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

Good morning, and welcome to the Americas Select Conference. I'm Tom O'Malley, semiconductor analyst here at Barclays. Pleased to host Micron. And from the company, we have Dave Zinsner, CFO. Dave, welcome, and thanks for joining.

Thanks, Tom. Appreciate you having me.

Awesome. So why don't we start out. Given some in the audience may not be as familiar with the company, I thought a good way to kind of start out would be spend a few minutes on Micron, the progress the company has made over the past several years [indiscernible] technology leadership perspective? And what are the goals the company has for the next couple of years here?

Okay. Great. Yes. Good question, and thanks, everyone, for your interest in Micron. So Micron, we're the fourth largest semiconductor company in the world as measured by revenue. If you look in the United States, we're the second largest in the United States. We participate in really 2 product areas, DRAM and NAND, which together, at this point, represent about 30% of the semiconductor market. So a big portion of the market is memory and storage. We've got a pretty diverse manufacturing footprint with facilities in Taiwan and Japan and a leading NAND facility in Singapore. And we, at this point now, try to drive for technology leadership in both DRAM and NAND. So we used to be -- if you look back kind of 5, 10 years ago, Micron was always kind of a laggard or a follower of the competition in terms of technology. And when Sanjay Mehrotra came aboard as the CEO, we really went through a pretty big transformation as a company to try to drive towards a leadership position in both markets, DRAM and NAND. And we did that in a lot of ways by just accelerating our pace of technology transitions. And then kind of investing more on the back end in terms of assembly and test to be more -- to have a more integrated model. And I'd say this is relatively simple in concept, but difficult in execution. And I think what Sanjay and the team he assembled really brought was a real disciplined approach towards execution, really sweating the details, really focusing on every milestone in terms of achievement. And that's enabled us to move a lot in terms of improvement. If you look at it from a financial perspective, we've moved EBITDA margins by 1,700 basis points from '16 to 2020 in an industry that really hasn't moved at all in terms of profitability. So a significant improvement on a relative basis. We've actually -- if you look at '16 versus '20, so that's a trough to trough kind of comparison on the revenue side. We've grown revenue 70%. We've seen gross margins through the entire cycle be 40%, EBITDA margins be 50%, our return on invested capital climbed to 20% through that period. Today, when you look at our nodes, we were the first to introduce 1-alpha in DRAM, the first to introduce 176-layer in terms of NAND. So really, the first time we've been in a leadership position, and we've done a remarkable job, and that's representative of the EBITDA margin improvement, really closing the gap relative to competition. I think now, at this point, it's about continuing to maintain the pace of the rest of the industry in terms of node transition. So continue the execution node-to-node that we've had. I think, also on top of that, we think we can continue to improve our product mix. We've done a lot in terms of NAND, moving our product mix to -- into the high-value portion of the market. And so when we exited '20, we were about 80% of -- in terms of percentage of the high-value market. And while we like that percentage, and that's kind of where we'd like to be, what we want to do is continue to improve the mix within the high-value solutions. So move more to the data center space. We bring out additional products in our NVMe, SSD area. So that's really what we have to do kind of uniquely within the industry to execute on the road map. But I actually think the biggest thing about the next few years is that there are so many really significant secular growth drivers that memory and storage will continue to be able to take advantage of: 5G in the mobile space; cloud, of course, which will continue to expand; AI within cloud and within other areas of the market; autonomous driving; more advanced capabilities within the automotive space. All these things, we see as big drivers of growth, both on the DRAM and the NAND front. And so the key for us is to take advantage of those growth drivers, continue to execute on the technology front, bring out products that really address customers' needs in a material way. And I think we'll have a really good opportunity over the next few years to drive meaningful profitability and growth.

Awesome. Yes, I think that's really helpful from a technology perspective and kind of how we got your perspective. So thank you for covering that. Why don't we take it from how we got out here to trends more recently? Over the past couple of quarters, it's really no secret that many of the major end markets have been strong, with even some being better than usual with the work-from-home and learn-from-home environment. Could you just give us an update on how you see some of the major end markets today in terms of demand and kind of how that work from home and learn from home has impacted some of those along the way?

Good question, Tom. So first of all, as you point out, COVID certainly accelerated -- it might have negatively impacted some markets early on, like mobile and automotive and some of the more consumer-oriented markets. But it definitely accelerated some of the other markets as we move towards work from home, learn from home, so forth. Cloud was certainly one of the areas that we saw good demand through the second half of the fiscal '20 period. We saw good strength in the notebook side as people are trying to procure notebooks to be able to manage their lives effectively now in a work-from-home environment or educate in home environment. So those markets actually did pretty well. I think as we transition to -- into '21 -- fiscal '21, which started in September of '20, we saw some of the other markets start to come back like the automotive space, certainly started to see some recovery. We started to see the recovery on the mobile front. But I think the other markets also have good legs to them. I think cloud will see strength through the year in terms of cloud demand. We're starting to see, and we mentioned this in the second quarter call that we're starting to actually see some percolation in the enterprise part of the data center space. That's also doing well. Of course, everyone's seeing the strength in the automotive has only accelerated from where it started the fiscal year. Mobile is definitely going to be strong with 5G being a big story for this year. So we think on the demand front, pretty much across all the end markets, it's looking extremely positive for us. I'd say as we look at this quarter, things are tracking very well for us. We still have 2 weeks to go, obviously. And those 2 weeks tend to be pretty important weeks for us. But so far, things have been going really well for us, both in terms of demand on DRAM and demand on NAND. The other aspect of things that I think is important to note is that the supply situation is tight, and that's a function -- particularly on the DRAM front, a function of kind of the CapEx investments that were made during the last, call it, year or so. And what has turned out to be incredibly lean situation across the supplier base in DRAM in terms of their inventory. So things are very tight. That tends to be healthy for the business. And so we -- and we expect that tightness to continue through the rest of this year and into early part of next year. On the NAND front, we were -- we started the year a little more bearish about NAND. We were concerned around overinvestment in that space and perhaps an oversupply situation. We talked in the second fiscal quarter that pricing had stabilized but continued to be cautious. And while I think we'll always be more cautious around NAND given that there are more players in the space, I think every month, we're incrementally a bit more positive about how NAND is progressing through the year. And so I think a lot of that's a function that demand has been better and supply has actually been a bit more -- a little bit more disciplined than perhaps we were anticipating at the beginning of the year. So this is a good time to be in the memory storage business for sure, with markets being pretty healthy across the board. As it relates to kind of like inventory at customers, we think those are all in pretty normal levels. I think a lot of customers are going to move to concept of just in case and away from something called just in time that we were all kind of adopting. And by the way, I'd say that's true for us, too. We're kind of adopting that approach too when we can build inventory. Again, it's tight. So there's not a lot of opportunities to even build inventory into a just-in-case model. But I think that's going to be the way things operate for at least some time given shortages that are somewhat pervasive in semiconductors and in memory, specifically. So we feel like everything is in a good place in terms of inventory levels. Demand looks super strong for the year. And so we would expect kind of healthy business this year.

Great. That's helpful. And since we kind of jumped to the supply center, I just kind of wanted to follow-up there, maybe on the DRAM side. So you mentioned on the last call, obviously, some drought issues. And you talked about how your bit growth in the back half of the year would be pretty much generated by your technology transition to 1-alpha. So could you update us on how are you dealing with these constraints right now on the DRAM side? And how is that transition going where you feel like, "Hey, we can actually get product to our customers with all this great demand that's going on in the second half?"

Yes. So that's a good question. So we have had a couple of, call it, hiccups in terms of the supply side back in December. Early December, late November, we had a couple of excursions. We had a power outage, we had an earthquake impact a little bit. Those things don't help when you're in a lean environment. And as you mentioned, we have the drought situation in Taiwan that's hitting us now. I thought we did a pretty good job working to overcome the early excursions we had. The drought, of course, is a bit of a challenge, but we have been able to procure the necessary water to run the facility at full capacity. It's costing us a little bit more. So our margin guidance was pretty good. It would have been a little bit better had it not been for this drought expense. But we've been able to kind of manage through it and mitigate it so far. So hopefully, this doesn't continue too much longer. Monsoon season, I think, starts in June there. So that, hopefully, alleviates the burden. But I think from now until then, I think we have a reasonable plan. Now your question, I think, is valid. Things are really, really tight. We have to be extremely adaptive to the demand environment. And so far, we've been able to manage through it without too many challenges, either on the memory side or even on the non-memory side of our manufacturing process. But things keep getting tight. So it is kind of a juggling act for us to kind of manage through it. I'm optimistic that things will start to alleviate on the non-memory front as we progress through this calendar year, so that will be helpful. I think for us, on the kind of DRAM bit side, we'll see relatively modest bit growth in the third and fourth fiscal quarter because of how tight things are. And then we get into the 1-alpha node, which should help manage the growth side of the DRAM front. So I think we're managing it effectively. But as I said, it's day-to-day as to how to look at it.

No, that's definitely helpful. Yes. And I mean, alongside that, you have a market that seems constrained. As you kind of highlighted on the demand side, things are very good. That obviously points to a pricing environment, which probably ends up being a bit better. Could you talk about -- I mean I know you can't forecast this specifically. But like when you see a market like this, is there a path in your mind to continue pricing increases? Or is there a point where you're not going to see any more increases? Have we seen already the amount that we're going to see? Like, can you just walk us through, is this a sustainable trend for a healthy market with the price going up like this?

Yes. That's a good question. We try to avoid specifics around trajectory of pricing as a matter of policy. But I would just say this that we think that DRAM demand this year, bit growth demand is probably growing around 20%. We think the industry in total will be less than that, obviously, and we certainly will be less than probably even the industry. I think this environment is going to be like this for the rest of the calendar year. And as I said, even into calendar '22. That generally results in a pretty favorable pricing environment. Maybe that's -- I'll leave it at that.

No, that's fair. Okay. So we dove into DRAM a little bit. I just kind of want to take a step back to some of the end markets. So the demand side sounds really good, but I just want to dive specifically into a couple that I've been getting hit on recently from investors. So 2 things really. One is that there's been some rumors out about some weaker Chinese handsets that are out in the market. And two, there are some concerns about PC ODM inventory, where certain players they're holding 10 weeks of inventory, whereas others who don't have this supply. Can you give us a comment because everyone is looking for the first crack in the market, right? Everything is so good. Everyone's looking for the first crack. So can you just address those two concerns and kind of what you're seeing in those specific end markets?

So we're not seeing a crack in the market. I think the mobile business, in terms of our outlook for the year, is very much intact relative to where we were thinking at the beginning of this quarter. I mean, yes, there are -- I'm sure there are machinations within customers in terms of their position -- market position and their outlook versus maybe some of the other players. So that is always happening. So I don't know where the data points come, but it could be from one or two that have an adjustment versus others that have an adjustment upside. But for us, the outlook is very much intact for the year. Inventory looks fine in that market. And we're expecting a really healthy environment. 5G being one of the big drivers. And also for us, with 5G comes content growth. In a 4G phone, DRAM, there's 3 to 4 gigs of DRAM. In a 5G phone, to get the 5G experience, a minimum of 6 gigs, and we've seen SKUs at 8, 10 or 12, and even 16 gigs. So this will be a great content story for us this year in an environment where I think 5G has a really good demand environment, just at the end unit level. So I -- for us, the outlook looks great. On the PC side, the PC side was the other question, Tom, yes?

Yes, yes. Certain ODMs with more inventory than others, et cetera.

Yes, I'm sure that's the case. I'm sure some have more than others. But I don't think anybody is thinking they have too much inventory right now given the situation. And so I think we continue to be pretty bullish about the notebook space. I think all this work from home and even when people go back to some return to office, it seems like a lot of them are going to have at least some kind of hybrid approach. Maybe there are some people that will have permanent work-from-home opportunities. And so I think that the idea of having multiple notebooks in a household is probably a trend that's here to stay. So we feel like that market looks really good. Yes, maybe some have a bit more inventory than others, but I don't think -- I think people are going to operate at higher levels of inventory. And again, it will be somewhat similar. There'll be some machinations, some will be adjusting down, some will be adjusting up. So I feel it will be helpful to us. The other part of the client space is the desktop area. Now that has been a pretty weak market for us, not surprisingly, because a lot of the desktops are enterprise desktops that probably there hasn't been a lot of spending on given everybody's mostly been working from home. I do think that at some point, as people start to migrate back, that's certainly going to happen in the U.S., I think, over the next few months, I would expect desktops to start to see some improvement from where they have been. And so that's another part of the client space that I think has some opportunity to see some tailwinds to it over the next year.

Great. No, that's really helpful. So another market impacted by DRAM more specifically, too, is looking at the data center. And you kind of circled around it in your intro remarks talking about how data centers should be strong, particularly the hyperscalers, for the rest of the year. And then you also said, "Hey, you know enterprise on the last earnings call would be coming back a little bit as well." Can you talk about what's driving the strength in the data center? Is it just units or is it content per box? And then kind of in conjunction with that, you've always seen the trade-off of enterprise versus cloud, like is enterprise going to come at the expenses of cloud? How do you see that market kind of coming back? Is it going to be greater than pre-COVID on the enterprise level before? Just any comments you have around those 2 markets would be helpful.

Yes. Okay. So let me just -- because I did meander a little bit. Let me go back and talk a little bit about the cloud and frame it up, if you don't mind. So like I said, third fiscal quarter of '20 was -- we started -- we saw some, I would call, unexpected strength, right, when this COVID hit, and suddenly, everybody was on Zoom and suddenly cloud became really important. We saw some improvement. And then in the fourth quarter, which was helped out by an extra week, but we saw a really good strength on the cloud front. Some of that, I think, was inventory in the fourth quarter of '20. So that was the August '20 quarter. And so not surprisingly, we did see an adjustment in the first quarter of '21, where it did kind of sequentially go down. Some of it was not having the extra week, but some of it was, I think, some adjustment. The second fiscal quarter started to -- we started to see some better business profile. And we're expecting that as we migrate through the rest of the fiscal and calendar year, we'll see increasing strength in that space. And that is a function of just build-out, of course, as the hyperscale customers continue to invest and expand. I think this COVID kind of catalyst to that business, it didn't really like, to be honest with you, I don't think it created much of a bubble. I think it has accelerated this drive towards the cloud. And I can tell you in Micron, specifically even, we are looking more at leveraging the cloud now, holding hat now, have more experience in the cloud. And I'm sure a lot of companies are looking in that direction. And then the second driver of, I think, improving market conditions there is what you said is content growth. So it's content growth because CPUs will add channels, and there'll be more memory and memory is clearly one of the major bottlenecks within the hyperscaler footprint in order for them to drive the compute horsepower they want. And also, as cloud workloads become more AI, machine learning, kind of driven, there's a lot more DRAM required in those systems. And it's like a 6x increase in DRAM to drive those systems. So that also, I think, is a catalyst for continued growth. I mentioned on the enterprise front, we have started to see that percolate. It really kind of fell off quite a bit during COVID, because obviously, your general corporate wasn't investing a lot on-prem. And I think companies that have a business model that requires big -- a big data center footprint for themselves, some of those industries gotten pretty negatively impacted by COVID, and they dialed back on their capital investments, and that also impacted things in the enterprise. Now those businesses, many of them are starting to come back. And I think that's helping drive some improving market conditions there. And also, I'll get back to -- I used Micron because I'm front and center in it, I used Micron as an example. We also are now starting to look at reinvesting in on-prem. And so the natural outcome is that there will be incremental investment there. I do, though, think that this whole like migration of a certain amount of applications to the cloud is going to be a trend that continues for a while. And so it may be that the enterprise space doesn't feel like quite the momentum back that maybe it was pre-COVID, while improving, not to those levels because cloud is going to sop up some of that opportunity as companies leverage the cloud. But enterprise will always be there. We'll always invest on-prem for certain applications, we'll need to. And as I said, certain industries will want to keep their data centers close to them so that they can optimize them. And so we -- I don't think the enterprise goes away. I think it's going to be a decent market, but clearly, cloud will be a market driver of growth for the future.

Great. It's helpful. So continuing our meandering here. So we kind of carved out some of the major DRAM markets. I think turning to the NAND side, you made some comments before, and particularly on the last earnings call that things were better from a demand perspective. There's no doubt about that. But you mentioned, I think, in your slides that supply is exceeding demand with some CapEx cuts that may be needed. So we've heard from others in the industry that NAND is tight, but we've also heard and seeing robust ordering from the semi cap group as early as the end of last year, really coming into this year. And people call for technology transitions, requiring more capital intensity, et cetera. Can you just walk us through the puts and takes of an environment like this? Like are you seeing this equipment come on too fast where the market is kind of out of control? Or is the demand kind of keeping up for now? And how do you see it trending in the next couple of quarters?

Yes. So in the near term, I think it's pretty clear. Demand is robust and supply is, in some ways, even struggling to keep up with that. We had, I don't know, like 130 something days of inventory, I think, coming into the second fiscal quarter. And we had pretty much gotten it down to 99 days, I think, by the end of the fiscal quarter. And NAND was definitely higher than DRAM in terms of its days of inventory, and it was quick. It was amazing how quickly that went back -- went down to normal. So we are seeing a much leaner environment there. I think as we look at the year for NAND, I think originally when we came into the year, we were thinking that the growth rate in NAND would be like 30%. And I think we moved it to maybe lower 30s, and now we're saying more like low to mid-30s. So it's clearly surprising us to the upside on the demand front. I mean there's going to be CapEx investment no matter what on -- in these technology transitions, they're expensive transitions. So I'm not sure just if at all our movement is and even the equipment needed to produce the next-generation NAND is that disconcerting. I mean there's -- we actually increased our NAND investment this fiscal year pretty meaningfully. In fact, most of the increase in our CapEx is really related to NAND. Now despite that investment, we feel we're going to be growing in line with the end demand over the next several years. So we are maintaining good discipline. So it's not to say that when you see that CapEx investment, that, that in and of itself, indicates a lack of discipline by the industry. As I said, that NAND is always going to be one where we're more cautious than DRAM. There's more players making investment. And there has a bigger opportunity for it to get undisciplined in terms of the capital investment. But for now, it's -- I don't see that showing up on the supply side. We'll have to kind of see how things progress through the year and into next year to see if bids come online from an industry perspective at a rate that exceeds demand. But for now, we're just not seeing it.

Fair enough. You mentioned a lot of players, obviously, going after the same market there. We've seen consolidation efforts in NAND, Intel, Hynix. There's a lot of moving pieces around Toshiba right now. Do you see there is an ability with the macro environment, one; and then two, the players that exist that there could be continued consolidation in the memory space? And do you think that you guys would ever look to be an acquirer in a market like that just because industry consolidation has been really healthy for you guys in the past?

Yes. It's hard for me to predict what happens on the consolidation front, obviously. And consolidation is one of those things. Acquisitions are -- you can't count on them. Stars have to align in a lot of different ways. It's got to make strategic sense. You got to feel like the integration makes sense. It's got to make financial sense for sure for investors. And for us, in particular, we have a high bar on all 3 of those things. And so it's why you haven't seen us be an active acquirer. I think for us, on the NAND front, we can't count -- consolidation certainly usually is more helpful to the market, but we can't count on that. It's not something that we have to kind of wrap a strategy around. So what we've got to do on the NAND front in our business is, we've got to be completely focused on technology transitions and the execution of those transitions, bring out world-class process technology and do it on a cadence that's aligned with the industry. We also have QLC, and QLC has good margins relative to TLC. And so we've been a leader in that space. And we got to execute in terms of bringing out QLC-related products and continue to drive the market in that direction. So that's an important part of our kind of road map. We call what we have is -- in terms of NAND as CMOS under the Array, which has got some cost advantages to it, too. And as we evolve the technology -- process technology, we need to maintain that capability. So that's also pretty important for us. And then lastly, while we -- like I said at the beginning, while we've moved the mix of our business and NAND from being mainly a component business to being a solutions business, 80% of our bits, as we exited fiscal '20, we're going into high-value solutions. There's still mix that we've got to do within the high-value solutions area. We are pretty underrepresented in the data center space. And that's largely because NVMe has really become the dominant technology within the data center space. We started NVMe kind of in the consumer and client area. We're starting to bring out the products in the data -- that address the data center part of the market now. We got -- those products -- we got to execute on those products, get those products out, [ coil ] them with customers, make sure we're supporting our customers in every way possible. And I think if we do all those things, I think the ROI of this business is solid. It may not be to the level of DRAM, but I think it's very solid and really rely on kind of heads down approach towards just managing this organically, making sure we got the right mix of controllers that we're doing internally and then utilizing merchant controllers where that makes better economic sense. So all of those things, I think, are the things that we got to focus on as a company and let M&A, if it happens, it happens, that's probably beneficial to us if the industry is more consolidated. If it doesn't, that's okay, because we've got a road map and a plan to drive good profitability in this business regardless.

Great. Helpful. Something bigger on the industry as well and something we've been hearing a lot about recently is when you look at overall industry investment and efforts by a lot of governments, including the U.S., Europe, Korea, to increase domestic capacity of semiconductor production, what are the opportunities available for Micron? I feel like sometimes Micron isn't mentioned, like some of the other big players, should be in the same sense, all right? So when you hear like these big numbers in Korea, like $450 billion over the next 10 years, that was recently announced, and you've heard several others, like how do you -- like how does this impact the global supply? And how does it impact Micron?

Yes. First of all, I'm not so sure that, that's something new. This concept of governments assisting semiconductor place, these are really expensive, really difficult to do, very sophisticated fabs that are getting built now. They require lots of capability in terms of human resources and lots of know-how in terms of IP and then lots of money, which requires some assistance from governments. We've gotten that assistance in the past in our footprint. So just, I think, a sign of the times of how the barriers and hurdles necessary to become -- to be a producer of high-volume semiconductors across the board, whether you're logic, foundry, analog and memory. And so that's not something new. That is something that's always existed. I think the new thing for us is the U.S. government's interest in getting into the fray here and putting money to work for their own interest. And I guess, I would say that, that's certainly something, I think, that we thought was needed as a company, as an industry. I applaud the U.S. government for approaching this and looking for solutions in that regard. As it relates to Micron, we're always thinking about matching our supply with our demand. That's kind of the ground rules for everything we do. And that's always the way we're going to approach any investments in DRAM and NAND. We also have a very good footprint of fabs that are highly capable, cost-effective fabs today. We have 1 DRAM Fab in Japan, 2 that we kind of look at as kind of 1 mega Fab within Taiwan and then a Fab in Singapore that does our NAND business. And then a legacy Fab that we have in Manassas, Virginia. So those -- that's -- we're happy with that model. Whether it will make sense over time to have another fab, that, a, aligns with our cost structure; b, supports this concept that we will always be disciplined in terms of our bit supply relative to the demand and whether we can take advantage of the resources of the U.S. government, I think, remains to be seen. So we'll have to kind of see how that goes. It's, I think, in -- relative in terms of the decision-making there. We're certainly looking at what will transpire there. And if we can take advantage of it and do something that supports the U.S. government's interest and also supports what's needed from our customers, we'll certainly take a hard look at it.

Thanks for that. Moving to broader technology, now that we've kind of covered the major bases here. So you announced in March that you would stop investment in 3D XPoint and shift focus to Compute Express Link, which you call CXL. I think that I hadn't heard much about CXL prior to the announcement. Could you talk about why you believe CXL is important and some of the progress you've made there just because I've seen it now announced across the industry a couple of times and understanding what that means for you guys to support?

Yes. So CXL, just in somewhat layman's terms, it's an open architecture approach that allows kind of different architectures within data centers for memory. There will increasingly be a need to optimize memory, different usages of memory within the data center and CXL kind of opens the door for that opportunity. By the way, CXL is -- could have also been supported in 3D XPoint. I think as we looked at 3D XPoint specifically as a technology, we felt that the market opportunities just didn't match up with the level of investment necessary to bring that technology out. And we're an ROI-based company. So we're always looking at the ROI of everything. And when we were continually updating it, and we got, I think, a better sense of what the opportunity set was, it was -- we ran them out and that just didn't make sense. And we also felt like there were some other things that we could do, which I'll refrain from discussing at this point, but we'll likely be talking about over time as we're -- we feel comfortable introducing those capabilities. We feel like CXL does unlock some of those other technologies that we think do have a more expansive market opportunity. The investment versus the payoff make a lot more sense for us. And we think -- I think CXL is exciting. I think we think it's exciting. It's got a really good capability to it that we think we can exploit with our own technology that really, I mean, at the end of the day, what we're trying to do is bring solutions to customers that really help them drive their needs, and we think we can do that with CXL and with kind of internal technology that...

Awesome. All right. Next one is kind of a 2-part question kind of relating to both that and then some broader spend. So there's obviously a Fab sale that's going on in Lehi in conjunction with this decision. You're going to generate a decent amount of cash there. And when we look at that cash and we look at the CapEx build that you guys have on a yearly basis, can you talk about, one, when you get flush with cash in the second half of this year, you look at just how the model is progressing, you're obviously generating a lot of cash. Now with this Fab being sold, where is this going to go? How do you think about the capital allocation strategy? And then more specifically on CapEx, can you talk about your little under run rate maybe last year, you're more close to normalized run rate this year. Talk about next year, and what that means now that you're generating a little bit more cash for your spending plans?

Run rate being cash flow, you mean, run rate, is that what you meant? Or?

CapEx, total CapEx spend, yes.

Okay. All right. Yes. Okay. So you're right. Yes, we will be selling the Lehi Fab here. We think we'll close by the end of the year. I'd say things are progressing well there. So -- and you're right, that will obviously put cash in the coffers for sure. We also went a little bit cash flow positive in the second quarter. We think we'll be pretty meaningfully cash flow positive in the third fiscal quarter. And for the year, I think this will be a solid year in terms of free cash flow. So high-class problem to have, I guess, as you say. Cash coming in on sale plus cash generated by the business, great situation to have. We are buying back stock this quarter. We have talked about that we want to return at least 50% of our cash in the form of buybacks, we remain committed to that. And also think that the price is a pretty good price to be buying stock at right now. So while sometimes, you're not happy about the stock going down, this is an advantageous time for us to be buying stock back. So that's good. I think you'll see in the future, we do expect this business to generate good free cash flow year-to-year. Some years will be better than others, but we would expect to be returning a decent portion of free cash to shareholders. On top of that, there are other needs that we have, clean up the debt. Sometimes we have some converts. We still have a convert outstanding, so there'll be some of that cash used to clean those kind of things up over time. And then, as revenue gets bigger, we have a stated policy that we're trying to keep liquidity in the low 30s as a percent of revenue. And so as we grow revenue over time, theoretically, the liquidity level, we'd want to be growing over time, too. So that will be another use of the cash. On the CapEx front, I would say maybe the best way to describe it rather than like calling out 1 specific year is what we're trying to do over time -- what we try to do is look at things over averages over multiyears because any 1 year can be -- can distort the picture either as a too high number or a too low network, quite honestly. And so we look at it over these multiyear periods. And I think, for us, what we're trying to do from a CapEx intensity is invest somewhere between 30% and 35% of revenue. Now that's not the number that -- that number doesn't drive the decision. We're trying to invest capital to grow the bits consistent with demand over time for both DRAM and NAND. So we think like longer-term bit demand for DRAM would be kind of low to -- I mean mid- to high teens. And on that -- DRAM -- did I say DRAM -- yes, mid- to high teens. And on the NAND front, we'd be like roughly 30%. So we think those are the demand growth profile for both of those products. So now our investments are around matching the supply to match up with that demand. And we think when you do that, the output of that is this kind of CapEx intensity that's somewhere in the kind of 30% to 35% kind of range in terms of intensity. So some years, we haven't like completely worked out what CapEx will be next year. And some years will be lower and depending on revenue and that year in terms of CapEx and some years will be higher. But over time, it should roughly kind of translate into those. The other thing I would say is -- on the CapEx front is there's a fair amount of CapEx for us that also doesn't go to necessarily big growth, right? I mean we're spending a lot in the back end to bring that in-house. We are, I think, more proactively just having kind of white space available for clean room expansion as we bring out new process technologies, which are more complex, which require more real estate. So some of this investment doesn't have any impact on bit growth in the future. Some of it doesn't have impact from -- for several years before that turns around and becomes an investment. So there's years where our CapEx intensity is higher even though our investment on the front end in terms of bit growth maybe not high, actually might be lower than the previous year. And so it's -- there can be some noise around it. But over time, I think if we drive to this 30% to 35%, I think that, that would be a pretty good outcome. And I think that would drive really good free cash flow.

Awesome. Well, I think that's all the time we have. Dave, thank you very much for joining us. Good luck in the coming quarters, and we'll chat soon.

Okay. Thanks. Thanks, everyone. Appreciate your time. Take care. Bye.