Good morning, welcome to the first day of J.P. Morgan's 51st Annual Technology Media and Communications Conference. My name is Harlan Sur, I'm the Semiconductor and Semiconductor Capital Equipment analyst for the firm. I'm very pleased to have Mark Murphy, Chief Financial Officer at Micron Technology, here with us today. Mark will start us off with some opening remarks, then we can go ahead and kick off the Q&A. Mark, thank you very much for joining us this morning.
Thank you, Harlan, good morning to everybody. I'll start with the Safe Harbor. I will be making forward-looking statements. Those statements have risks and uncertainties associated with them. I refer you to our risk factors disclosed in our public filings. This morning I'll open my comments with a brief discussion about the trends in the business, and then I will separately address recent developments in China. On the business itself, overall, we're seeing trends that have been consistent with the overview or the outline that we provided on our last earnings call, which was at the end of March. Pricing and volume trends have generally been consistent with our expectations. We are seeing bit growth across DRAM and NAND. As we said in March, we feel that we're closer to a period of sequential revenue growth.
The bits bottomed in November as we had expected. We saw bit growth in the February quarter, and then we see bit growth again in this May quarter. We expect bit growth to continue in the second half of calendar 2023. In our view, we do see the supply-demand balance improving. Supply growth is slowing and demand continues to increase. The trend this quarter, again, things are progressing generally consistent with what we expected. There have been more signs this quarter of industry suppliers cutting supply, now we have all suppliers have indicated that they've reduced supply from previous levels. We are seeing pockets in the marketplace of price stabilization. In some isolated cases, we've actually seen some price increases.
An area of positive price has been DDR5, which has had better price trends than some other areas in the marketplace. Importantly, we're also seeing customer inventories coming down at a gradual pace, down from levels that we've seen over the past year. Q4, we expect pricing declines to moderate from the sequential declines that we've been seeing the last several quarters. We expect bit shipments again to increase May to August quarter, improving sequentially. Now this, we're in the process of assessing the China decision, which I'll talk about momentarily. These comments on the, on the trends of the business, of course, do not contemplate that at this time. We received notice yesterday morning that the Cyberspace Administration of China concluded its review of Micron's products. This review commenced on March 31st.
The CAC concluded Micron products present a cybersecurity risk. As such, the CAC notified operators of critical information infrastructure in China to stop using Micron products. We will continue to cooperate with the CAC and are working to understand the details about their security concerns with Micron products. We remain unclear as to what security concerns exist, and we've had no complaints from customers on the security of our products. We've served customers and have operated in China for over 20 years, providing the most advanced products of the highest quality, and we look forward to enabling our customers' success in China within the confines, of course, of local laws and regulations in China. These are very new developments, and we continue to assess the CAC's conclusions and are evaluating its impacts on our operations and our business.
As we have disclosed in our filings, China and Hong Kong headquartered companies represent about 16% of our revenues. In addition, we have distributors that sell to China-headquartered companies. We estimate that the combined direct sales and indirect sales through distributors to China-headquartered companies is about a quarter of our total revenue. We are evaluating what portion of our sales could be impacted by a critical information infrastructure ban. Depending on what type of customers, products, and end applications are included in the definition of critical information infrastructure, the potential impact would vary. We are currently estimating a range of impact in the low single digits % of our company total revenue at the low end, and high single digit % of total company revenue at the high end.
We have work ahead of us to get a more precise view on the impact of the restrictions. We will continue to assess the situation and work with our local authorities, agency officials to get greater clarity and, of course, our customers.
Perfect. Thanks for that, Mark. On the Cyberspace Administration of China, or CAC review and decision, I know this just happened yesterday, U.S. time. Micron is still trying to assess the situation, so I really appreciate the transparency with the limited information. You did level set us, China direct exposure plus indirect exposure on distributors that ship into those customers about a quarter of your business. There are a number of value chain manufacturers in China that build systems in China, ship them outside of China. Are these value chain manufacturers also subject to the CAC restrictions, even if the systems are built for export?
We don't know. I think we're working to assess and get greater clarity on the restrictions. They are customer-driven restrictions, we're not restricted.
Mm-hmm.
It's the customers that are unable to buy from us. We will learn more about the range of customers impacted. What we've tried to do is apply what we think are reasonable expectations of what those, what the range of outcomes could be based on what we know at this time. You know, we'll continue to evaluate, and to the extent there's a material change, we would work to update the market at the appropriate time.
You were able to give us and I know this is an early and rough approximation of the potential impact to the business. Did the CAC spell out what they mean by critical information infrastructure? The end markets, at least that come to my mind, is China data center and enterprise markets. Did they provide you a list of end customer applications that fall under that definition of critical information infrastructure?
We do not have detail at this time. Again, we took what we believe was a reasonable and sort of classic definition of what that could be. At the lower end of that range, you know, you can envision that's, you know, principally networking companies.
Mm-hmm.
At the higher end of the range, it's, some different markets, but what is not included in our range, are, for example, smartphones.
Yeah.
We tried to draw the line, you know, and, you know, range this in a way that we felt, was reasonable.
Have any of your customers in China already communicated with Micron on what types of products they'll stop procuring from the team, or are you still gathering that information at this point?
As of the last information I have, no. Again, this is very new developing situation, and have nothing to report on that at this time.
What's the next step from here as it relates to your dialogue with the CAC team?
Oh, we continue to work constructively-
Mm-hmm.
Respectfully with the CAC, provide them any additional information they would need, work to clarify, what the security concerns are with our products and, you know, again, just, work to do all we can to, serve China in the proper way.
My final question here, and obviously I know the team will keep us up to date on the situation as it becomes more clear. My final question here is, Did the CAC decision, in your view, given the trends in the business, is the CAC decision going to impact the May quarter?
It's of course happened very late in the quarter.
Mm-hmm.
You know, as a result, we have no update on the quarter. The quarter was trending in a good way, and again, this happened at the end of the quarter. Now, in the August quarter, you know, we need to take stock of how this will play out, so we will provide an update on the August quarter during the earnings call in June. Nothing, but as far as the May quarter, no update to our previous guidance.
Well, I'd like to move on to some of your commentary on the fundamental environment, but I do think the CAC thing is important. Before I move on to the business fundamentals, are there any questions in the audience about the CAC review and decision? Any other prepared commentaries by Mark? If you do, please raise your hand. Perfect. Let's start off at a high level. We'll go into the detailed end market dynamics a bit, but obviously you gave us a great update.
Sounds like things are, broadly speaking, tracking to plan, at the highest level, the, you know, the team talked about on the last earnings call, broad-based expectations of broad-based improvement in customer inventories and expectation that inventories would continue to improve as the year unfolded, driving sequential bit shipment growth in both DRAM and NAND through the second half of the year. Is that how the team continues to see it play out in the second half profile based on the demand signals, based on the inventory, excess inventory work downs?
We do. You know, and as you know, Harlan, we acted very early.
Mm-hmm.
When the downturn hit, I'd look to any other company to compare to on how well we responded. We, you know, from last summer, we began to take down CapEx utilization, expenses. We intensified those efforts through the fall. Even through the last call, we indicated taking utilization down more and spend discipline continues. You know, the, you know, the profile of the recovery, I'd say, is largely the same as we laid out last fall. Or in, I'd say, the December call.
Now what we communicated in the March call was that, you know, the, you know, the sort of recovery, so to speak or the improvements were, you know, a bit elongated and, you know, inventories, were still coming down more, in that February quarter than we expected. You know, the profile is still the same. We see customer inventories declining, albeit gradually. We see supply actions being taken across the industry. Ultimately, we believe that combination of supply working out of the system, along with continued demand growth in memory will drive memory and storage will drive, you know, a more balanced market and improved financial picture into next year.
Yeah. I wanna drill down into. I appreciate that. I wanna drill down into some of your end markets. You know, your PC and mobile business started to see the weakness first last year. You've been working down inventories, expected to see those inventories improve through the year. You expected PC and mobile bit shipments to improve in the second half of the year as well. Most semi companies, I think, have seen, you know, that have PC exposure, reported recently that, yeah, they are seeing signs of, you know, second half improvement. However, it looks like the component inventory burn in mobile is taking a bit longer. Talk about the PC profile. Has the mobile profile changed at all relative to your expectations back at earnings?
Yeah. PCs and mobiles are sort of ground zero for all of this.
Mm-hmm.
Last year, and, got hit very hard in the summer. You know, the inventories have been getting worked down in PCs and smartphones for a long time now. Actually the PCs and smartphones inventory levels are, we believe, healthier than, for example, data center.
Yep.
You know, unfortunately, though, and, you know, the market is still finding its footing.
Yep.
In fact, in the, in our March call for the February quarter, we actually called down calendar 2023 outlook for both PCs and smartphones. We, we believe now that PCs are gonna be down mid-single digits when before we had believed that they would just be down slightly. Then smartphones, we now believe, will be down slightly when we thought they'd be up, flat to up slightly. Those markets are still soft, though, you know, hopeful that when demand improves, and eventually replacement cycles and things will kick in, when demand does improve, the inventory levels are such that, you know, we'll see that positive effect.
The data center was later to see the spending weakness, as you pointed out, and team took proactive measures on the supply side. You called the February quarter the bottom in your data center business and expected revenue growth, this quarter, May quarter, and a return to normalized customer inventories exiting this calendar year. You know, data center has been relatively mixed through this past earnings season. Curious to hear if the team's view for this segment is still relatively unchanged and if you anticipate sequential bit shipment growth in data center through the second half of this year.
Yeah, there's no change from our last update. The data center was a disappointment in our February quarter.
Mm-hmm.
I mean, it was much weaker than we expected. It was just customer inventory levels coming down more than anticipated. You know, was that related to a pause in hyperscale, you know, spend? Was it related to enterprise-related server? You know, not clear. You know, it was very weak. We expect, you know, we see May quarter stronger, we believe, you know, stronger second half. You know, that will help the inventories, you know, reach relatively healthy levels by the end of the calendar year.
You know, you mentioned this in your prepared remarks, but, you know, the team has reacted exactly as we would have expected, right? From a supply discipline perspective, cutting utilizations, cutting wafer equipment spending, holding back bit shipments on current output, and even slowing the pace of technology migration. Your industry assumptions on supply and overall supply-demand normalization assumes that all of the major memory players act in a similar fashion, right? Seems like they are from a CapEx and wafer start perspective. But based on the metrics you monitor in your end markets, the metrics that you monitor in your channels, are your competitors being disciplined from a pricing perspective as well, and any differences in DRAM versus NAND?
Well, we've said, and you can see it in the results, that the pricing has been really challenging for both DRAM and NAND. It's probably a bit worse in NAND 'cause the inventory situation's worse in NAND. We do expect, you know, an improving trajectory in the back half of the year. We expect the rate of decline to slow. Eventually, prices will firm up and increase, we believe. You know, already spot pricing, just due to all the supply actions, you're seeing you know, some little positive movement in spot.
Mm-hmm.
Pricing here and there, which is good and early indications, and that causes customers to take note and think about, you know, replenishment of their inventories and creates, you know, hopefully eventually a virtuous cycle as vicious as this last cycle has been. We are just continuing to work hard to drive the best pricing we can, you know, while also bringing value to our customers. You know, but in the end, you know, we're focused on what we can control, being agile and smart in the marketplace about where we put our bets and the value that we get for our bets. Reducing, you know, supply on our end, and right-sizing our inventory.
You know, we expect customers to continue to work their inventories down. You know, we've talked about seeing better levels in the second half. You know, I think we've done a pretty good job so far.
Mm-hmm.
We just gotta stay focused.
You talked about some green shoots in pricing that you've seen, some stabilization in pricing and in certain cases, maybe even some near-term pricing improvements. I'm wondering if you can just maybe sort of give us a snapshot of what are some of the end market or end applications where you're starting to see these green shoots of stabilization.
I think the clearest has been DDR5.
Mm-hmm.
You know, there's the, there's the new compute platform rollout, and then the supply is a bit better balanced in that market. You know, we have a, we have an outstanding, you know, DDR5 product, you know, on our 1-beta and, you know, that product, you know, our 1-beta technology, has over 35% greater bit density, has 15% better power consumption than 1-alpha. We're, you know, that's gonna be a good product for us, and the market is, you know, healthy, so to speak.
Any questions from the audience? I'm gonna move on to a little bit more maybe mid to longer term dynamics. Any questions? Raise your hand if you do. The adoption of large language models has really catalyzed an AI arms race amongst the cloud titans, right? While they're all trying to figure out how to monetize AI, they're all spending on compute, networking, storage, and memory infrastructure, right? AI-based servers represents only about kinda mid-single digits of total server shipments based on our internal analysis. An AI-based server carries anywhere from 3x-5x more DRAM and NAND content versus a general purpose cloud server. Is this AI arms race having a tangible impact on driving a faster work down of excess inventories in your data center market? Does it change your long-term growth assumptions around DRAM and NAND?
Yeah, it's a super exciting area. It's hard to not even the opening presentation.
Right.
Here, it was heavy on the AI. It's hard not to get sort of sucked up into it all. We do hope it sucks up a lot of memory. The, you know, the, you know, the reality is we knew that AI would be driving growth. In fact, if you go back to our Investor Day, we talked quite a bit about the mix shift it would create for our business, the growth potential for our business. AI is very memory intensive, you know, applications. That's both on the training side and on inferencing. You know, it's going to be a strong growth factor for the business.
You know, since we were aware of it and had built it into our forecast, we're right now trying to discern whether all the excitement is, you know, is it incremental, and you know, how large is that incremental change to our outlook. I think, I think it's safe to say our view is probably neutral to positive, but how much we're still assessing. We may talk more about it on, you know, on the earnings call. You know, one other note on AI, and it's gonna drive, you know, high-bandwidth memory.
Mm-hmm.
It's gonna drive DDR5 product volume now. On HBM, that's a very small portion of revenues right now. you know, but we have a good product that's gonna be qualified later this year, that we think it puts us in a great position as that market grows out for AI. I mentioned DDR5 earlier. you know, 1-beta is designed for DDR5.
Yeah.
You know, we just think we're really excited about the quality of that product built on the, you know, high-value technology that we have in 1-beta.
From a technology position, the team has done an outstanding job, right? You've got a strong position in DRAM and NAND. First to market 1-beta DRAM, first to market with 200+ layer NAND. In response to the supply-demand challenges, I mean, you've slowed the ramp of your 1-beta, 232-layer NAND, and pushing out technology roadmap for subsequent generations of DRAM and NAND technologies. Bears would argue that competitors will use this down cycle as an opportunity to close the gap with Micron, right? How does the team retain its technology leadership given the near-term slowing down of technology migrations? Is it just as simple as, you know, we start to focus more on product performance, differentiation, and higher value-added products? How do you guys think about that?
You gotta start with we don't need the bits.
Mm-hmm.
The technology's gonna drive more bits. Our priority at the moment is to get supply-demand back in balance. You know, we've got inventories to clear. We have good. You know, for example, our 1-alpha is a very good node, so we've got good inventories and we're in a good position with that. You know, that's the first concern, is getting the market back, you know, healthy, inventories down, eventually utilization up. Now, as you noted, we did take the decision to slow the ramp of 1-beta and 232-layer.
Mm-hmm.
As luck would have it, this downturn hit just as we were, you know, installing tools and beginning to ramp two great nodes. You know, both 1-beta and 232-layer NAND. On 1-beta, as I mentioned, you know, over 35% better bit density, 15% more power efficient. You know, it's a fantastic node, and we're excited when it does ramp. We're already getting tremendous yields on it because we've had, you know, lines running and getting yields up in low volume periods. When we're ready to ramp, that thing's gonna ramp very efficiently. On 232-layer, same story in a way, but a NAND of course. It's also a very good node. You know, double the write bandwidth, and nearly double the read bandwidth. 50%, you know, better, faster data transfer than 176-layer, which was a good node too.
Again, and as with 1-beta, because we've been able to, you know, work on manufacturing efficiencies, its production yields are already very high. In fact, both of them achieved, you know, high yields in record time. Again, we're in a position with that node when volumes pick up, when our inventories are at a good level, we'll be able to ramp that very capital efficiently. You know, so we're working to be capital efficient in the use of, you know, technology advancements. You know, we, you know, we believe that we can sustain our technology position.
Yeah.
We're, you know, you can see we're still spending. Yeah, nearly $3 billion on R&D, you know, even in this down period. You have some things to look for. We're still, you know, for example, working on 1-gamma in DRAM. As we said on the last call, that's where we see EUV being required. There was a lot of discussion about whether we would be ready for EUV. We were able to delay it because of our multi-patterning technology. We have EUV tools in Taiwan and in Idaho. We've been working with EUV. In fact, we inserted it into one of the layers in our 1-alpha node. We've been able to demonstrate it in production and are getting good yields. Of course, it's higher cost. You know, we will delay, you know, deployment in the technology until we, you know, in the principal flow until we need it.
That's just one example of, again, how we're continuing to invest in technology and advancing it, staying ahead.
Any questions from the audience? Right up here.
Hey, thank you for all the comments. I just wanted to clarify on the DDR5 ramp. How should we think about how you might be influenced by some of the competitors in this space, and how you might react to if one of those were to perceive to be ahead at any point? Just over the medium term, how the interaction between you and competitors in ramping that space, please? Thank you.
I mean, we have an excellent product and, you know, we're, we've, you know, changed some of our manufacturing, configuration to be able to produce more DDR5. You know, we would expect to, you know, gain, you know, have our fair share or more of that product. We've got an excellent product.
I have a question that came from one of our clients that's listening in. Given your discussions on end market demand and customer inventory trends, does the team still anticipate sequential improvements in gross margins through the year?
Yeah, we've not given gross margin guidance out through the year. We've said that, you know, profitability will be challenged.
Mm-hmm.
You know, at these pricing levels. You know, as the supply balance improves, supply-demand balance improves, you know, we would expect, you know, gross margins to improve. We've said that we expect to return to operating margin profitability, within our FY 2024.
Okay. One last question. Shifting to your cross-cycle financial model you had laid out at your Analyst Day, you know, revenue CAGR target of high single-digit % growth, operating margin 30%, EBITDA margin low 50%, capital intensity mid-30s, free cash flow margins greater than 10%. With the memory downturn in full force, obviously, which wasn't the case at the time. I mean, what is your confidence level at these cross-cycle targets?
We're confident that we'll deliver results in line with the cross-cycle model over the long term. As you said, you know, the downturn was severe, caused by a confluence of unprecedented factors. It's gonna take some time for that to normalize. In the near term, we're making the investments in the new fabs, you know, and that will pressure free cash flow a bit on the cross-cycle model. Longer term, you know, we're confident the profitability of, and growth, of memory and storage. It serves such a crucial role in the data economy. And, you know, and as evidenced by our comments on the last call for a record TAM in calendar 25.
Perfect. Well, we're just about out of time. Mark, I wanna thank you for your participation and all of your insights. Thank you very much. We'll look forward to catching up with you at earnings.
Thank you, Harlan.
Yep. Thank you.