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

Thank you for standing by, and welcome to Micron Technology's Post-Earnings Analyst Conference Call. [Operator Instructions] Please be advised that today's call may be recorded. [Operator Instructions] I would now like to hand the call over to Farhan Ahmad, Head of Investor Relations.

Thank you, Latif. Welcome to Micron Technology's Fiscal Second Quarter 2020 Sell-Side Analyst Callback. On the call with me today are Sumit Sadana, our Chief Business Officer and Interim CFO; and Manish Bhatia, our EVP of Global Operations. Today's discussion of financial results is presented on a non-GAAP financial basis, unless otherwise specified. A reconciliation of GAAP to non-GAAP financial measures may be found on our website. 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 filed with the SEC, specifically our most recent Form 10-K and 10-Q for a discussion of risks that may affect our future results. Latif, we can now open the call for Q&A.

[Operator Instructions] Our first question comes from Aaron Rakers of Wells Fargo.

Two, if I can, real quick. The first question is, I guess, last quarter, you talked a little bit about the visibility you have in terms of long-term agreements. I think it was 75% of your revenue under some kind of contractual longer-term agreements. Could you provide us any kind of updated commentary on that? Has that changed? Do you see that possibly further extending in terms of contributions going forward? And then the second question real quick is on the NVMe enterprise SSD ramp. You guys still have a relatively smaller share in enterprise SSDs. And so I'm curious of how much breadth you've seen in terms of qualification cycle at this point? How many hyperscale customers do you have? Or just how we should think about the trajectory of that going forward?

Yes. So I can provide some color on that. So Aaron, in terms of your first question regarding the LTAs, we have been working on this topic of LTAs for 5 years almost, and when we started, less than 20% of our business was under LTAs and now over 75% is under LTA. And just to reiterate, these LTAs are quarterly volume agreements going out typically 4 quarters. And when we say greater than 75% revenue, keep in mind there is a certain part of our company business like the Crucial business, for example, that is very consumer-based business. There are certain parts that we sell through distributors to a whole range of smaller customers. And so those types of revenue cannot practically be under any kind of an LTA. And consequently, the way to think about the 75% is -- sorry, greater than 75% LTA is that practically all of the large customers, the substantial majority of our large customers with whom we do business are now under LTAs. So based on that, we have pretty good visibility. And even though these agreements are not take or pay type of agreements, they are written documented agreements, and consequently, both sides try to do their best to live up to them. And of course, our approach has been that when the industry conditions are tight, we'll preferentially provide some supply in those type situations to customers who have gone above and beyond to keep up their end of the bargain in terms of working with us on these agreements and living by these agreements. And so there is a very natural feedback loop that exists. And I feel like there is a really strong relationship with customers to try and make this work in a way that, I believe, over time will reduce the volatility, improve the planning, improve the visibility. And particularly as we have, I believe, created this approach in the industry, this happened well before COVID hit. So we got to these very high levels even before COVID happened. So while the rest of the industry thought about longer-term deals, we had already gotten to these very high levels. And so we believe that as the rest of the industry increases the amount of LTAs, it just has a dampening effect on the volatility. It reduces the ups and downs. Everyone plans better and it just becomes a really positive thing for the industry. So that's why it's a very exciting development for us and it has paid huge dividends. And so don't expect that number to increase in any material way because substantially all of our large business is done under LTAs now. In terms of the second question on NVMe. Yes, our data center NVMe SSD share is low right now. We have very high share in SATA SSD, but really the data center market focus is NVMe, and we have obviously been working on that for some time. And we have introduced 2 products in quick succession, the 7400, which is our first vertically integrated data center NVMe SSD, and then the 176 layer NAND-based 7450, which is our second-generation vertically integrated product that literally was introduced just a few months after the first one. And so you can already see the benefit of the platform investments that we have made that we were able to bring out the latest technology node. So this now becomes the first 176 layer vertically integrated NVMe SSD in the world. And it is a very robust platform. It has exceptionally good QOS capability, quality of service capability, compared to competitor drives. And we are seeing significant interest in this drive across a lot of the hyperscale customers in different geographies. And so we feel that this product is going to be a really strong product for us. Already, we are seeing that benefit. And so Sanjay mentioned in his prepared remarks that the data center SSD revenue had doubled year-on-year in FQ2. And as we go through the rest of this calendar year, we expect our data center revenue to continue to improve as a percent of our overall NAND business. And that should have a positive impact on gross margins because this is by far the most positive gross margin business in our portfolio.

Our next question comes from Harlan Sur of JPMorgan.

Congratulations on the strong execution. Two questions from me as well. So in DRAM, can you guys just true us up on where you see the current level of channel and customer inventories by end markets, data center, PCs, mobile and embedded and your expectations as we progress through the year? And then second question, on the strong cloud and enterprise data center demand dynamics over the past few quarters and also looking forward and then you combine that with your solid server DRAM portfolio, is server DRAM now the largest part of your DRAM mix surpassing mobile?

All right. So I would -- let me just answer the first question first. So in terms of the inventories at customers by end market, let me just make some high-level comments first. So our customers generally have their own inventory strategies that differ from customer to customer. So there isn't like a segment-wide strategy that we think about every customer does its own thing. And so that's one piece of it. Having said that, by and large, what we are seeing is our customers have become very concerned about inventory getting their revenue, getting their own market share. And so they have gone from this just in time type of inventory approach to being more prudent about the level of inventory they need to carry increasing the amount of inventory they need to carry just to overcome some of the macroeconomic challenges that occur trade-related challenges that occur, COVID-related challenges that occur, and now disruptions coming even from geopolitical events like the tragic situation we have going on with the Russian innovation of Ukraine. So I think based on all of that, customers generally have been carrying higher levels of inventory than pre-COVID times. And I don't feel like that inventory is going to go down to the pre-COVID times. Certainly not until a time when everything is clear and no one is concerned about COVID anymore and people are not concerned about shortages, people are not concerned about supply chain challenges, not concerned about trade-related issues, not concerned about geopolitical events. So your guess is as good as mine as to when we get to that regime. But suffice to say, it's a pretty safe bet that it won't be anytime soon. And so we feel like customers have been largely carrying inventory at a higher level based on that. But outside of that comment, I would not say that we are losing sleep on inventory-related issues. There are customers who have been making corrections to their inventory when it tends to get out of hand. So late last year, late calendar 2021, we mentioned to you that the PC customers, several PC customers were making reductions in their memory and storage inventory in the last few months of calendar '21, and that has -- that is behind us. And so we have seen pretty robust demand from our PC customers. And yes, the unit volume expectation is flat, but it's a big mix change that is positive towards corporate sort of laptops. So that helps. If you look at the mobile customers, there has been some level of weakness in mobile demand in China. And consequently, some Chinese customers have started to make some of their inventory reductions. We saw some of that happen in our FQ2. So there are customers who will make these changes over time, and we don't lose sleep over that, because over time, these things tend to get normalized. And because of these LTAs, we are able to work with these customers and have it not be terribly disruptive to the extent that, that can practically be worked out. And so look at the overall demand trends, we feel like the demand trends are pretty constructive. And hence, we feel that the inventory is being staged to meet the demand.

Got it. And then my question on server DRAM, is that the biggest part of your DRAM mix now?

Yes. Server DRAM is the largest part of our DRAM. It is now larger than mobile, no doubt. And of course, if you look at the overall compute business, that, of course, is what we report in our compute networking business unit is, depending on the quarter, 40%, 45% of our revenue or mostly DRAM. And so you can see that, that is a very big and growing part of the business. So it's a very critical business. And we had also mentioned to you that the data center TAM has now become the largest TAM in the market, right? So this is just a reflection of that.

Our next question comes from Tom O'Malley of Barclays.

I just wanted to ask on the cost side. You had mentioned that you expect cost to outpace the industry now. And particularly on the NAND side, I think you had previously said that the industry was kind of in a mid-teens cost reduction. Has there been any change to the outlook of what the industry can do from a cost reduction perspective? Or do you think that, that holds still and you guys can get better cost reductions than the mid-teens?

Manish will talk about this.

Yes. So Tom, thanks. No, I think that's right. Our long-term production for the industry is in the mid-teens. And yes, we do think that this year, we're able to outpace that industry average on the front-end cost for NAND. We've done a really good job with our -- as Sanjay highlighted on the call, with our 176 layer ramp already representing the majority of our bits and ramping across the entire portfolio of our products, now that we're launching it in data center as well. We're already seeing those cost benefits in the client side and in the mobile side. And the rest of the industry will have some challenges with even reaching that kind of long-term average given the incident that's happened at one of our competitors at [indiscernible], as well as some public comments from others about their pace of technology cadence this year, in particular. So I think we feel really good about where we are with our 176 layer, and we still have a little bit more room to run on 176 layer in terms of penetration here before we get to our next node that will come after that, as Sanjay mentioned, in calendar year '23.

Got you. And then just one follow-up. Just on the cadence of bit growth in NAND, you're still matching the industry about the 30% growth rate. But it implies very strong growth in the back half of the fiscal year. Can you just talk about, one, are you seeing an elevated jump up obviously by your guide and the fact that you said NAND is growing faster, you're seeing a big step up in May? But can you talk about the linearity between May and August, obviously, both numbers need to be pretty big to get to that industry growth rate? Any color on the linearity would be really helpful.

Yes. So in terms of the growth rate, of course, we have mentioned the 30% on NAND and mid- to high-teens on DRAM as a calendar year sort of a number, not so much fiscal year sort of a number. So definitely, if you just look at the rest of the calendar year, which goes out through the end of our fiscal Q1, that's the expectation that we have. So I would not try to slice this into a fiscal year sort of a thing. But having said that, I do believe that we will have a decent growth in bit shipments in FQ3. We have already given you the guidance for FQ3 in terms of revenue. The additional color I can provide on the FQ3 is that we expect that the NAND bit shipments will be twice as fast on a sequential growth percentage basis compared to DRAM in FQ3. And then we are not guiding FQ4, but I will just say that we do expect there to be a bit shipment growth sequentially happening between FQ3 and FQ4 as well. And that is consistent with the strong second fiscal half commentary that we have provided to you in terms of a step-up from the first half to the second half. And then when you take it through fiscal Q1 that ends in November, that's when we expect to get to -- for the full year, which includes December, so first month of fiscal Q2 as well to that approximately 30% on NAND and mid- to high-teens on DRAM. So I just wanted to make sure that it's a calendar year sort of commentary.

Our next question comes from Sidney Ho of Deutsche Bank.

I have 2 as well. The first one is if I look at your fiscal third quarter guidance, you said gross margin is going to improve for both. But specifically on the DRAM side, with more muted front-end cost reduction you talked about, but still expected gross margin to grow, are you assuming DRAM ASP is going to be up, I think, it just pointed as that. But I know you don't talk about that. But is there any factor that like product mix that we should consider when we do that math? And then I have a follow-up question.

Yes. It's a good question. I mean, we did not intentionally provide any pricing guidance for DRAM. Now I will say that the overall pricing environment is strictly benign for DRAM. Of course, a lot more positive for NAND for FQ3. But DRAM pricing is also constructive. But you can tell a little bit the math from our gross margin guidance. I gave you a little bit of color in terms of the ups and downs for Q3, some of the 100 basis point plus headwinds that we are seeing that still end up with a positive midpoint sequential growth from FQ2 to FQ3. And so I would just say, very modest growth in DRAM gross margin sequentially and then a stronger growth of NAND gross margin sequentially in FQ3. So that's the best I can provide right now.

Okay. That's helpful. Maybe my follow-up is, considers some of the cost improvement on the DRAM side, but you also talked about some higher costs that are coming in. How should we think about the full year cost improvement for both DRAM and NAND in calendar '22, maybe compared to what that to what you expected a quarter ago? Again, what is the mix impact on some of these things that might impact the cost improvement side?

Manish, do you want to say something?

I think our -- if we try to lay out, Sidney on the Fab side, the front end side, our execution on -- the question was about DRAM, but we actually sit on both DRAM and NAND on our leading 1-alpha for DRAM 176 layer for NAND has been really strong. And so our outlook for both of those is better for the fiscal year than -- we were more confident in what we are able to deliver from a cost base on the front end than we were before. The ramps are going better than planned. However, we do continue to see a lot of costs, some of which are in the front end with specs of the chemicals and gases that we are having to pay premiums for, as well as primarily the back end where we have COVID mitigation costs, as well as other component expedites that we're paying for that kind of bring us back to the kind of cost reduction range that we had been expecting before. We still think our cost reduction range for the year is very competitive on a like-for-like basis. And then the only other thing on the cost that I'll say is we do have a pretty strong mix shift going on towards high-value products like SSDs or data center SSDs during the back half of the year, and same in DRAM with more module and DDR5 server modules, which have a higher BOM cost as well. So those are just a couple of things that are kind of working in the opposite direction of the strong execution on our front-end costs in both DRAM and NAND this year.

[indiscernible] just to qualify that.

Sorry, just to build on something that Manish said that's very important. So we do have these mix changes happening that Manish highlighted, both on the SSD side in NAND due to SSDs and in DRAM due to some of these DDR5, et cetera, modules. Just keep in mind that even though the cost impact of this mix is to increase cost because they are just higher content type of products, they have a positive benefit to us on the gross margin side. So the ASPs of these products goes up more than the cost goes up, and that's a positive for our gross margin. So that's why when we talk about our cost performance and we very much believe that we are in a leadership position this year with our cost performance versus the industry, it's on a like-for-like basis because the things that impact mix are gross margin accretive anyway. So we tend to not create confusion with that sort of commentary. So just to make sure that's very clear.

Our next question comes from Raji Gill of Needham & Company.

Just another follow-up question on the margin. So in terms of the sequential improvement, the revenue is going up about over $900 million, and the gross profit is going up about $400 million on an absolute dollar basis. So that's a little under 50% fall through. You talked about the negative 100 basis point impact relating to bonus accruals and the kind of unfavorable mix shift to NAND. But wondering, as you kind of progress throughout the calendar year as volume continues to ramp, should we start to see kind of a higher gross profit fall through, as you mentioned, driven by the better shift in NAND as well as a higher volume? And is the bonus -- I would assume that's temporary for fiscal 3Q or is that going to potentially continue throughout the year?

Yes. So that's an important question. So let me just address that. So in terms of the bonus accrual, we obviously forecast our profitability for the year, business based on what we think we are going to achieve. So when we have to -- when we see a higher profitability forecast for the year since we accrued in FQ1 at a lower rate and FQ2 at a higher rate, we have to do a catch-up accrual in FQ2. So the accrual ends up being higher than what the rate is. Now of course, a lot depends on what the trajectory in future quarters is for our profitability assessment. But there is a slightly higher accrual that happens. Now we booked that accrual, for example, in expenses already in that FQ2, but it flows into inventory and then into costs in FQ3. So that's why the impact in gross margin is felt in FQ3. So I just wanted to make sure I emphasize that. Having said that, I don't think of this NAND growth as like a permanent trajectory of constantly increasing NAND mix that will have impact on our consolidated gross margin. There are going to be quarters in which NAND grows faster and has this impact on consolidated gross margin, but there are quarters in which DRAM will grow faster and the gross margin will benefit from that. But I think the thing that is important from an ROI perspective is at least in FQ3, and we shouldn't lose sight of that, both NAND and DRAM gross margins are expected to increase sequentially, and that is critical data point. And I think just looking out in terms of how we look at the DRAM business, there will be quarters that DRAM business will grow faster, and we certainly see that will happen. So I think overall, the gross margin performance is on a very good trajectory, and we feel really good about how we are doing there. So there will be certain quarters where the gross margin pull-through will be higher than what you are seeing in FQ3 in the bottom line.

And just sticking on the gross margin, if I can, Manish and Farhan, you talked about that some of the cost improvements got pulled into fiscal 2Q. Can you just remind us of why that happened? Why was it pulled in fiscal 2Q? And you're kind of reiterating your long-term cost reduction is kind of in this mid-teen level, and this year, it will get outpaced by the NAND side. So it paints kind of a very favorable picture on the cost structure side as you continue to kind of ramp with 1-alpha and 176, and then you have maybe a positive mix shift both within NAND and overall. So maybe you can just sort of provide some clarity on those points as well.

Sure. So I think we also said that on the DRAM side, we'll also outpace the industry. So I want to make sure we're clear on that. And again, the 1-alpha is doing very well. Really what it is, is that there's improvements in costs that we -- improvement in output and yield on both of those 2 nodes from the Fabs, in general, but on those 2 nodes in particular, really helped us with our costs throughout the first half of the year. While we achieved those, get close to maturity somewhat earlier, look at the -- and then they'll help your overall year's cost position. But on a sequential basis, when you think about how much cost decline we have from FQ2 to FQ3, that's what we're trying to say that we won't see quite as much cost reduction sequentially because we get better than expected. And obviously, once you get to -- as you get closer and closer to maturity, the incremental impact of costs on these leading nodes is limited.

So you saw most of the cost reductions in fiscal 2Q?

I don't say most, it's just we did better than expected. And now that we're getting closer to maturity or close to maturity on both of these, there's not as much more cost reduction to go when we look sequentially.

Got it.

Right? But when you think about it on a year-over-year basis, we expect at least on the front-end costs led by both of these 1-alpha for DRAM and 176 layer for NAND. We feel better about where we are on an FY over FY basis now than we were projecting last quarter.

And just the other -- the only other thing I will add is this is how things just go. I mean, we ramp a node, then we start ramping another node and that drives cost reductions. So if you just look at it from an annual perspective, we still expect to be very strong, very competitive even in 2023 and in future years because there will be nodes of technology we will ramp, and of course, we expect to retain our industry leadership and technology as well.

Our next question comes from Srini Pajjuri of SMBC Nikko.

Two questions for me as well. First, on the inventory, Sumit, I see why you guys are building raw materials and work in progress inventory. Just curious, have you noticed any meaningful change in your customer behavior in terms of how their inventory is being procured after the, I guess, the recent -- the war started in Russia, Ukraine. I mean, are you seeing any more interest in terms of building some additional buffer since the war started?

Yes. I mean, I think the war is just another example for our customers that we continue to be in an extraordinarily uncertain environment. There are these uncertainties that come from just every different direction it seems. So there is these geopolitical uncertainties and then there are trade-related uncertainties there are uncertainties related to COVID shutdowns in different parts of the world. And then there are uncertainties just coming from just tightness of component supply due to demand overwhelming supply in several parts of the supply chain. And these are different for different segments of the market, of course. So this is just another reminder for our customers that carrying extremely lean inventories is dangerous in an environment where it can start to gauge revenue. And so I would not say that this particular event has created any new behavior that we see. It is just reinforcing to them that they cannot let their guard down in terms of how they manage their inventory because of all of the overall uncertainty.

Okay. Got it. That makes sense. And then a question about the mobile market. I think you reported 4% year-on-year growth in that. I'm just curious because on one hand, the content is increasing 50%, and you also said your expectation for unit growth for this year, for the smart phone market is somewhere in the mid- to high single digits. So year-on-year, it only grew 4%. So is it because of the allocation issue? Or did you see any inventory correction in that market? Or is it any market share dynamics there? Any color would be, I think, helpful.

Sure. Yes. I mean, that's right. I mean, in terms of our smartphone demand, it is in the low single digits in terms of units for calendar '22. So I just wanted to mention that not mid- to high single digits, but low single digits. But having said that, you are absolutely right that the content growth is meaningful. And when we look at bit growth in this market, definitely the bit growth for the calendar year, but even with these assumptions, is pretty decent. So we have been focusing on significant growth and increase in mobile share over the last couple of years. So we ramped up a lot of our bits allocated to the mobile business over the last few years, last 3 years or so. And that paid dividends. We gained a lot of share in that business, and our DRAM share is at very steady levels, NAND share climbed, just like Sanjay mentioned on the call, from low single digits to high teens. And so what we are now focused on is getting the ROI on a lot of our investments in SSD platforms and also continuing to drive growth in automotive. So there is now a lot of internal competition for the supply that Manish's team produces from the Fabs. We have very strong capability on the automotive side, very strong capability on industrial side, very strong SSD portfolio and continuing strength in our mobile capabilities. And this is how we like it because now we can do better on the portfolio optimization front because we have so many opportunities and we focus on the strategic relationships. We focus on long-term goals in the segments. And we can do better in terms of how we manage to maximize the profit pool that we go after based on an optimal mix of different segments that we target. So I think we are allocating -- fair to say, we are allocating preferentially more to drive the growth of our business on the SSD side. But we are not really taking the foot off the accelerator too much on the mobile side. There are always quarter-to-quarter variations. As I mentioned earlier, we do expect the mobile business growth to pick up when the overall product cycle again drive seasonality and things like that. So quarter-to-quarter, there will always be fluctuations. We do expect in a couple of quarters, you will see much better growth on the mobile side, but we have been definitely driving more growth on the storage side.

Our next question comes from Nik Todorov of Longbow Research.

Two questions, if I can. One, you mentioned that you're seeing some pockets of semi component markets being -- starting to improve, and I think your guidance kind of implies that your starting to see some improvement in NAND controllers. But you also said that some areas are being more constrained than you expected. Can you give us some details on where do you see improvement? Where things are more constrained. And on the constrained side, which end market do you expect to be more impacted for a more prolonged time?

Sure. I can take that one. Nik, we're starting to see some improvements, and that's why you're seeing us being able to achieve strong growth in SSD and forecast that going forward. Because obviously, the SSDs, a lot of these analog and other components, just generally higher BOM content there. Same thing for DRAM modules. And as we talked about in DDR5, we were able to execute and kind of lead the market in the DDR5 transitions. So these are a couple of areas where we've been able to secure more components, and we are hopeful that we will continue to see those things continue here as we go through the back half of our fiscal year and towards the end of the calendar year. There are still another area that I think we've seen some of our assembly materials, we see things improving in that regard, as well substrates and printed circuit boards. Those are getting a little bit better. But other areas that are continuing to be constrained. We talked about assembly and test capacity and services at the OSAT. Logic foundry is still very tight across multiple different nodes, some legacy nodes, some leading nodes and logic foundry looks to be tied into '23 as well. So something we're working very closely with our suppliers on there. And analog, we see pockets improving and others, that are still a challenge. So just kind of a color across a few different areas.

Okay. Very helpful. And if I can follow up. There's been some data points that just some of the new CPU architectures are going to be delayed. So I just wonder, do you still expect server DDR5 to be, I think you mentioned about 20% of server DRAM mix exiting 2022. Has that expectation changed from your perspective?

Yes. I mean, I think definitely, the DDR5 ramp is heavily dependent on how the processor schedules align to customer deployments. And so we are not going to comment on the timing of the platform availability of some of these new CPU architectures. Suffice to say, we are going to be very adaptable to how the situation changes and evolves if it does change and evolve. And so we are working very closely with our ecosystem partners, the CPU vendors, our customers. And so if the CPU ship on time, then those numbers are very viable. If they don't, then we will adjust.

Our next question comes from Steven Fox of Fox Advisors.

Just on the inflation pressures you're seeing away from the 100 basis points. So you called out OSAT, the gases' logic, et cetera, you're not breaking that number out. It's within the overall cost reductions on the nodes. I'm just wondering, one, if you would put a little bit more color on what that number could be. And then at what point do you think about not absorbing it within your own cost structure and maybe having capacity on to customers?

So I'll turn it to Manish to answer some of that. But I just want to point out one important thing. I didn't quite understand exactly what you said about the 100 basis points. But I wanted to point out that the 100 basis points comment that I made was related to the variable compensation accrual, which is the bonus that we accrue for our employees and also to the NAND mix and not related to the inflation. So yes, you might even -- we didn't provide inflation-related impact that is correct. I just wanted to clarify that nothing to do with the 100 basis points. So that -- all right. So yes. So Manish, any...

So I don't think we're going to break out the detail for you, Steven, but I can give you a little bit more color on sort of how it flows. We're seeing some of the assembly cost increases that I mentioned, whether we're paying for additional or higher prices for assembly and test services or if we're paying for some of the assembly materials or the components that go there, whether those are controllers or our analog components. Those tend to make it into our results a little bit earlier because they are integrated closer to the time of revenue. So those tend to make it in earlier. Some of the other costs, like we mentioned, noble gases where we're having to pay premiums or even some of the energy pressures that everyone around the world is feeling. Those will -- as those come in, those will be longer-term dynamics because those will flow through our COGS on our balance sheet before they -- flow through COGS to inventory on our balance sheet before they get into COGS. So those impacts will be felt a little bit later. So not really ready to kind of break that out. In terms of your second half of your question on working with customers, certainly, we're making a lot of investments. We are definitely paying premiums to make sure that we can provide continuity to our customers. We are making investments, as we've mentioned several times throughout this pandemic over the last 2 years, in our assembly and test footprint to provide flexibility to be able to build products in different locations so that if we are impacted by COVID, as we were in this last quarter in Xian, that we're able to have the flexibility and the capacity to be able to move production builds to other sites and maintain that continuity to our customers. And we are making, of course, investments in the leading edge technologies. And across all of these, whether you look at materials or inflationary pressures, whether you look at the investments in capital and leading edge technologies or whether you look at the investments in assembly test capacity to be able to provide that continuous supply, we are working closely with our customers. And I think they are recognizing the value of working with a top-tier supplier like Micron, and hopefully, we'll be able to see that in the returns we get on those investments.

And ladies and gentlemen, that does conclude today's conference call. Thank you for participating. You may now disconnect.