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

[Audio Gap] to speak with us, Mark Murphy, the CFO; and Satya Kumar from Investor Relations. So gentlemen, thanks both for attending today. We've got a lot to talk about. Mark, maybe I'll turn it over to you for your safe harbor and some opening comments, and we can get ready for some Q&A.

Sure. So great to be here, Chris, and thank you, everybody, for joining us today. So starting with safe harbor. I'll be making forward-looking statements. Those statements have risks and uncertainties associated with them. I refer you to the risk factors disclosed in our public filings. So some opening comments. We have no change to, or no update to our second quarter guidance. On the third quarter -- fiscal third quarter, we do expect revenue growth. We now see fiscal third quarter gross margins to be lower by a few hundred basis points sequentially. This is primarily due to a higher mix of consumer-related or consumer-oriented mix in fiscal Q3, lower C Q1 consumer-oriented pricing and NAND -- overall or broader NAND industry conditions. As a reminder, our underutilization in NAND will weigh on fiscal third quarter margins as well. We do expect industry conditions to support improved margins beyond fiscal Q3. As it relates to the industry, we are seeing improvements in client and smartphone inventory levels, and we expect them to be healthy by the spring. We see, in calendar '25, continued content growth in PC and smartphone. Of course, as you can see from recent announcements, we expect continued growth in data center. And then we do expect supply reductions in NAND to support that market over time.

Well, you've given us a lot to chew on for that. So maybe I'll kind of dig into some of the things that you said there. So you talked about the gross margin decline next quarter being driven principally by mix. So is that a function of -- again, seasonally in this point of the year, you'd expect consumer to be a little bit weaker. So what's driving that mix in this point of the year?

Yes. We're just seeing stronger volumes than we had anticipated in a number of consumer markets. And that could be mobile, it's various channels. And we would -- it's just driving our mix higher than we expected and contributing to what we see as revenue growth in the third quarter. We do see bit shipments up in the third quarter. We do see second half bits being stronger than first half bits. So these are just stronger signs for our later of the year industry conditions. We are getting good signals. As I mentioned on the PC or client and smartphone, we are seeing inventory levels improve. And so if we look at the conditions for demand, be it unit growth, be it content growth, be it customer inventory levels, for all 3, we see positive signs. In the case of end market demand, we do see low to mid-single-digit growth for PC, smartphone markets. For content growth, AI is having an effect, both the Windows update and then AI-related content, we're seeing 16 gigabyte mins in PC refresh. In smartphone, we're seeing content uplift of 50-plus percent for AI-related devices. And then finally, again, on inventory levels, in PC, smartphone, we are seeing those inventory levels improve at customers and think they'll be in a healthy place by spring.

And maybe you could speak to the nonconsumer part of the market. And I guess, is this a function of the consumer markets actually turning out to be stronger than you thought? You maybe haircut those because what was going on. What about the nonconsumer markets? Are they not picking up as much and, therefore, declining as a percentage of the mix?

Yes. I mean the nonconsumer markets, we've talked about automotive and some transition there. But data center demand remains strong on DRAM. And I think in nonconsumer markets, we have talked about how our data center eSSD has got a bit of a digestion period. So that's an area that, while volumes are up, we would -- that's an area that's not up as much as consumer parts of the market.

And that's about the same drag that you talked about in the December quarter. So...

We have not updated that. Yes.

What about with respect to HBM? That's an area where you're expecting to be ramping. There's a product transition at the largest customer there as well, which kind of has some difference between mix of 8-high, 12-high, and eventually HBM4 towards the end of the year. How are you progressing on that? And how is that going against your expectations?

Progress on HBM continues to be great for Micron, commercially and technically and production. So as it relates to this year's product cycle, so we will be ramping here shortly the 12-high stack for HBM. And it's an impressive product. I mean it's 20% less power than competitors' 8-high stack at 50% more capacity. So it's an impressive product, and we would expect the majority of our HBM volumes to be on that product second half of the year.

There's an impact in -- is it really kind of May quarter impact? Or is this some 12-high...

It's beyond that. Yes, beyond that. But our position is good. Our technology is good. And so we expect to continue this leadership in HBM4. So time to market there and technology should be -- we feel we're in a great position, and that would ramp in '26. And then as we go into 4E, what's exciting becomes more exciting. So there's the opportunity for customization on the base die. And of course, we've got engagements with customers already on these sort of technical solutions. And I think that puts Micron in a good position for this to continue to be a strong contributor to the P&L.

And your goal is still to get what your typical share is as compared to the commodity markets, and still second half of the year is still a reasonable expectation for that?

That is. That's on track, and that is our stated intention to be in line with our broader DRAM share.

One of the questions we get is that, obviously, the big guy in the block, Samsung, is not participating as fully now as they are in the commodity markets. Stated intention of getting there. They're working out some technical challenges publicly. They're Is expected to be there for HBM4. I guess the question for you is, if and when Samsung enters the market, what does that mean for Micron? Would you still achieve that 20% share whether or not Samsung winds up coming back into the market and taking whatever share they wind up getting?

Yes. I mean it starts with the fact that we've got the best technology in the business. We've got unique design attributes that are durable. We've built our product on 1-beta technology, which is a great process technology and has been very stable to launch the product and ramp it. We will be moving to 1-gamma over time. And this knowledge about the product, our position, the fast cadence, the process technology leadership, all these things will add up to that we believe will sustain our technical advantage on the product category. We do believe that Samsung will be successful. That's our base case. And so we will -- with our superior product capabilities, technology capabilities, we would expect to be able to achieve our target DRAM share -- our target HBM share. And it's important -- supply/demand is an important part of this market as well. So it's important that we keep that in mind.

And the margins and the profitability of the HBM business, you still expect that to be accretive to overall DRAM and the corporate average?

We do. We've stated from the outset that it's accretive, which is impressive, because obviously, it was an early ramp at one point. We stated last quarter, I believe that it's significantly accretive. We would expect it to be accretive in this quarter and next quarter. And based on the superior value proposition of the product and our position, we believe we sustain that accretive level from what we see.

Is there any sort of HBM impact implied in what you're talking about for May quarter margin? It's like the mix is higher of consumer, but I generally would have felt that HBM would be ramping in May as well, which would be an offset to that. So is that offset perhaps a little less? Or I don't want to put words in your mouth, but that's probably one of the questions folks will ask.

Yes. I think your -- I mean, we obviously had contemplated an HBM ramp, and that's going well as well or better than expected. And that's a favorable effect on margins, as we've talked about. But there are, as we said, some headwinds on margin as we go into the third quarter. And it's a higher mix of consumer-oriented products. It's lower pricing that we're seeing in these consumer-oriented markets in calendar Q1. And the NAND industry conditions overall are still weak in our view. Now we see industry conditions improving beyond the fiscal third quarter. So that general profile that we talked about in the last call is still there, but the third quarter will be lower on gross margin than we had anticipated. But again, as we go through the year, HBM will get bigger in our view today. Other data center products which are higher margin will continue to grow. And then our eSSD, which is a higher-margin NAND product, will continue to grow through the year, and that will also be with the backdrop of better NAND market conditions, hopefully. So all those things should play out and we should be able to resume margin growth in the back half of the calendar year and our fiscal fourth quarter.

It sounds like it's sort of a picture where May quarter margins are probably low for the year, not to, again, put words in your mouth, but...

We're not guiding specifically beyond -- we gave this a few hundred basis points down sequentially in the third quarter. But we've not given anything beyond, but the factors we talked about, industry conditions improving in the back half of the year, calendar year; the inventory levels that are improving in many end markets; the supply response that we've seen in NAND helping. And so these factors should contribute to third quarter being the low point.

I want to dig in on NAND a little bit, because that was certainly a focus of the call in December. And I think what you said at the time was that you had a view of the industry needed to slow tech migration because of the number of bits that were being driven by the higher layer counts. What's your optimism on that as you go into the second half -- into next year? And again, part of the question is, we heard from Samsung, since you spoke. And Samsung seemed like they took a little bit of a different view than you did. We heard from SanDisk yesterday. It seemed like it was a bit closer to your view about slowing tech migration. And I guess there's things that you can control, which is your own road map on what you can do and then there's the other is what happens with the industry. And based on what we've all heard over the last 90 days, what's your level of optimism that we get to a supply/demand balance in NAND?

Well, there are some encouraging signs. But this is one of these things that there needs to be continued discipline. There needs to be a continued drawdown of inventories in NAND. And that's happening. You've seen the supply response in the industry, and we see it in some of these end markets that the volumes are a bit higher and the inventory levels are improving. You've got -- there's ample supply at the moment and there's inventories, and there's a couple of ways to deal with that. You can you can delay node migrations and what that does, right. It doesn't -- a node migration will produce more bits per wafer. So pushing that out, delaying those migrations as we've done is, I think, prudent. And then I will say, we did launch our G9, which is industry-leading, but the ramp on that has slowed. And then you've got -- and then you can take wafer capacity down as the other option. And we did a lot of that during the downturn. But even in this relatively short period of oversupply, we've brought our utilization down immediately. And you've seen others generally do that.

Right. And I guess it's a little different where there's -- the conflicting narratives are that I think everyone agrees right now, there's oversupply today, which is why people are taking production actions. And the other part of it is that because pricing has come down, there's a view of, well, we need to get our costs in line with pricing. But the problem, and I think this is what you guys expressed very well in December is, the problem with that is by lowering the cost, you're also creating more capacity.

Right.

That's the difficult situation. It's hard to understand how that gets resolved quickly unless that tech migration slows.

Yes. I mean the bit demand, as we talked about in December, that we lowered the bit demand for '24 and '25 down to low double digits, and it had been obviously much higher than that before. So I think it's just a case of -- we always try and manage -- we're very sensitive to understanding the demand signals and creating the right supply for that. And it's just that continual maintenance and being disciplined on our capital spend and operating conditions that's important.

To what extent is -- are there potential disruptions from China? And maybe I'll start with the NAND market, because NAND is one of the areas where the Chinese supplier there is perhaps a bit less impacted as compared to DRAM and your ability to get process equipment and migrate to the next node. How much of an impact does that have on the market? Because we would assume that they will not slow tech migration, because it's not in their interest to do so. And are they big enough in the market to have a sort of persistent effect on pricing and supply/demand?

I mean, they're subject to the same market pressures that we are. And I think that they serve a certain part of the market. We generally will serve other parts of the market. And I mean, at the end of the day, the industry needs to get supply in line with demand, and that occurs through tech node migration timing, in this case, delay, and wafer reductions or not putting in additional capacity. And then on China, I guess, DRAM also you're saying. So on DRAM, we serve the higher-performing parts of the market, including in China and have a good capability in China and serve our important customers there. For example, if you take DDR5, we're on our third generation of DDR5. And our speed rates are near double the competition -- Chinese competition there. So we have a performance advantage and that's -- and we serve different parts of the market as a result. And we'll continue to advance the technology, and we will move to 1-gamma here. It will go into volume production this year. And then that will support architecture developments for us and performance improvements for us going forward.

What will be the principal product on 1-gamma, I assume, that goes in HBM first or...

I mean, no, I wouldn't say that necessarily. I think product road map decisions are complicated and depends on customer ramps and stability and all that. It's going very well. So I think we have the option to move things there, but it's a complicated decision because obviously, a lot of work is already underway with customers on many different products. So we'll intersect the product road maps at the time that's best for our customers and best for our efficiency, profitability.

I'll also open a question if somebody has one in the audience. We've got about 15 minutes left. I don't think there's a mic, but I can say it. If you want to ask one, sir? Oh, he does have a microphone. Sorry, I didn't realize.

I was just wondering if the pressure is mainly driven just by NAND, which seems to be problematic in terms of the gross margin pressure? Or do we also see pricing maybe for commodity DRAM as a source of gross margin weakness for Q3?

So it's a good question. I mean, primarily what we see is the factors I laid out. We see the gross margin decline driven by the consumer-oriented mix, the higher mix in consumer. And in C Q1, we saw consumer-oriented pricing down in both DRAM and NAND. Now we have seen, in the past few quarters, some price decline in D4 products. I think that's known. But we'll see how the market looks going forward. It's important to keep in mind that our leading edge is tight. So our 1-beta production is basically between HBM and DDR5-related, LP5-related products. That market, we're tight on supply, and inventory levels will be lean by end of year. But we have seen, in the past few quarters, some erosion in D4.

And a lot of that was driven by China in general and competition there as they put up some capacity. And you talked about your D5 performance. And I guess as one of the other fears is that, I guess if this China impact is isolated to D4 and LP4, the market is moving on from that pretty quickly. So if you maintain that competitive advantage, it means that China would be, therefore, less disruptive going forward. There's technology restrictions they have in doing that. Do you think that plays out that way, that it's more difficult for Chinese competitors to be more disruptive on D5 and LP5, and that provides kind of a floor to pricing a better supply/demand balance as you go in the back half of the year into next year?

Yes. I mean the performance stat that I gave was architecture common. So it's still going to be on the older architectures as the previous question posed. We've seen pricing weakness. As we go forward, what helps either reduce any pricing pressure or limit the duration of it is the fact that we're serving performance-driven markets and our process technology serving those performance-driven products is tight, in part because of HBM and its trade ratio and capacity demands. So certainly, pricing can be impacted by a number of factors, particularly in the short term. But I think that, as I laid out in the comments, we've seen primarily erosion in consumer-related market pricing, and we saw that in C Q1. And then to date, fairly limited outside that.

What about from an industry capacity standpoint?

Oh, and I should say, and broader NAND products. Yes.

I think what we've been hearing from a while from yourselves, your competitors is that in DRAM, there's a reluctance to add bit production. However, there's still spending associated with HBM. And sometimes a little hard to disaggregate that, right? That there is capacity being added, but it's being added to leading-edge node. As you said, it's being absorbed by these leading-edge products. What's your level of conviction that the spending of the industry right now is kind of leading itself to as we get to the second half into next year that we get tighter supply just because of -- maybe it's a function of supply and bit demand, right?

What gives us confidence in the second half?

Second half.

Yes. So I mean, a number of things. If we start with PC, smartphone, as I mentioned earlier, if I look at the 3 fundamental drivers of demand, end market units, content and customer inventory levels. All 3 of those are positive in the sense that we still see growth in PC, smartphone low to mid-single digits. Importantly, we are seeing this content uplift trend on PC, smartphone. And that's clear. And then finally, as I mentioned, we are seeing inventory reductions at customers in PC, smartphone or client and smartphone. And we expect them to be healthy by the spring. Now on data center, the market got spooked a bit with this DeepSeek announcement. So everybody is, of course, watching the CapEx commentary by the hyperscalers. And that was certainly a positive data point taken in aggregate and staggering numbers, right, 50% year-over-year growth projected when it was 50% growth last year. And then nearly double or around double quarter year-over-year, 30% or so quarter sequential. Those are pretty staggering numbers. And then a couple of other interesting data points around that. We noticed comments about shell capacity is not the constraint, more in the year projection it server is the constraint. And then even the useful life being ticked down by one of them sort of to me suggests sustained reinvestment. So I mean all these things, again, support strong data center growth. And there's more memory, as we've talked about, coming into those architectures, be it higher performance HBM, the high-capacity DIMMs, a lot of LP DRAM going into servers. So it's a very good story. And so as mentioned, we see, between those generally good market indicators on demand and supply response on NAND and then continued leading-edge constraints on DRAM supply, we would expect industry conditions to be better in the second half and support the recovery.

From a demand standpoint, I wanted to dig on NAND a little bit as well. One of the bull cases on NAND is eSSD applications for AI. And I guess the question, and I know that's going through a period of digestion now. But how big can that be as a percentage of the NAND business? Is it big enough to move the needle in the absence of some of the price elasticity we've seen in the past? How impactful is eSSD in your view?

Well, for our NAND business, it's incredibly important. And we've done extraordinarily well. So we benefit from having the best technology in the business. And then we did an overhaul of that particular segment, appreciating its performance expectations and the size of market. And those efforts have paid off. And you see that in our -- we just launched another 6550 product variant of 60 terabyte, best performing high-capacity SSD in the marketplace, fastest. We also previously announced this 9550, fastest in the industry. So we've got the fastest eSSDs in the business and a good value proposition for our customers. And so you saw us move the needle on that business. Business reached $1 billion run rate. We're in the sort of period of digestion, but we've said that we expect that business to be multibillions in FY '25. So it's a good business for us and important for Micron specifically.

Can that -- I mean, with that achieving that multibillion, can that be done irrespective of kind of an overall recovery in NAND? In other words, rather than pay attention in the broader NAND business, should I be paying attention specifically to the eSSD business?

Well, we see volumes picking up in the third quarter already and our projection. And we'll give more color on that in the earnings call. It's not that far out, and can give a more specific picture on eSSD recovery and on that recovery in the context of the broader NAND industry.

Out into next year, and I mean, this is the difficult question where we've been waiting for this cyclical recovery, the improvement in margins for a while. And it's not just you, it's pretty much every company I cover. We've been sort of working through this. What's sort of the end game here? If you're right, we've got this kind of recovery in the second half of the year. What's the kind of outlook into next year if it goes as you're hoping right now?

Yes. We're not -- I'm not going to give next year guidance. And we may provide a bit of color on subsequent years beyond just our normal bit growth numbers in the next earnings call. But what I can say is that in the PC, smartphone markets, we talked at length about how the replacement cycle and a content-driven replacement cycle should be helpful. And we believe that will take hold. We're seeing positive signs there. And then on the data center side, we couldn't have had a stronger set of data points in this last earnings cycle about the commitment to investment there and how this -- this is a generational technology change. And Micron is in the best position it's ever been in its history. We're the leader on technology, on DRAM and NAND. We have the best product portfolio in the business. And we're operating very well, very efficiently. And so the combination of those things, I think, put us in a tremendous position as the market conditions improve later this year and as the industry grows over the next several years. We did put out some markers on, for example, the size of HBM and going from $16 billion TAM last year to over $100 billion by 2030. So those are big new markets, and then you've got a number of associated other product lines that will grow sort of in that manner as well.

So it's a function of getting the inventory cleaned out, letting HBM grow and assuming that the CapEx commentary doesn't change as a result of [indiscernible]. I guess it sounds like what you're saying right now, it hasn't.

It doesn't appear to. I think in the end, there are questions about how that model was trained. There are hallucinations in the model and there's security issues that I've read about. However, credit to that team for a lot of innovative technologies in that model. And the most important thing is it's lowering the cost of inferencing, which will -- in the electronics space, these things proliferate. And in the end, we think that's good for memory.

It's just interesting how that news sort of came out about the same time as very aggressive, very favorable CapEx numbers from the end customers, which is something we have to digest, but that's what we do for a living here. All right. I think we're at time. So I think we're going to leave it there. Mark, Satya, thanks for joining us.

Thank you, all.