not that this demand is perishable. This demand, in terms of the trend of SSDs replacing HDDs in client notebook computers, where the attach rate continues to increase. In 2017, attach rate of SSDs to PCs is around 35%. That attach rate over the course of next few years continues to grow to around 50% in 2018, and by 2020 timeframe, expected to go to around 75%. These demand trends are secular in nature. It's the same thing on the enterprise side and on the cloud side, that the attach rate of SSDs as well as the average capacity requirements on a per-server basis continue to go up as well. These are really very solid secular trends here that are long-term in nature.
Of course, the trends of mobile devices adding new rich features such as augmented reality, such as rich displays, all of these are trends that also continue to drive higher average capacities in mobile devices. I feel very good about the demand trends on the NAND side.
Just to follow up on the CapEx. If you look year-over-year, we just reported the totality of fiscal 2017, you would get a reasonably similar number within the ranges that we provided. We would expect on a hard dollar basis, perhaps $100 million-$200 million more in that enablement and technology piece of the business, and that's predominantly related to the centers of excellence that we've been talking about relative to wanting to consolidate a lot of our back-end operations very close to our front-end operations.
Great.
Some of the product enablement CapEx is related to back-end captive assembly operations, which will help us improve our cost position going forward. Of course, also, there is CapEx associated with upgrading of the infrastructure that is needed to realize the technology transitions.
Great. Thank you so much.
Thank you. Our next question comes to the line of Chris Danely with Citigroup.
Hey, thanks guys. Can you just give us a little color on the, I guess the three main end market server, mobile, and PC on tightness or relative tightness and how it trended during the quarter, and any information, either qualitative or quantitative you have on inventory levels out there at the channel or at customers, just on DRAM.
Sure, Chris. I think as has been the case for a few quarters now, there's actually a fair amount of tightness across those three channels that you mentioned. It would be hard to distinguish one from the other, relative to any nuances there. We don't think there's an inventory issue. Certainly, if you actually take a look at inventory levels that are reported, which are typically financial numbers, and adjust those for dollar cost of how pricing has changed over the course of the year, you do get a bit of a different perspective on inventory as reported by a variety of customers and channel partners in those areas. The environment continues to be strong.
The supply-demand circumstances continue to be fairly tight, and we're working very closely with our customers to make sure that we stay in close sync with them as they think about their plans going forward.
Great, thanks. For my follow-up question for Sanjay, by the way, welcome to the DRAM party. You were not afraid to engage in M&A at your previous shop. Can you talk about the willingness or appetite for M&A at this point for Micron versus the desires to continue to improve the balance sheet?
We do not rule out M&A in the future. Right now, our focus is on the priorities that I mentioned that include cost competitiveness and strengthening the high-value mix in our revenue. Of course, if and when we were to engage in M&A, our focus would, of course, be to try to strengthen our future opportunities and make it an opportunity that absolutely provides a strong ROI. Beyond that, I would not speculate on this matter.
Okay, thanks, guys.
Thank you. Our next question comes from the line of Karl Ackerman with Cowen and Company.
Hi, good morning, gentlemen. Ernie and Sanjay, I wanted to ask more of a strategic question. Two of your competitors in the NAND industry have been at disagreement on the trajectory of their manufacturing partnership. Are you and your existing NAND partner inclined to abstain from future partnerships? I have a follow-up, please.
I don't think the experience that I think you're referring to colors our perspective on partnerships, that all partnerships are a very individualized, partner-specific set of activities. Certainly, while we continue to become informed, just as you do, our color would be very specific to the circumstances at hand vis-a-vis any potential partnerships or the continuation of any partnerships.
I think we have a long history of valued partnerships here and knowing how to make the partnerships work well. Partnerships are important, absolutely.
Understood. Appreciate the color. For my final question, this was somewhat addressed earlier in the call, I just want to move back to just some of the content growth that we have been talking about. Our field work during the quarter indicated that DRAM fulfillment rates, at least in the server market, improved from last quarter, but they were still well below 100%. Are you seeing more active engagement today with tier 2 customers for longer-term contract agreements that gives you greater confidence in your capacity planning assumptions?
We are definitely seeing strong demand trends from the entire spectrum of our customers. Large, what you would call tier 1 or tier 2, as you termed, although all customers are important to us, and we do engage meaningfully with them, and we work closely with them to understand their demand requirements, and we apply our own judgment to their demand requirements as well in terms of assessing the overall industry demand trends. Based on those demand understanding on behalf of our customers, we project healthy industry demand supply environment in DRAM and NAND.
Thank you.
Thank you. Our next question comes from the line of Vijay Rakesh with Mizuho.
Thanks, guys. Congratulations on a great quarter. Just looking on the DRAM side, you mentioned the crossover of 1x nanometer. When do you expect to see that? If you can give some color on what the cost reductions are as you transition the mix to 1X.
Sure. We had talked about being at bit output crossover for 1x DRAM by toward the end of the calendar year of 2018. As we have previously shared, for the 1x node, we see somewhere in a roughly 45% increase in bits per wafer versus 20 nanometer and about a 20%-slightly more than 20% cost reduction on a cost per bit.
Great. Got it. On the NAND side, obviously pretty nice numbers on the storage business side, up to 21%. Could you parse out how much is enterprise mix for you on NAND? Obviously, you guys are seeing some good traction there. Where do you see enterprise mix exiting next year, let's say? Thanks.
We said that our SSD mix was about 20% of our NAND revenue, and that consists of our sales to client customers as well as to the hyperscale cloud and enterprise customers. Both are roughly about the same in both of those categories. Beyond that, we don't provide further breakdown.
All right. Thanks.
Thank you. Our next question comes from the line of Tristan Gerra with Baird.
Hi, good afternoon. Looking at your production cost reduction timeline, is it fair to assume that there's going to be some slowdown as you reach a crossover for 64-layer in NAND, meaning that after that, there could be a little bit of a slowdown in terms of your ability to reduce production costs? If you could reiterate the production cost reduction targets for NAND and DRAM for 2018.
Sure. I think that in terms of Gen 3 NAND, we'll certainly talk more about that as we are able to, in terms of what we might expect and whatever is appropriate for that. It is true, just on a mathematical basis, that as you progress up layer count on an absolute basis, there are less incremental bits, and there is an association between cost reduction and bit growth opportunity. In terms of providing specifics, it's very early to share that with you. If you look at 64-layer versus the first generation, it's somewhere in the realm, as we've talked about before, of two times the density and roughly, in aggregate, a 30% reduction in bit cost.
In general, the technology complexity increases with subsequent technology generations, both in DRAM and NAND. Also, given the increased technology complexity, gigabytes or gigabits that you gain in NAND as well as DRAM on a per wafer basis tends to decline with advanced technology nodes. All of those factors play a role in terms of cost reduction capabilities as well going forward.
Okay, great. A quick second question. You talked about the adoption rate trajectory of SSDs in notebooks. Could you also give us your expectation both now and, say, in a year or two, in terms of the adoption rate for SSDs in a data center and traditional enterprise server?
Sure, I can do that. I just want to add a comment to the previous question on cost. My comment was related to cost on a per wafer, gigabytes per wafer basis from one technology node to the next. Of course, cost also depends very much on how you deploy those technologies into volume production. This is one of our focus areas in terms of deploying advanced technologies faster into production. Specific to your second question regarding the attach rates in enterprise and server market. SSD attach rate is around 50% there in terms of on a SSD per unit basis, and opportunity there is greater. Average capacities are definitely moving fairly fast. In fact, enterprise and data center is one of the fastest growth segments for flash in terms of year-over-year bit demand increases that are projected.
Average capacities in enterprise and data center for SSDs are over 3 terabytes. That's the average capacity, and that trend continues to increase by some projections tripling almost to 9 terabytes by 2020 timeframe. As I was saying earlier, these demand trends for increasing attach rate of SSDs in client and data center cloud computing applications, as well as increases in average capacities are secular trends.
Great. Thank you very much.
Thank you. Our final question comes from the line of Mark Delaney with Goldman Sachs.
Yes, good afternoon. Thanks for taking the questions. First question is a follow-up on CapEx. I was hoping you could clarify the comment about no new wafer capacity additions. I know you said that explicitly for DRAM, but can you clarify if that also applies to NAND flash? On that topic, if you could help us understand to what extent you may need to invest in new clean room capacity as part of that CapEx guidance next year. Obviously, even without adding new wafer capacity, the amount of factory floor space is going up for both one X nanometer and more so for 3D NAND. I know the company had some available, but just wondering if any of the spending relates to clean room capacity due to the floor space requirements.
Yeah, Mark. Certainly the statement about no new wafer capacity in fiscal year 2018 applies equally to both DRAM and NAND. Relative to your second question, there is not anything material relative to the CapEx guide that we shared that relates to construction costs or whatnot. As you pointed out, there is obviously incremental clean room space available or could be available at pretty low cost, but it would not be material in course of the overall guide that we provided.
Okay. If I could just follow up specifically on smartphone demand trends, if you could touch on what the company's seen in terms of end demand for some of the flagship models specifically, but also if you could talk about some of the Chinese domestic OEMs, if you're starting to see any pickup in demand trends for memory from those customers.
I think in the flagship models, to answer your specific question, the demand for DRAM as well as for NAND, average capacities continues to go up. Of course, the mix of these high-end smartphones also as a part of the total smartphone market continues to go up. Specifically to DRAM, average capacities of DRAM in high-end smartphones going from somewhere over three gigabytes in 2017, projected to go up to over five gigabytes, closer to six gigabytes by 2020 timeframe. Specifically to NAND, average capacities of NAND in high-end smartphones in 2017, somewhere around 70 gigabytes, let's say, projected to double or more than double by 2020 timeframe as well. Again, the average capacity increase trends in smartphones continues to be solid, not only actually in high-end smartphones, but in value segment of the smartphone market as well.
Thank you.
Thank you. That concludes our question and answer session for today. I'd like to turn the floor back over to Micron for any closing comments.
Thank you. That concludes our call today. We really appreciate your support. As a reminder, we will be posting the prepared remarks from today's call, as well as a webcast replay on our website later this afternoon.
Ladies and gentlemen, thank you for your participation in today's conference. This does conclude the program, and you may now disconnect. Everyone, have a great day.