Micron Technology, Inc. (MU) Earnings Call Transcript & Summary

Thank you. Welcome back, everybody. Thanks for being with us, and a real special thanks to the Micron -- folks from Micron. Sumit Sadana is with us. He's EVP and the Chief Business Officer. Sumit, welcome. Farhan is with us as well. Everybody knows Sumit and Farhan. Before I begin, I just wanted to say I wish you all the best. Most of our listeners have been listening to other conversations we have had since yesterday. But really, I really appreciate you joining us given what most -- a lot of us are facing in the Bay Area. So welcome. Thank you.

Thank you, Ambrish. Really appreciate it.

And Sumit, maybe before we go into the Q&A, just give us a quick overview. And everybody is very familiar with Micron. But just wanted to hear it from you, from your vantage point, the investment summary and the investment case for Micron.

Sure. So first of all, thanks for having us here and for this opportunity to talk to all of you. And I also wanted to mention that I may be making some forward-looking statements today. So please refer to our SEC filings and risk assessments for the comprehensive view of the context of my comments. In terms of just looking at the big picture, these are very interesting times for sure in the short term. But if I zoom out and look at the big picture, really some of the structural trends in the industry give us a lot of excitement and cause for optimism about the future. We have tremendous changes taking shape in the landscape on an ongoing basis. And starting over the near term, medium term, we see really big growth drivers in 5G, really big growth drivers in AI and machine learning driven applications, especially in the cloud, that are structural 10-, 20-year type of growth opportunities for us. And these are going to spawn a whole host of new architectures like tiered memory, where Micron is uniquely positioned to offer NAND, 3D XPoint and DRAM as the only company in the world with that kind of capability, and so that is tremendously exciting for us as that unfolds over the next few years. And on top of that, we have some company-specific initiatives that we are very excited about, that I'm happy to continue to touch upon, including the transformation of our product portfolio that we believe has come a long way over the last 3 years and certainly technology focus, which has enabled us to do really well on both the DRAM and NAND side, innovation engine in terms of our 3D XPoint capability, dramatic improvements in our cost structure. We have taken out $9 billion of cost and expect to do more over the next few years. Very close relationships with key marquee players in the industry as our customers, and the transformation under Micron -- underway in Micron on the cultural side that we believe will be creating long-lasting results for us and our investors and our customers. So between the industry and disciplined supply, alongside great demand and the Micron story that continues to evolve on a multiyear journey of making Micron a leader in the industry. So I'll just pause there.

Great. Great. And those are some of the very topics I wanted to discuss with you. So it's a great opportunity to have you with us. But if you don't mind, maybe I'll just start off with the near term first. With about 3 weeks left to go in the quarter, Dave had provided us -- provided an update on the business conditions. So a couple of questions. Are there any updates to that given you have a little bit more visibility now? And then how would you characterize the inventory situation, whether at the customers or in the channel?

Sure. That's a great question. So we still have about just over a week to go in our F Q4, and this current quarter has been more back-end loaded than previous quarters, but I have no new update to provide for our current quarter. As we look into F Q1, which is our next quarter, of course, Dave shared some thoughts on that a couple of weeks ago, and I just wanted to build on that. First thing to keep in mind is our F Q1 is a regular 13-week quarter, whereas our F Q4 has been a 14-week quarter. So that's one important piece. The other is, as Dave had mentioned earlier, we have seen some changes in our demand compared to what we had updated our investors at the end of our third fiscal quarter, which is that we are seeing some incremental weakness on the enterprise side, largely stemming from small and medium businesses being weaker due to the economy as well as certain segments of the economy that have not been doing well, like travel and leisure, hospitality, oil and gas, et cetera. Their combined impact on the enterprise space has been an issue in terms of the demand environment for the second half of this calendar year. And of course, there has also been some inventory adjustments that customers have been going through that is important. I think another change that has happened over the last few days, as all of you have seen, is Huawei, which I just wanted to spend a couple of minutes on. As you know, we are continuing to work to understand the details of the latest U.S. orders regarding Huawei, but it is clear that our prior licenses that we had under which we were shipping to Huawei have been impacted by this new ruling. And it seems that most of the semiconductor shipments to Huawei will have to stop after September 14, 2020, unless, of course, the U.S. administration happens to extend or change that deadline. As of now that's the deadline we are operating under. And of course, this is a challenge for the whole industry because the typical cycle time through the fabs and the back-end assembly, packaging and test is about 4 months, including shipping. It's through the whole fab, assembly, packaging, test, logistics, so it takes about 4 months to flush out the entire inventory that was in process. And we were given a 1-month deadline. So clearly, this is an impact. Huawei has been as high as 13-plus percent of our revenue but lately has been just under 10% of our total revenue. So below the peak but still fairly substantial. Of course, we will be able to make up some of that impact by targeting those products towards other customers. But clearly, there will be an impact in F Q1 if that deadline remains. And as we go forward, a smaller impact in F Q2, but we believe that beyond F Q2, we shouldn't really have much of an impact. And obviously, much depends on how the mix changes also occur because any shipments that Huawei doesn't make in terms of smartphones, for example, other smartphone companies will take over some of those volumes in the marketplace. And we have very strong relationships with the rest of the smartphone industry and customers. And so we'll be able to take advantage of that demand. But of course, there is mix-related issues that we have to just go through. So this is something I view as a near-term issue primarily and something that will improve as we go through F Q1, F Q2, F Q1 being the biggest impact, F Q2 being smaller impact, and then it should be okay beyond that. We have also seen actions by the U.S. government around companies like ByteDance and Tencent. We also sell to those companies, of course, nowhere close to the scale of Huawei, and those are things that we are going to manage within course of our business, but certainly, those things add up as well. So I wanted to mention those as some of the changes that have occurred in recent days and weeks. Overall, I think these are short-term impacts. As we get into calendar year '21, we expect a lot of improvement both in the market environment as well as in the development of our own business, which we hope to continually improve beyond that.

Okay. Thank you. That's a lot of good color, Sumit. I just wanted to make sure, in Dave's commentary, he was -- and he was taking into account all these factors that you just mentioned, correct?

I think the Huawei factor is a new one. A lot of what Dave was talking about was really focused on the changes to the market demand environment since our last earnings call. Of course, the impact of Huawei is something that we continue to study. But more of that has come in focus as we have been working to see what the overall profile of shipments is going to be. And clearly, as things stand today, we don't think we will be able to ship to Huawei beyond September 14 because September 15 is the cutoff.

Right. Right. Okay. Okay. Let me -- with that out of the way, let me switch over to maybe start with the medium term. And then I really wanted to focus in on the long-term drivers and also talk about the structural changes in the industry and also at Micron. So maybe just medium term, and you talked about it a little bit heading into calendar '21. How should we think about the supply demand as we exit this year and head into next? This is something we haven't seen ever, what the industry did last year. I think Micron was the first to cut CapEx very aggressively. And then others followed as well, at least publicly. So with that in mind, how quickly the industry reacted and what we're seeing now, how should we think about supply demand as we exit this year, both for DRAM and for NAND?

Yes. It's a great question. So let me share some thoughts on that front. So I'll make my comment based on the calendar year rather than just the fiscal year. And I'll mention, then I'm switching back and forth because our calendar year versus fiscal year tends to be different for sure. So if I look at calendar year '20, we certainly saw the first half of calendar year '20 to be stronger in terms of the demand balance. The second half is somewhat weaker than the first half, and I'll provide some color on the different segments as we go through, and then I'll talk about 2021. So in terms of the cloud demand, again, a very important driver long term, very strong in the first half of calendar year '20. We had some pockets of shortness of supply because of the strength in cloud. In the second half of calendar '20, still healthy demand from the cloud, not as strong perhaps as the first half, but still very healthy. Enterprise, I already mentioned, second half of calendar '20 between the end market demand as well as some of the inventory adjustments that are taking shape. In different end markets, the enterprise demand is weaker in the second half. The consumer and client devices, for example, laptops, notebooks, Chromebooks, very, very strong. More enterprise laptops in the first half of calendar '20 was strong. Second half of calendar '20, more consumer-type notebooks are strong. Chromebooks have been strong throughout, driven by the work from home, et cetera, trends and a lot of shortages in the Chromebook market due to displays and other things have been in short supply. Of course, the Chromebooks and consumer PCs have lower average capacities of memory and storage. That's just one thing to keep in mind. But on the unit volume side, they are doing well. Desktop PCs have been very weak throughout calendar year '20. So just a little bit of color on that front. In terms of mobile, first half calendar '20 was very weak, a lot of declines in mobile, smartphone units around the world. But things have been recovering from that point. Mobile, as you know, is a very important market, and there are some really exciting data points coming out of mobile, not just in terms of the recovery that is taking shape but also in terms of the 5G mix that is happening. So if you look at China, it passed the 50% point in terms of 50% of the handsets being 5G versus all the prior generations, 4G and below. And we passed that point already. And so every month now, there are more 5G handsets being sold in China than all the other generations prior to it put together on a monthly basis. So that momentum should continue to build going forward. Industrial demand is hanging in there, relatively lackluster but should pick up as the economic activity improves. And finally, automotive, that has been the weakest segment in the market for sure in calendar year '20 because it had both a demand and a supply impact as some of the factories shut down around the world. But even those factories are starting to come online and demand is starting to improve, supply is starting to improve. So we are past the weakest point there as well. So in a nutshell, we do expect that as we get into 2021 calendar year, the improvement should continue. And so if I may shift focus to now talk about calendar '21, I am very optimistic about calendar '21, particularly as we go past the early months of seasonally weak time for the consumer right after the whole Thanksgiving, Christmas time frame. Getting to the first quarter tends to be seasonally a little bit slower. But as we get past that, the rest of calendar '21, I believe, will continue to be accelerating to the upside in terms of demand. And there are lots of things to be very excited about, and let me just go through some of those real fast, and happy to dig into any of the details that you like. So first and foremost, 5G. 5G mobile cycle should be super strong next calendar year. We expect about 200 million units of 5G handsets this year in calendar year 2020. We expect that to more than double next year in terms of units. And for us, in terms of memory and storage, the really exciting part is the average capacities are going to be up quite a bit alongside those improvement in 5G handsets, 6-gigabyte minimum DRAM, many handsets with 8 gigabytes, 12 gigabytes, et cetera, of DRAM. And most configurations of Flash, NAND Flash to be in the 128- to 256-gigabyte range. Of course, there will be some 64, some 512, but 128 to 256 will become the sweet spot. And of course, then looking at other segments, cloud growth should be super strong. The economy continues to focus more on cloud computing. More companies are making those choices to move to the cloud, a lot of advantages, a lot of horizontal capabilities with machine learning, functional libraries that they can take advantage of. Automotive and industrial growth should snap back. A lot of average capacity improvements in automotive as well. And if you look at the server platform, 6-channel platforms going to 8-channel with new CPUs, so higher average capacities for servers, and a lot of -- a lot more QLC penetration on the SSD side in servers, so really exciting for us. So bottom line, 2021, as it goes through the calendar year, should see accelerating momentum, and we are very excited about the opportunity there. And overall, industry CapEx seems to be coming in, if you look at the public statements from our competitors as well as equipment companies, relatively constrained. So it should set us up for a good calendar year '21.

Okay. Okay. That's a lot of areas you went into. Between NAND and DRAM, if you just go through each of those, could you -- because for the full year, whatever the expectation was for bit growth, and I think you did not update that in the last earnings call, but can you just help us understand where you expect the industry to land now with all these -- all the memory? My assumption is that all the memory companies will have the same -- similar kind of headwinds that you are facing. So where do you think bit supply and demand lands now with, what, 4, 5 months to go into the year?

So for calendar year '20, we think that the DRAM demand ought to grow in the mid- to high teens range, most likely. We're still looking at the numbers and how they play out for calendar year '20. And I think on the NAND side, less than 30% most likely in terms of bit growth for calendar year '20. Of course, a lot of moving parts, differences between first half and second half, as I was saying earlier, but those are the numbers. And then, of course, like I mentioned, next year, things should be in pretty good shape. We haven't really provided numbers for next year. We will do that in future discussions after our earnings, et cetera.

Okay. Okay. That's fair. So coming back to the long-term drivers, and we really wanted to drill in a little bit more with Micron-specific -- some of the initiatives that you have been working on or you're leading both on the NAND and on the DRAM side to address some of the secular trends that you laid out, which range all the way from consumer side, all the way to data center, AI, clear trends, which I subscribe to as well. So we just -- just help us understand what are the Micron-specific initiatives that will help you differentiate?

Sure. So I think when we look at our longer-term view for Micron, we have 5 important things we are trying to do to set us up for long-term success, and I'll just go into the 5, but I'll try to focus more on the product and technology part. So the first one is product and technology leadership. It's very critical, and I'll talk a little bit more about this in a moment after I touch upon the other 4. The other one, the second one is cost competitiveness. We have spoken about this in the past that we were on a path to take out $9 billion of cost from our cost structure. We have reached that goal already. And so between 2016 and 2019, we did take out $9 billion of cost. And we do expect more improvements on the cost side in the future, of course, less so in fiscal year '21, but '22 and '23, we should see really good acceleration of some of our cost improvements. In terms of the third piece, the third pillar, really innovation, innovation is very, very critical. We have several aspects of innovation across our portfolio. On the NAND side, we do this with QLC, a lot of good QLC products coming out, a lot of good QLC momentum. And of course, our 3D XPoint product, very, very exciting long-term opportunity for us. It's going to be slow to ramp as the whole industry and ecosystem have to really work towards that tiered model of memory and fast storage. But we have been doing some early work with some leading customers and are tremendously excited about what the next 5 years will bring on that front. So that's the innovation part. Customer intimacy is number four, very, very critical to us. We have been building partnerships with some of our top customers in different segments. And then, of course, the last one being talent and culture, which is very key. So really going back to the technology and product leadership, I just wanted to spend 2 minutes on that because we have been doing a lot of work across the company on that. So if you see Micron's DRAM capability, for example, we were trailing competitors on DRAM from a technology perspective, and we became the first company to ramp 1z DRAM node of technology in the marketplace. We were also the first company to come out with MCP products for mobile phones based on LP5 DRAM. And so you can see a lot of areas where our DRAM innovation and product leadership and technology leadership are taking shape. High bandwidth memory, which is specialized product geared towards machine learning and artificial intelligence applications, we were not in the market with that product. And now we have introduced that market. It is -- now introduced that product into the market, and it is something that we are going to be ramping over the next several quarters. And so we are really excited about that. It really rounds out our DRAM portfolio, so really exciting capability there. And switching over to NAND. We have been on this push of high-value solutions. As you know, looking back to 2016 time frame, we had a small minority of our output in NAND going into high-value NAND type of products. And now we have hit the 75% mark as of F Q3, and we do expect to get to the 80% number we had spoken to investors about a couple of years ago. We do expect to get to that 80% high value solutions mark in fiscal year 2021. And of course, a lot of focus on QLC type of capability, a lot of focus on NVMe part of our SSD portfolio. Over the next 6 quarters, we should have several new NVMe products coming to market, which we are very excited about, and that should help us gain share on the SSD side. And we have had a tremendous track record of gaining share on the mobile side using our NAND capability as well as our DRAM capability in MCPs. So a lot of good traction there. So we hope to continue this momentum that we have established over the last 3 years into the next couple of years as well as we make Micron a much bigger force on the product side in the industry.

Yes. Yes. Seems a lot of initiatives. Just one question on NAND, just to remind -- just to calibrate our thinking. This year is a transition year for you. What should we think about cost down once you make the transition on the technology side?

Yes. I mean our cost reductions in NAND will be driven largely by our adoption and ramp of our second-generation replacement gate technology. As you know, we have been ramping our first-generation replacement gate technology and making good progress on the second-generation replacement gate technology. And the products that we ultimately ramp in our second-generation replacement gate are what will drive the cost reductions because, as you know, the first generation was meant to be a much smaller portion of the output because it was a big change from our floating gate technology. So that second-generation replacement gate should ramp over a period of time over fiscal '21. And as that output becomes more material to the overall mix, that's when we'll start to see some of the cost improvements happen on the NAND side.

So the right way to think about is low double digit once you normalize for the second generation being meaningful enough. Is that the right way to think about the numbers?

I think, certainly, when we get to a good portion of our output being in second generation, then we would be able to get to the cost reductions from that technology. I would hesitate to provide specific ranges at this point because we also have to focus on the mix changes that are taking shape in our portfolio, which, once the portfolio stabilizes, those mix shifts are fairly small. But right now, we have fairly significant mix shifts happening because our portfolio had not been so high or mix of high-value solutions. And these higher-value solutions do come at higher costs, but they also come at higher gross margin as well as higher stability in those margins compared to, for example, selling components. So they do have an impact on the cost but also have a positive impact on the gross margins.

Right. Okay. And this was actually going to be a lead-up to my question on the industry side, on the industry profitability, and you touched on it, the structural changes at Micron. So in DRAM, there's been massive consolidation. And the results have shown up in the last cycle that we've seen also in the downturn, including for Micron versus in the past cycles. The situation is a little bit different in NAND. So now we have YMTC making noise on their 128 layer, and it remains to be seen what volume they get to. So if you could just take us -- provide your perspective, from your vantage point, kind of how to think about the structural changes in the industry, both DRAM and NAND. And then you've taken out a lot of cost from the Micron model since this management team has been there. So how should we think about profitability through cycles for -- both for NAND -- for DRAM first and then for NAND, given that at NAND, we've seen the reverse of consolidation.

Yes. So happy to address that. It's a good point. Certainly, on the DRAM part of the business, we have 3 major players in the industry. If we look at other industries that have consolidated over the decades and take lessons from them, I would say that this type of industry structure does provide good strong ROIs. And so we have been seeing good profitability, great ROI out of our DRAM business for sure, and we expect that to continue over time. We are very positive about our DRAM business over time and its ability to generate high ROI for our investors and great value for our customers. I think in terms of NAND, certainly, the industry is a lot more fragmented compared to the DRAM industry. And right now, we have 6 players in the industry who produce in good volume and have good capabilities. We have some -- certainly on the horizon, a newer player, like you mentioned, YMTC. A lot remains to be seen about YMTC in terms of their capabilities, in terms of them being able to deliver quality at scale of manufacturing, reliability, endurance that sustains over time in the field with the products that they build. And so far, certainly, they are -- they seem to be -- if you look at the environment and the marketplace, they seem to be there in some of the really low-end products where these issues are not much of an impact or not easy to discern the quality of their output. But a lot of the test for them as a company to be able to cross those significant hurdles lie ahead and so remains to be seen. We are, of course, always watching, always remain paranoid. And we have in our market models have them growing as a share of the total supply just so that we do our planning with our eyes completely open, wide open. Of course, the industry structure is such that there are some stronger players and some weaker players amongst the 6 existing players. And our goal is to continually improve our NAND capabilities which historically had been weak, but we have come a long way in improving the ROI on the NAND portfolio. And given our focus and progress on the high-value solutions, given our share gains in mobile, on the NAND side, given our improvements in terms of the SSD portfolio as well as our optimism regarding our road map for the next 6 quarters, where we will be introducing a lot of new product and ramping second-generation of replacement gate, we feel good that we will be able to continue to improve the ROI of our NAND business. Of course, ultimately, the returns and the profit structure is heavily dependent on the industry structure as well. And here, if there is consolidation in the NAND industry, as there has been in other industries before like the HDD industry or the DRAM industry, so if NAND does consolidate, it will certainly improve the ROI of that business for all of the remaining participants.

Yes. Yes. Makes sense, Sumit. That's great, great insights. I'm going to pause and take questions from the line. [Operator Instructions] Okay. There is -- there are a few. Maybe we'll start with the first one. And this one is looking to compare your business with NVIDIA. The question is why has NVIDIA had such a higher growth rate than Micron in recent quarters if your DRAM and NAND is embedded in their GPUs going into the data center?

Yes. I think that's certainly a good question. We have been an important partner to them. They are an important customer to us. And certainly, NVIDIA's business is multifaceted. They are into multiple different segments of the market, certainly, data center, of course, gaming, even in automotive products, et cetera. We feel that this partnership with NVIDIA, as well as with other processor platform companies, will really pay dividends for us over the long term because we are in the very, very early phases of growth in the marketplace. And I will also mention that the AI and machine learning type of applications that some of the processor companies ship into, we have had a hole in our portfolio as it relates to high-bandwidth memory in the past years. We have not been present in high-bandwidth memory. It's the principal way in which AI and machine learning applications utilize memory because the traditional way of using either graphics memory or straight DRAM memory is not as efficient for very bandwidth-hungry and latency-sensitive applications that exist in AI and machine learning. So this is a new product for us. We are tremendously excited that we have introduced this product to market, and we are going to be ramping this product going forward. And as we ramp this product and continue to gain the benefits of the partnership that we have because of our leadership in graphics between high bandwidth memory and graphics, memory, where we have leadership already, we believe we'll be really well positioned to take advantage of all of the growth areas that NVIDIA and other processor platform companies, both on the GPU side as well as the CPU side, as well as the custom ASIC solutions that are coming up increasingly in AI applications, we'll be able to work with all of these companies in a much closer way with a much better fit on the portfolio side because of these recent improvements in our portfolio. So we certainly expect going forward for things to be on a good trajectory. Of course, it goes without saying that companies like NVIDIA don't have some of the issues that we have to deal with on the memory and storage side as it relates to some of the supply/demand dynamics in the industry. But as I mentioned earlier, as the demand environment gets past some of these near-term impacts that we discussed and as we get through calendar year '21, we should have accelerating momentum on the demand side as well, which should dramatically improve the environment.

That's, in some ways, not -- maybe not a fair question, but it is. It's a good question because it got out the HBM part of the story. So just to remind us, HBM starts ramping this quarter or the quarter after?

Yes. We are just in the process of starting ramp of that product.

Okay. Good. Thank you for that question. That's actually got us a decent response. The next question is, how has server demand held up? Recently, other analysts have claimed it has slowed considerably. Has GDDR6X begun shipping yet? And then what higher-value products is Micron selling now? And what percentage of revenue comes from higher-value products as opposed to raw chips? Bunch of questions in there.

Bunch of questions in there. Let me see if I can remember all of them, but let me try some of those. So in terms of server demand, we had mentioned that the first half of calendar year '20, the server demand was exceptionally strong, and it was strong for a couple of reasons. Number one, the cloud demand was exceptionally strong. Even before COVID happened, we were expecting a strong first half of calendar year '20. But once COVID happened, the demand really accelerated quite a bit on the cloud side. And again, it was because of some of these trends of e-commerce, which has certainly accelerated because of COVID, work from home, entertain from home, online gaming, all of these things have increased the demands on the data center, especially in the cloud. And so we have seen all of those benefits in the first half. In the second half, server demand continues to be healthy on the cloud side, weaker on the enterprise side for the reasons I mentioned and weaker in the channel as well because the channel is where you see small and medium-sized businesses transact typically. So those aspects have been impacted on the server side but still decent and healthy on the cloud side in the second half. As we go through calendar year '21, as I mentioned, we do expect continued strength. We see cloud growth to continue. We see some of the structural factors driving cloud growth, and we see important transitions happening on the server platform side, where 6-channel memory support on the processor side will go to 8-channel with some of the new CPU platforms coming out in the next several months. And so that should really improve the amount of DRAM that can be connected to processors. And one important thing I just wanted to mention that several of our investors know but important to emphasize that as these new CPU platforms get introduced by all of the CPU companies and GPU companies, et cetera, the number of cores is escalating very rapidly. And so the bandwidth per core is falling as the amount of DRAM that can be connected to these processors is not keeping pace, and so these increased number of channels is very, very important so that all of these additional cores and these new platforms can be fed with memory. Otherwise, what happens is the processor is setting that [ dwindling ]. It comes waiting for data, and that's not terribly helpful to have a superfast processor with lots of cores that can't be fed the data. So the memory subsystem is so critical, and that is why these innovations with graphics, memory and high-bandwidth memory is important because the bottleneck does not become the processor. Ultimately, it shifts to the processor, memory, subsystem. And that's where our ability to differentiate becomes very critical. So as we look ahead, we see more of our memory over the next few years being shifted to high bandwidth memory, graphics memory, cloud-oriented memory, enterprise memory rather than the client memory and client devices, which is more of a commodity type of an approach. So lot of that shift is going to occur on the DRAM side. Another high-value area on the DRAM side is also automotive and industrial memory. We have extended temperature grade, cross temperature qualifications that we do there, longevity for our customers in terms of multi years of shipments from our fab in the U.S. So we provide a lot of value, which is why we are #1 in the world in terms of automotive. And we see automotive as the one segment which will have the highest demand CAGR over the next 5 to 10 years. So really exciting for us that we are already #1 in that segment. And that's also going to be high-value bits for us on the DRAM side. And NAND, I have already discussed in terms of the high-value solutions.

Right, right, right. You gave us a percentage of the high-value which the target you should be able to hit by fiscal '21. But is there a way to answer the question on what percentage of revenues on the DRAM side are high value, if you were to capitalize...?

We don't get into the percentages on the DRAM side. We haven't spoken about that. Suffice to say that our goal has been to improve the portfolio mix. And now with some of the changes that we have finally introduced on the high-bandwidth memory side and some of our continuing innovations on the graphics memory side, we feel very good that our DRAM portfolio has never been stronger in our history. And one other point I will mention in terms of high value, which is, I think, very, very critical, which is this comment I had referenced earlier, which is this move to multi-tier memory, 2-tier memory. And this is where we think that as our compute architectures evolve over the next couple of years, we will have new introductions like CXL, which is a completely new bus, a new way to connect memory and storage over time, will be introduced into these computer architectures. And this will enable us to really take advantage, full advantage of our portfolio of technologies. And this is where the 3D XPoint discussion comes in as well. So as we move to 2-tier memory, as we move to an ability to take advantage of superfast storage, our portfolio of DRAM, 3D XPoint and NAND will stand apart from everyone else in the industry, and we are tremendously excited about those possibilities.

Okay. Let me just go to the next one, which is what percentage of revenues is from Huawei. I think you answered that. You said less than 10% earlier on in your remarks. So we got that, less than 10%. Thank you. And then this one is could we see DRAM growth above 20% next year at -- and also at current spending levels, can -- I think I need higher resolution glasses. At current spending levels, can production growth satisfy that?

Yes. That's a great question. We do think that DRAM bit growth should be higher next calendar year compared to this one. This calendar year, we said it's been -- we think it's going to be in the mid- to high teens. We do expect that it should be towards the higher end of the range, if not slightly higher. We think that the 2 important -- 3 important drivers are the ramp in mobile 5G, the continued growth in AI and cloud computing and the bounce-back in automotive demand, which has increasingly higher attach rates for DRAM going forward. And on top of all of that, on the server platform side, I mentioned the processer transitions to 8-channel from 6-channel. So I think it's shaping up really well for calendar year '21, '22 to be having a tailwind in the industry. Of course, we just need to get through the next couple of quarters, and we look forward to getting much improved momentum beyond that.

Great. Great. Thank you very much. Let me just -- I think there's one more in the queue. I was wrong. I think we got them all. Thank -- Sumit, thank you very much. Thank you, Farhan. And good luck on the front that we were talking about. We wish you all the best. Thanks again for joining.

Thank you so much, Ambrish. Really appreciate you. Thank you.

Thanks, Ambrish.

Thank you. Bye.