Micron Technology Aktie

Hello, everyone. Thank you for joining us, and welcome to Micron's Technology and Product Leadership Webinar. [Operator Instructions]

I will now hand the conference over to Satya Kumar, Corporate Vice President, Investor Relations and Treasury. Satya, please go ahead.

Thank you, Christine, and welcome to Micron Technology's webinar on Technology and Products Leadership for the AI era and beyond.

On the call with me today is Dr. Scott DeBoer, President and Chief Technology and Products Officer. AI is driving significant changes in memory and storage requirements. And we're excited to have Scott here with us today to provide investors a deeper look at our technology road map and innovation strategy.

Before we begin, I want to remind everyone that today's discussion includes forward-looking statements, including regarding our technology and product road maps, investments and business opportunities. These statements are subject to risks and uncertainties that could cause actual results to differ materially. Please refer to our SEC filings for a discussion of these risks.

With that, let me turn it over to Scott.

All right. Thanks, Satya, and good morning, everyone. Thanks for joining us today for this pivotal moment in our industry where AI is really changing so rapidly and redefining computing. Today, I'm going to talk about how Micron is shaping that future and making it real at scale. The next era of AI will definitely be won by advancing memory, storage and compute together. And Micron is in a unique and strong position, leading in bandwidth, capacity, power efficiency, reliability and increasingly focused on supporting our customers to execute their extremely complex systems going forward.

So today, I'm going to talk on a number of points and try to give you a feel for how Micron is positioned. And definitely, we are in an unprecedented position in time with an opportunity for memory and storage in particular. And through this first stage of the AI transformation, Micron has executed and delivered extremely well. Micron's technology leadership really is foundational. And I'll talk today about that foundation and how it's going to evolve through time. Truly, the differentiated industry-leading products that we have today and that we'll have going forward are all built on a core technology leadership position, and that's a focus for Micron going forward.

In addition to the technologies that we're creating in our products, AI also accelerates our team members' ability to innovate and execute. And it is a major focus of Micron, and I think we are leading the industry in our implementation of AI products and technology capabilities across our whole development road map, really from starting physics through design, through ultimately reducing variation in manufacturing and using smart manufacturing across our ecosystem to ramp faster than we've ever done before.

I'll also touch today on our longer horizon R&D investments that we're focused on to really enhance Micron's future competitiveness. We're making substantial investments and really changing the way that we're looking at longer-term R&D investments from many aspects, and I'll give you a feel for some of the things that we're doing there.

I've said before several times, and we've emphasized that co-innovation really is a key part and it's very different today than it was in the past. Micron's co-innovation with our partners, both suppliers and customers really are a core piece of how we deliver these leadership solutions and help Micron position itself with the optimal memory and storage to support our customers' next-generation products.

All of this adds up to Micron being positioned to really extend our technology and product leadership for the next decade, and we're uniquely positioned, I think, with the strongest opportunity that we've ever had and probably the farthest extended road map in terms of technology and advanced products that I've ever seen in my career, truly a unique period in time.

So first, just to touch on a few background pieces relative to where we're at today. The products that we have focused on and are underpinning Micron's success up to this point really have been training focused and through the first part of this inferencing wave. And really, there's a wave of AI increases in memory demand that continually come at us, and we're still in the early innings of the adoption and the evolution of the AI systems.

And each of these is going to drive significant changes in usage and demand on the memory products, not just in terms of bits, but definitely in terms of capability and what type of products in the memory and storage area that we build. And it's really an interesting and unique period of time where Micron's technology leveraged -- our technology position can be leveraged to really position the company stronger going forward.

When we look across the architecture that supports these waves of AI transformation, Micron's portfolio really goes across all of them. It goes across all of them through time with different types of products, different types of innovations that we focus on with our customers and really spans the portfolio from HBM today in terms of ultra-high-performance memory, evolving to custom and then future vertical HBM solutions that I'll talk about today, supporting our customers' needs at the very highest performance levels.

In addition to that, we've seen over the past couple of years, really the importance, the growing importance of high-capacity DDR modules and the exceptional performance that's provided by low-power DRAM in the data center. And that continues and expands through the next several years. And our focus, as I'll talk about today, is on making sure that the next generation of both DDR and low-power DRAM modules are truly optimized for what's coming ahead in the AI world. We've had great progress on the SSD front in terms of data center SSDs. The business and technical teams have really delivered strongly over this past couple of years, and that's been noted in our earnings calls and in several other public disclosures relative to our progress on data center SSDs.

Today, I'm going to talk a little bit more about what's next in our focus really in enabling AI-class NAND, which is higher performance, different varieties of higher-performance NAND products that support opportunities that our customers are looking at in terms of leveraging that to get better and better performance out of the end systems. And finally, high-capacity SSDs as we leverage our technology leadership and our cost structure -- cost per bit structure of our advanced nodes, higher capacity SSDs continue to be a focus and continue to be a key important point for our data center customers.

But memory is not all just about data center and the technology and AI implementation into multiple other spaces is critically important as well. And a few of those on this slide noted in terms of autonomous vehicles, robotics and of course, AI PCs and mobile, all evolving with AI and all driving unique requirements from the memory technology. So what is great AI technology for data center may not be optimized and truly is not optimal for these other edge applications. And we're focused on new types of products across all these edge spaces and looking at the opportunities with our customers to build optimal memory across each of these different types of non-data center focus areas going forward.

So if we look at that first wave, we've had tremendous success and building on core technology leadership to enable a number of industry firsts and best-in-class products. And these really span across the product portfolio that I mentioned on the previous slide, driving LPDRAM in unique ways to have industry-first and sole-source positions with LP-based products. The SOCAMM modules leveraging LPDRAM, SOCAMM2 modules, leveraging LPDRAM have come out and really been a key driver of that part of our business. We've been a leader, of course, in SOCAMM2 and driven the highest densities available in the market with that.

On the data center SSD side, we've leveraged our technology leadership and really a long-term partnership, both with customers and an internal focus with our SSD team to enable PCIe Gen6 first in industry and deliver it into mass production. Micron has come a very long way from just 4 or 5 years ago in terms of our position in really being at the forefront of data center SSDs, and that's something that we're very focused on, and I'll talk a little bit more about today how we're going to continue that path.

This slide highlights a number of other industry firsts that are of critical importance. And of course, HBM4 strength is one of those. And as we've mentioned, we're in high-volume manufacturing with HBM4 now, very successful product ramp, and a world-class product there and moving on to HBM4E that I'll talk about in the rest of the road map as well.

And then one of the most recent accomplishments noted on this slide is our high-density RDIMMs. We've sampled a 512-gigabyte DDR5 RDIMM recently. It's the world's first module of this sort with exceptional performance and positioning for high-density module optimization and usage by our customers, again, coming from a position a number of years ago where we were not in the leadership role and have put significant focus over the last several years in making sure that every aspect of this product portfolio is intentionally positioned to take advantage of opportunities with our customers.

There is no question as we look at the waves of AI coming towards us that inferencing raises the bar on memory technology. And memory technology is fundamental to how AI works. And I think it's relevant here to just talk about all the touch points as we go through what happens and how memory evolves into this and engages into the AI ecosystem, starting with just the example of a simple prompt in the context of information being stored in DRAM and coming from storage SSDs for that prompt, holding the background information.

And then the LLM inference really reading that prompt and utilizing low-power DRAM to optimize energy efficiency and working with capacity to drive the prompt into understanding what the answer is to that question. And then going through a number of calculations through the KV cache and then really having the HBM engine drive the data back and forth with the GPU to get to that answer, which is then generated one token at a time. And those tokens live in memory and memory is the foundational piece of the ultimate performance of the system really driving and living in the memory.

So the position of memory has never been more important as it is relative to AI today, and this opportunity just keeps getting bigger as the models and the capability continue to grow.

So when we look at the pieces of how memory drives and continues to scale to meet these waves of oncoming AI opportunity, we have to foundationally continue to scale memory. And I'll talk today about our confidence in continuing to be able to drive memory technology to meet this challenge. Absolutely, Micron has a history of successful scaling, and I'll talk a little bit about some of the sustained technology leadership we have on the next slide.

This slide just highlights the foundation that we're building on. And it is a little self-serving since 1995 is when I started at Micron. So I made the graph start in 1995, but 22 nodes of DRAM technology since 1995 to today, ending in our 1-gamma node, which is a very high-volume leadership node in the industry today. And we're building on that to really ramp this next generation of AI products over the next year. That will be our main running node. And for context, this graph represents about 400x as stated here, more bits on a DRAM wafer since 1995.

When you look at this 1-gamma node, which is the workhorse of Micron on the DRAM side now, positions itself with a kind of a unique understandable explanation for how small it truly is in fitting more than 1,500 capacitors across the width of a human hair on this node. Micron's leadership, as we've talked about in multiple forums, has been sustained over the past multiple nodes, really 4 nodes of DRAM consecutive leadership, 3 nodes of HBM generations in power leadership and NAND focusing on delivering 3 consecutive leadership nodes from G7 to G9.

And today, I'm going to talk about what comes after these, but this accomplishment really is a result of many focus areas across Micron of our outstanding technology and innovation team driving technology across many fronts and executing flawlessly over a number of years to position Micron with this leadership. And we absolutely don't take this leadership for granted. We're focused on making sure that this execution continues and that we truly continue to leverage this global world-class team to enable leadership in DRAM, HBM and NAND going forward.

One last time I'm going to use this slide, I've used it at a couple of external events, but I really like the illustration of the pace of innovation in memory relative to other things. And as a look at a Porsche from 1995 with 270 horsepower and cost about $60,000 and had 23 miles per gallon. The illustration of what the pace of the memory industry versus the automotive industry really would be if automotive had matched the pace of the DRAM industry. This vehicle would now have about 2.7 million horsepower, cost $5 and have 23,000 miles per gallon if it were capable or if the auto industry had moved as fast as the DRAM industry. So really unique piece of how fast things move in this -- in the technology cadence across memory.

Another interesting piece here before I shift to the technology road maps is really the piece of recognizing how outstanding the performance of HBM4 truly is. And this just gives some context for exactly how powerful this device is, how complex this device is with 1 HBM4 stack now delivering 2.8 terabytes per second of bandwidth and looking at the total Internet bandwidth in the world right now, this is across the Internet Google data. Worldwide Internet bandwidth is 282 terabytes per second total. So about 100 HBM4 stacks can match the worldwide Internet bandwidth.

Okay. But as we all know, memory is currently a focus. And as we've discussed, memory is at the center of AI, and it is absolutely the limiter in terms of ultimate system performance and capability at this point, and we take that very seriously. As memory has become the AI bottleneck, our focus on making sure that memory capability continues to evolve, memory architecture continues to evolve. Both of these things must happen, and I'll talk about our road map to deliver that. Ultimately, memory shapes AI capacity and performance and cost per token. So our focus is on working with customers to make sure that we deliver the optimal product solutions to make that happen.

So if we look at now stepping into Micron's technology position, and again, it really is a price of entry situation. Technology leadership absolutely underpins our ability to deliver the differentiated industry-leading products. It's been core to the previous ones over the past couple of years and will continue to be really that key foundation that underpins the next set of products that are coming along. So I'll step into exactly where we're going with core technology. So first, just a general point on memory technology and the evolution.

On the DRAM side, planar DRAM has been scaling for decades, and Micron has been in a strong position, as mentioned. It's foundationally shrinking the XY dimensions of the DRAM. Of course, it's a very vertical process now. So there's 3D aspects to it, but we don't call it 3D DRAM at the moment. It is just a function of making the chip -- the cell smaller and smaller. DRAM is nearing atomic limits at this point, although we have confidence in several more planar nodes. We are very focused on making sure that we introduce 3D DRAM at the optimal time from a cost and performance point of view. So our intention, as I'll talk more today is truly around making sure we have 3D DRAM prepared and that we'll implement it at the proper time for the market.

On the NAND side, I think everyone is familiar that planar NAND ended scaling in around mid-2015-ish time frame. Micron scaled that to about a 16-nanometer planar node before it wasn't capable of scaling any longer and successfully transitioned into 3D NAND, which was a function of primarily Z scaling, adding more layers to 3D NAND across multiple nodes.

The current industry direction and Micron's direction is scaling in both by adding layers, but also by shrinking the cell dimensions. And that complexity will continue to provide benefits in terms of cost structure and performance over the next several nodes. And then we'll be looking at future alternatives, which I'll talk a little bit more about on some upcoming slides. But again, confidence in both the DRAM and the NAND scaling path. And I think from a visibility point of view, have more confidence in the next several years of scaling on both DRAM and NAND than we have in quite some time.

On the DRAM road map, we have multiple, and I'll go through the picture of exactly where we're at. As I mentioned, the 1-gamma node is in high-volume manufacturing today. It's the basis of all the key products that are coming out in the 12 months and very confident in that industry-leading node. It was the node we introduced EUV at first, and that's been a successful introduction and our manufacturing team has done an outstanding job of ramping that and leveraging that capability.

On our 1-delta node, which I'll talk about starting a little bit more on the next slide, that is the node that we're focused on in Boise, but it will be U.S. manufactured as well as in other facilities. And this technology is coming along rapidly, and our intention is to ramp that into volume in the second half of next year. And then beyond that, we have a number of nodes in progress, and just we'll talk a little bit about that path. Our 1-epsilon node leverages wafer bonding and optimizes the bit growth and technology capability to deliver truly unique solutions in the AI space, and that's well underway.

And then path finding we have going on both early development for our 1-zeta node, 1-zeta angstrom node and our eta angstrom nodes beyond that, bringing in combinations of high-NA EUV, new materials and a number of other innovations to continue this planar road map successfully over a number of years. So lots of confidence -- high confidence in this DRAM road map through this period of time. For reference, we have looked at and investigated 4F2 DRAM really over the past 2 decades. Micron invented 4F2 DRAM.

The path with 4F2 DRAM has always been an option through time, and we have evaluated it and not pursued it due to increased complexity and cost through this past 2-decade period of time, but development has continued continuously through there, and we've had a number of prototype products to compare through time. Two challenges are holding Micron -- have convinced Micron not to go down this. One is that the scaling path is not viable beyond a certain point. So there's no actual benefit to going down that path.

The second piece is there's a bit of a misnomer about what actually drives the bit density per wafer on DRAM. And foundationally, it's important to understand that the capacitor pitch, the capacitor size and dimensions is actually what sets the DRAM bit density, not whether it's 6F2, 4F2, 8F2. Everything is optimized around the capacitor pitch. And then underneath, actually, you can optimize with 6F2 or 4F2 and fit those for a given capacitor dimension, however you'd like to. So it's really about optimizing process complexity, process margin and cost relative to a given capacitor pitch, which is the foundational element of how we build DRAM.

When we look at 3D DRAM, we have that on the road map, as I mentioned, and it's shown here in between 1-zeta and 1-eta. But in reality, this is a parallel path, and we'll implement it in a fashion that is ideal for cost and complexity and device performance. And the timing of it will depend on Micron business decisions from a technology and a product point of view, it will be ready to implement in the optimal time as we've done with previous major technology transitions.

Okay. A little more on our 1-delta node. This is a focus in Boise today. The fifth consecutive industry leadership node is our anticipation with this ramp in 2027 second half. A few early mentions on capability of this node. This is going to provide in the mid-20s percentage bit density improvement, focused on optimizing costs while providing the densities that are required. The performance will be optimized to meet our HBM5, our vertical HBM and our vertical EIM products, which I'll talk about in more detail on some upcoming slides.

This node really is optimized for AI, and it's aimed to deliver the kind of power and bandwidth that is going to create those leadership products that I talked about we've done before and will enable in the future. So this ultimately is going to be a great node. The picture in the middle is a picture of the cells on this 1-delta node and we're very, very excited about this node ramping over the next year.

When we look at -- besides the node itself, we look at the DRAM CMOS logic road map, which over the past several years has been what we have leveraged to create differentiated products. So when we look at DRAM CMOS, it is not one CMOS for a given node. We -- within a given node, for example, in beta, we will have multiple variations of CMOS that we will implement into specific products on that node. So optimized for LP performance or optimized for HBM performance, we'll have CMOS that meets the requirements to deliver to the specifications that our customers need. This has become a very significant focus for Micron over the past 7 years in enabling the right technology for the right time.

And we've gone through a number of different variations that we have capability for now. These have created from 2020 through 2030 projection will create about a 10x improvement in performance in terms of drive at a given leakage current to enable the kinds of performance that we need. So we have multiple nodes of power and performance leadership and a road map that continues to deliver that through time and will become increasingly complex as we go forward with different generations of both planar high-k and then 3D CMOS to support the needs.

When we look at DRAM based on this technology and the products that we are delivering in the next generation, three that I want to highlight today are our LPDDR6 for data center and mobile, which has been a major focus from our design teams doing a great job of putting together industry-leading products here that will lead the industry in power with 35% improved power versus LP5 and more than 35% on some densities. So big, big push on LPDDR6, which is going to be a key memory type, both for mobile, obviously, but increasingly important for data center.

On the DDR6 front, Micron is positioned to be a leader in DDR6 implementation. We partner with key customers to deliver this higher bandwidth 2x versus DDR5 that can be leveraged across data centers for server applications. So lots of focus on these transitions to LP6 and DDR6. And then in the meantime, I had mentioned the 512-gigabyte RDIMM that we announced recently. The next generation will be delivering a 1-terabyte RDIMM for the industry's highest capacity RDIMM products. So many different focus areas for specific customers across our DRAM product portfolio.

When we look at HBM, we're building on our HBM leadership through from the evolution from standard HBM like our HBM4 product today to what we've talked about in terms of custom HBM, both standard and custom will coexist in the ecosystem depending on the customers. We are focused on this co-design element with our customers, co-optimization with customers to implement both memory IP and logic IP into our base die to then deliver ultimate system performance with the customers who want custom. And of course, we'll have high performance on standard HBM for those customers that don't have custom design.

As we go beyond that, we're focused still on custom HBM and standard HBM, but we will add in a product focus that's very important going forward, which we call vertical HBM. This is an implementation of memory capability directly on top of processing units and is a key element of our customer partnership going forward in delivering this solution, which provides ultimate bandwidth in combination with co-design and also still utilizing HBM around the edges of -- in many cases, of the DDR XPUs.

So supporting that execution is a number of advanced packaging innovations, all required to deliver the HBM performance and the vertical HBM performance that we see as necessary with customers. Interconnect density is really the driver of -- requirement and driver of many of the technology requirements. Ultimately, hybrid bonding technology is a major focus for Micron. Our capability is strong. This still needs to evolve through time to be prepared to implement it in a high-volume manner.

But ultimately, the driver of hybrid bonding winds up being interconnect density. So the pitch of the pads in order to get bandwidth requires closer and closer connections, spacing -- closer and closer spacing between connections. Hybrid bonding is the solution to that technical problem. In addition to the well-publicized thermal benefits of hybrid bonding, ultimately, you can't build products with a certain level of bandwidth without hybrid bonding. So it's a major focus for Micron, and we intend to be leaders in that space.

Stacking technology is still key. And ultimately, there will be implementations going forward of higher and higher stacks because memory capacity does play a major role in driving token costs down and in system -- ultimate system optimization. So this particular picture in the middle of the slide indicates 16-high HBM stacks built both with micro bumps on the left and with hybrid bonding on the right. So different implementations of 16-high products. And ultimately, the hybrid bonding in this case is beneficial, as mentioned, for both bandwidth enablement and thermal improvement.

So stacking technology continues to be a key major technology capability that Micron is focused on leading, and it's in a strong position at this moment. It becomes even stronger requirement as we look to the transition or the inclusion of 3D vertical HBM integration products in terms of enabling, as I mentioned, the tighter interconnects and thermal management. Thermal management becomes even more important as we look at vertical HBM products, which has to be optimized around heat structure, both of the base die -- heat flow, both from the base die and from the XPU that the product is mounted on.

So significant complexity requires very complex integrations with our customers, co-design and the designs have to be thermally aware. It's a key piece. One example of an innovation that Micron has implemented in the future it's in our designs now is the heat spreader, which takes a position on the HBM cube over the hottest part on the HBM base die and is effectively serves as a chimney for heat from the base up past the HBM cube. So this innovation is a key piece and the positioning and the design of that innovation to cover that fire area of the base die is a key enabler going forward.

Token efficiency focus is an urgent requirement, and it brings about a diversity of different types of products. And I'll hit on just a couple of these right now, but they're important ones. Both the vertical HBM that I mentioned, the vertical edge inference memory, both have commonality in terms of structure, in terms of the packaging requirements, the thermal requirements and the need to co-design with customers. As you look across this time line of these products coming in, one thing to note is trade ratios continue to increase as you go from left to right.

As these products come in, the trade ratios continue to be stressed in terms of the number of wafers it takes to build HBM, vertical HBM and vertical EIM relative to building standard DRAM bits. So that pressure remains through this road map. When you look at the different packaging requirements for vertical edge inference memory versus vertical HBM, while the structure looks similar in terms of the packaging capability needed, the products are optimized in different ways, where vertical EIM has lower bandwidth requirements, but substantially more sensitive to power requirements.

So it's really designed in a way to get bandwidth capability to drive these edge products in an effective way, but optimized for power. While vertical HBM, as it is implemented in parallel with other varieties of HBM, it comes in with huge bandwidth benefits. So it's very bandwidth focused and improves in capacity over time, but really bandwidth focused is the deliverable of the vertical HBM technology.

Now I'm going to shift over to NAND. We've been in a very strong position with our Gen9 NAND, and this leadership is going to extend with Generation 11. And this is a significant leap. We have optimized this memory. We've actually gone back and ensured that this memory is optimized for the AI products of the future. And intentionally, which is why it goes from Gen9 to Gen11, we've made substantial improvements that this node really brings relative to any prior NAND.

A few of those are doubling the bit density versus our Gen9 NAND, but then also a 1.5x I/O speed target and then focus on data transfer energy and really bringing that down relative to G9. So this G11 technology is the world's most advanced NAND that will ramp in the second half of next year also. It's based on wafer-on-wafer bonding technology, which is like I described before, we had mentioned we'll bring in wafer-on-wafer bonding technology at the right time when it's cost effective and it brings in capability that's important for the product specs at the time, and this is the right timing for that technology.

We'll be at more than 400 layers, and we'll give more specifics on that next year closer to the launch dates. But as mentioned, it will be ramping in the second half of 2027. And then beyond G11, we have a strong road map, I'm not going to go into great detail on it today, but really focused on new materials, densifying the cell architecture and continuing to focus on that xyz-scaling going forward. So big generational leap with our Gen9 or Gen11 NAND technology rolling out and excited to see how that really underpins the AI-focused products we have going forward.

When we look at those products going forward, really 2 focus areas. Micron's focus is to ensure that we are capable of meeting customer needs on a variety of different higher-performance NAND products. And there's really 2 focus areas here. One is on high-IOPS AI-class NAND, which is SLC-based designs implemented in small blocks to design ultrafast memory storage products. And this is a key piece that meets several emerging use cases noted here and really is a major focus for Micron in terms of getting these products out to market over the next couple of years.

On the other side of this architecture and design path is high-bandwidth AI-class NAND focused on other emerging use cases. And these ultimately are more costly and have to be implemented in partnership with customers in a way to ensure that they're meeting the mark and the cost point to drive the value proposition. But definitely enabling TSV-stacked NAND with logic base die to put closer to the accelerators. So not along the short line of the GPU as we don't think that has merit, but along closer positioning points to the compute itself to make use of this high-performance NAND and the viability of that path.

And then there's a number of -- looking at the segment view, a number of different segments that utilize a combination of those 2 architecture choices. As mentioned, if you go from data center to automotive, there's different cost sensitivity levels and different interactions with customers to determine which ones are appropriate and which ones we want to provide ultimately to customer demands.

Next, hitting on a couple of last topics. First, on AI accelerating AI. So how we use AI across Micron to enable innovation and execution and really position our team members with all the tools to deliver industry-leading products going forward. And a couple of examples here just to highlight where we're at today. One on data center controller design, and this is a demonstration that is confidently being executed right now, delivering this controller in 2x faster pace from silicon to tape-out. So 50% less time from our prior generation of SSD controller to the new controller utilizing a variety of AI tools.

Second example also confirmed now is developing an entire data center SSD product. In this case, a little bit faster than 2x, 60% less time in going from the prior generation to the next generation, really leveraging AI-generated firmware, root cause analysis, enabling agents to do a significant amount of the core determination of product viability and then optimizing around the path to make that product work. So besides those 2 examples, definitely focus on basic physics all the way to smart manufacturing, which are impacting every single engineering employee at Micron and making them more successful.

The next couple that I want to highlight today are around R&D investments really to enhance our future competitiveness and then a little bit about how we co-innovate with customers. So first on the long horizon R&D investments. We've announced several lately that are significant changes in how we've operated and how far we've looked out into the future, definitely ramping up our Micron Ventures efforts, and we've had some press on the increased capital commitment relative to Micron Ventures, focusing on strengthening the ecosystem engagement.

Albany NanoTech, our announced focus on advanced lithography there, really driving the high-NA EUV road map for DRAM, in particular, at Albany with our team there, focused on the next 5 to 10 years. And then the big announcement we did this summer on the $10 billion over 10 years focus on Micron Research Labs, really with Micron taking a broader position in long-term research outside of our current product road maps and shaping what's next for the memory and compute systems of the future.

And we look at an example of how we combine that for customer advantage, an example of customer collaboration really goes from a good example with WEKA and Micron here as a company that we invested in from our venture fund and have built that into a partnership in supporting them and working out business arrangements with them to help them grow. So starting from an early investment to leading into business. So really interacting all the way from a start-up like that to the focus with Meta, which is highlighted on this slide also relative to LPDDR6.

And we put out a white paper that was released this morning, really a great partnership there in optimizing LPDDR6 for the data center and in partnering with them to understand how we take advantage of the unique server capabilities and higher reliability capabilities built into LPDDR6 and optimize that together. So really from start-up through hyperscaler, engaging the customer interaction to improve Micron's future AI-related products.

Overall, as a general point of view, really across our AI co-invention opportunities, we're co-imagining, co-designing, co-developing, co-validating and then delivering at scale. And this opportunity is truly unique. It's very different than the way we interacted with customers even 3 or 4 years ago, but the opportunity to partner with customers to really understand from an early point of view, what can be done with memory products, how customers can leverage that and then ultimately building that into really having engineers working together on co-designs and solving problems together all the way to the point of getting that final qualification with the customer, co-validating it and then ultimately ramping it at scale.

Okay. So all the emphasis today hopefully has given a little bit more perspective on how Micron is positioned. We feel this is the strongest position Micron has ever been in, in terms of technology and product leadership. Partnerships with customers have never been stronger. Our connections into the ecosystem are growing, and we're going to continue to expand that. Really in a unique position and a very proud position, but absolutely one where we're focused on future execution in this growing era of complexity and making sure that we can leverage AI to deliver innovation and execution at an even faster pace.

So with that, I will conclude, and I think we're ready for questions.

[Operator Instructions] Your first question comes from the line of Vivek Arya with Bank of America.

Scott, for the very informative webinar. My question has to deal with how do you think the end customers are dealing with the fact that memory is becoming a bigger part, right, some say half or more of the bill of materials. There is some talk of destacking and making memory more workload specific.

So the kind of the specific question to you is, do you see DRAM content becoming more workload specific? Is it possible that when we look at inference workloads that the DRAM content might be lower or different than what it is in terms of training accelerators, would that be a change from what was assumed before? So just how do you, in your customer interaction, see them dealing with this rising impact of memory? And would there be a change in inference versus training DRAM content?

Sure. Thanks for the question. So I think the starting piece to this answer, and I'll go into a little more detail. Starting piece of this answer is for a given system, memory content and memory bandwidth capability will make the system operate at a higher level as it increases. And that always there are ways to optimize for systems, but ultimately, the system will work better with increasing memory content and increasing memory performance.

So as we work with our individual customers to create the products that fit with their system design through time, they will, of course, make decisions relative to what the minimum system requirements are, and they'll optimize relative to shipping their products and their business as well. But the underlying piece is their product will work better with the highest memory bandwidth and highest memory content. So eventually, this will evolve through time.

And as we've said, we're focused on both creating -- increasing our supply with all the investments that Manish and we've talked about through earnings and other times, public times, but also through making sure our products can deliver higher and higher bandwidth through a number of these things like vertical HBM and vertical EIM so that the systems can be optimized.

And for sure, if you look across the space of products, edge inference, mobile products, different kinds of small model products will have different memory optimizations. But in general, the underlying piece here is memory is still core to AI capability and AI system capability, and we'll continue to press to -- customers will continue to press to optimize that. And inherently, more memory does equal better system performance.

Your next question comes from the line of Timothy Arcuri with UBS Securities.

Scott, you said that there's no viable scaling path for 4F2. Can you just double-click on that? Why would -- we're hearing other competitors pushing this? And why would they want to do that? And like why do you think that there's no scaling path forward? Is it like an equipment readiness issue? And maybe do you think that there will be a time for 4F2? And if so, when?

Yes. So I can't comment on why other customers or other suppliers would do things. I can only say from our investigation of this technology over the past couple of decades. First off, it's a bit of a myth that 4F2 actually provides anything versus 6F2. It's the capacitor pitch that defines the DRAM. You can build the cells, the transistors underlying in different ways, but they still have to fit under the capacitor pitch, which sets the DRAM. So some of that is just kind of smoke and mirrors in terms of talking about 4F2 versus 6F2.

Now when you look at a scaling path, you can go back a few nodes and absolutely, 4F2 can be built back a few nodes. No question about it. It's more complicated. It's more complex, it's more costly, but you can do that. When you look out from Micron's point of view as the leading edge technology company, when we look at scaling 4F2 beyond where we're sitting today, it's physics. There's fundamental problems with 4F2 scaling that ultimately will lead to 3D DRAM coming in, instead of 4F2.

Your next question comes from the line of Ben Bajarin with Creative Strategies.

So Scott, just a quick one on how you think about memory road maps as it relates to next-generation compute scaling for racks, meaning more of the compute domain that everybody is talking about. So not just having 1 rack to run a model, but networking those together to 2 or 3 or 4 and how that might impact memory when they can run larger models within compute racks networked as compute domain?

So I think there's going to be a lot of different configurations of compute depending on the customers. And our focus is making sure we have the products that can optimize in any of those domains. As we've said, the next few years, we don't see a period of time where memory is not going to be in short supply in the next 2 years, as Sanjay said, we actually don't see a time when it comes into balance. So our focus, whether it is in these different kinds of configurations or in the leading-edge configurations of data centers today or even the edge and other devices that we create the types of memory products with the bits that we have that optimize the solution for those customers.

So I think there are going to be different kinds of solutions. They are all going to be memory intensive. And whether we optimize -- whether our optimized LP solutions, LP6 solutions or DDR6 solutions are required in bigger volumes for some configurations or other configurations require high-density or medium density products or more HBM. I think we just are going to leverage our close collaboration and connections with customers to be sure we deliver -- we have the products that they need for how they're going to build their racks.

Your next question comes from the line of Srini Pajjuri with Royal Bank of Canada.

My question is on the base die for HBM. I know you guys talked about moving to a foundry process node in the future. I'm just curious, I know you didn't announce exactly what process node, but it does seem like it might need an advanced process node going forward. And NVIDIA talked about potentially moving the memory controller into the base die. So if you could talk to us about what makes potentially the advanced process node necessary for the base die?

And then as we go vertical, do you see anything changing as far as the base die is concerned? Or do you think that the advanced process is still kind of necessary depending on, I don't know what kind of functionality I guess, get into base die versus the GPU XPU in the future when we go vertically. Just trying to understand how you see the road map going forward.

Sure. So to start with the base die in the custom sense is absolutely a decision jointly with the customer that we build the custom product for. So in the example that -- actually, I won't use a specific customer, for example. But when we co-design a base die and the certain logic elements that are desired to be put in that base die are generally an offload from GPU circuitry anyway onto the base die to put it closer to the memory. And so the best leverage is to utilize the advanced node that's appropriate for the customer.

We have flexibility generally to a certain extent to optimize the memory IP around -- certainly around the more advanced node, but we would optimize it around the most cost-effective node that meets the performance. And going forward, as we said, we're on both our standard JEDEC design products and our custom products, we'll be using foundry node for both those base dies. And we haven't stated what nodes we're using. We're using advanced nodes for both to enable capabilities. Really, it's around power and capability into the base dies.

So as we've talked about, and I think I talked a lot about this time last year, for HBM4, optimal power and performance didn't require that. Some competitors used it. Our HBM4 didn't. And it has power performance strength as we said it would at that time. But for HBM4E, the performance, the speed and the power optimization does require more advanced CMOS. So when we look out to vertical HBM, it will again have functionality in the base that connects directly to the GPU.

And it will provide -- so it will still be advanced nodes in the base on vertical HBM, and it will have increasing functionality in that base die that can be leveraged there that connects directly to the GPU, again, optimizing for quality -- certainly quality and reliability of the HBM product itself, but also optimizing for ultimate system performance, power and leveraging advanced silicon to optimize the overall efficiency of what's on the accelerator chip below it. So moving what makes more sense onto the base die will still be a reality in vertical HBM time frame as well.

Your next question comes from the line of Daniel Newman with Futurum.

Scott, I thought your slide where you showed sort of the AI increases on memory demand across training inference and agentic and then physical really told a very interesting story about sort of how Micron needs to both align and be more strategic with its partners in order to scale. But in order to meet what we expect to see continued growth across all these areas across inference, agentic and physical AI, sort of what are some of the key areas of innovating, partnering and building processes in both serial and parallel that are going to enable you to keep up and deliver on sort of the insatiable demand that continues to grow?

Yes, that's a good question. Thank you and definitely one that we've put a lot of focus into.

So I think a couple of things that I would point out is one, although I only have one slide and a couple of examples on AI across our execution process, it is a major focus across the company. And that improvement of 2x time line, 2x shorter time to execution from innovation to high-volume manufacturing, that is a key piece that we're well down the road on and is a major focus, but also a key piece that we are depending on to increase our bandwidth and being able to cover the kinds of solutions coming at us from the whole ecosystem.

Second piece is very focused and targeted product development and alignment with customers. And then in parallel, on the technology front, enhancing our partnerships with key suppliers to make sure we're delivering specific capabilities. So whether it's across lithography areas or across packaging, hybrid bonding enablement or film processing, different things with key partners that we have need for enablement and we can see what to deliver certain aspects of the products that our customers need, we're strongly focused on deepening those partnerships. So all the way across our supply chain and then up to our customers, making sure that we just focus very extensively on execution of this wide variety of things coming at us and the bandwidth that we have across our engineering team increases substantially through the use of AI tools.

Your next question comes from the line of Atif Malik with Citi.

Scott and Satya, for putting this together, especially the 3D DRAM versus 4F2 road map update.

My question is on the physical AI or near memory where Micron is also taking a different approach with edge inference memory, EIM, which involves stacking low-power LPDDR on top of the logic die versus peers' processing in memory, which embeds computing units inside the memory. Scott, can you talk about the merits of your approach versus peers?

Well, I can say that actually, we have both product directions in flight. The performance of -- so it's a bit of a balance. So when you look at processing in memory, there are certain capabilities that can be built in cost effectively into LPDRAM. And that balance and the system complexity of that is one set of considerations that we work with our customers on. But absolutely can say that we have processing in memory solutions in flight and in discussion with customers as well.

The performance that we get from our vertical edge inference memory is just substantially higher. So it reaches a point where the customer needs more bandwidth than can be accomplished through standard PIM type capabilities. And then a capacity that is also beneficial in these edge opportunities. So there -- it again is -- and we can also reach lower powers because of the use of a base die with LP and the logic and memory IP that can be built into that base die, we could have a substantially higher performing product for edge and vertical edge inference memory than we can with standard PIM inside of an LP.

So it really is down to the customer interaction and customer collaboration to what type of solution benefits them most, and then we optimize the memory around that. Edge inference memory can provide a higher point of performance, which I think I had a few examples on the slide. But that can be tweaked with cost structure depending on how far the customer wants to go in terms of how much bandwidth and how much power.

There's really a spectrum all the way from those numbers of sub-picojoule per bit kind of powers and those bandwidths that are good enough for edge devices, the spectrum really goes from there all the way up to HBM. Since the structure is similar, it's more about segmentation and design of the DRAMs than it is about the core technology. Core technology is largely common. It's more of a design element difference between that edge memory and the vertical HBM, and we can optimize anywhere in between there depending on the customer needs. So it's really a platform that can be leveraged to meet different kinds of applications.

Our last question comes from the line of Mehdi Hosseini with Susquehanna International Group.

Just I want to go back to segmentation of HBM. You highlighted introduction of vHBM. I wanted to double click on this. Can you help me understand how bandwidth is necessitating the vHBM? And how should we think about the trade-off between bandwidth and density? I think there are some confusion. And in my opinion, bandwidth is priority #1 and perhaps vHBM would give you that extra bandwidth and this is why the diversification, but I would like to hear how you see especially as it relates to the underlying technology?

Sure. Thanks, Mehdi. So the key piece with the HBM is bandwidth, as you point out. That is the enabler. If you think about the kind of bandwidth that can be accomplished and the power levels that can be accomplished by putting the structure directly on top of an XPU, that is the piece that can't really be done in a different way in terms of that much bandwidth gained.

Now capacity, of course, is more complicated directly on top of GPU because it's a much more difficult thermal management challenge to go to multiple, multiple die stacks, although you have to go to higher capacities on vertical HBM as well, it doesn't start out as a product that can be built with a large number of die stacks and actually have the power and performance benefit in thermal management.

So capacity scales at a different pace there. And the system product, ultimately for most systems or for many systems will be HBM5E in combination with vertical HBM to get the bandwidth and the system will be optimized around both those things. So capacity optimization is key, and we're working hard to get the capacity up on vertical HBM. But these things -- there's also the interplay of these things with trade ratio.

So when you look at vertical HBM, it again has a very high trade ratio because of -- both because of the stacking, but also because of the optimization for bandwidth inherently makes the number of I/Os much larger and which inherently reduces the die per wafer on the DRAM. So it's another balance point to all this. But the system vertical HBM ultimately is -- it's going to be a huge benefit to our customers as it comes in and provides primarily the bandwidth boost that you mentioned.

There are no further questions at this time. I will now turn the call back to Scott DeBoer, President, Chief Technology and Products Officer, for closing remarks.

Okay. Thank you for the opportunity to talk with you all today and good questions.

Just wanted to reemphasize that we believe Micron is at its strongest position in history at this point in time. The technology underpinning our products is a key strength of Micron, and we're very focused on the execution it takes to continue that technology leadership. Our products built on that core technology are enabled in a very strong way with our customer collaboration, and we're set up to win in that space due to the depth of partnerships that we're building and have established. And ultimately, in order to manage this complexity and the large number of customer demands, we are leading the way with AI implementation from basic physics through design and high-volume manufacturing and really leveraging that AI to make sure that our execution is stronger than ever.

So thank you very much for your time today.

This concludes today's call. Thank you for attending. You may now disconnect.

Hello, everyone. Thank you for joining us, and welcome to Micron's Fourth Quarter Post-Earnings Analyst Call. After today's prepared remarks, we will host a question-and-answer session. [Operator Instructions] I will now hand the conference over to Satya Kumar, Corporate Vice President, Investor Relations and Treasury. Satya, please go ahead.

Thank you, and welcome to Micron Technologies Fiscal Fourth Quarter 2026 Post Earnings Analyst Call. On the call with me today are Manish Bhatia, President and Chief Operating Officer; Dr. Scott DeBoer, President and Chief Technology and Product Officer; and Mark Murphy, our Chief Financial Officer.

As a reminder, the matters we're discussing today include forward-looking statements regarding market demand and supply market trends and drivers and our expected results and guidance and other matters. These forward-looking statements are subject to risks and uncertainties that may cause actual results to differ materially from the statements made today. We refer to the documents that we filed with including our most recent Form 10-K and upcoming 10-Q for a discussion of risks that may affect our results.

Although we believe that the expectations reflected in the forward-looking statements are reasonable, we cannot guarantee future results, levels of activity, performance or achievements. We are under no duty to update any of the forward-looking statements to confirm these statements to actual results. We can now open up the call for Q&A.

We will now begin the question-and-answer session. [Operator Instructions] Your first question comes from the line of Ben Reitzes from Melius.

Congrats on the promotion, Scott, I'm not sure if Manish is on, but congrats. So I wanted to ask about Congrats. So wanted to talk about 2028, new commentary here regarding and you guys felt compelled to kind of say you see things tighter in '28 than this year as well as '27. I wanted to know a little bit more about what changed and what that means for margins. You gave more margin commentary than you usually do around '27, which was great, by the way. but I just don't see any reason why margins would change in '28 from the level you implied for '27. So just a little more comment around '28 would be great.

Sure, Ben. I can start, and then if Mark wants to add anything, and thank you for the shout out. In -- in terms of 2027 and 2028 in our commentary, we are seeing stronger demand drivers than we've seen before. We commented about the server units continuing to grow into '27. And of course, I think everybody is talking about how agentic AI is really growing fast, and that's creating a CPU-driven demand stream as well. And so as we've gone through working on our fiscal year '27 and we commented that we have more than 75% of our shipments committed for the year that shows a strengthening demand and allows us to shift our conversations on allocation with our customers out to 2028.

And so that's one element of our confidence is that '27 is shaping up with stronger demand than maybe we had seen before. And then, of course, as we also have now more SCA conversations with customers, including extensions that we've now negotiated that gives us more confidence in the long term as well. And so the combination of those two things is a very strong outlook for demand through 2028.

And then on the supply side, there's still the same structural constraints on supply growth that we've talked about before, which is diminishing returns from technology transitions versus past technology transitions. HBM growing faster than conventional DRAM through 2028, which means that as a portion of the industry's output capability, HBM is growing in terms of that share. And of course, the trade ratio, not just today's trade ratio for HBM, but future more complex HBM having higher trade ratios that's also going to be constraining supply.

And then it just takes a long time for these new clean rooms that the whole industry is working on to be able to build and then qualify and then equip and it takes -- even from the time they open, it takes a few quarters for meaningful shipments to come. So that really gave us the combination of that demand outlook and this supply outlook. We made the statement that we really don't have line of sight to when supply and demand balances.

Yes, Ben, I would maybe just add to Manish's comments that as you noted in '27, we had indicated for the balance of the year that we would see margin expansion relative to Q1 a function of continued price increases, albeit at a more moderate pace. And we've said that for some time that eventually price increases would moderate. And then at that point, in addition to just price increases at a lower rate, we would -- we have a better mix of products, and we would work our mix based on our technology and product leadership.

And the market conditions, as we said, we would expect to remain tight, and that would be supportive through '28. And of course, the partial offset to price and favorable mix would be start-up costs. But these are all things we're managing. We've been managing them, and we would expect to be able to sustain strong financial performance.

Great. And if I could just sneak one other in. I mean, Sanjay had a great seat at the Trump, I guess, launch dinner, whatever, lunch. And just wondering, did you come away feeling good about the industry's ability to grow and self-regulate and the future coming out of that meeting, did that lend a hand any of the upbeat guidance he gave? Or just any color out of that and how we felt coming out and even if memory came up a lot at the meetings.

Well, Ben, Sanjay is not here, but I'll give you what conversations I mean, for sure, we're very happy to participate in that forum. And it does show the importance of memory that Micron was invited there along with the model companies as well as the accelerator companies. I think that the framework that in the white paper or the -- that was published out of that, that many of the model companies have already signed is that, that framework is constructive towards continuing to have advancement in AI infrastructure and AI, in particular, AI hardware infrastructure.

And I know that one of the concepts that got discussed quite a bit was that the way to manage some of the security aspects is to actually have security solutions, which will require more advanced hardware, including higher performance memory, lower latency memory, as part of those security solutions, higher bandwidth memory because if you can imagine, setting up a gateway to be able to manage much of the security aspects that could be proposed in the future, those -- the responsiveness of those gateways are going to depend a lot on the availability of performance and low latency members.

Your next question comes from the line of Melissa Weathers from Deutsche Bank.

My congrats as well to the 2 new promotes. On the -- and -- in past quarters, you've given us a view on where you think the HBM total TAM could be by 2028, 2030. Clearly, the pricing environment has changed since those updates. I don't know if you want to give us an official new number for where you think HBM TAM could go. But like directionally, can you help us sort of try to size like how much is coming from bets and how much is coming from pricing? And just any updated views on how big you think that market could be?

Sure, Melissa, thank you again. We are not updating that TAM outlook right now. But what we have said is that HBM shipments, we expect to grow faster than conventional DRAM. That means that HBM will continue to gain -- to grow as a portion of the industry's capacity year through 2028. And the pricing for HBM, at least for us, we did comment that we have increased that pricing significantly for calendar year '27, which will reset at the beginning of the calendar year to narrow the profitability gap with conventional DRAM. But beyond that, we haven't made any comments on the specific outlook for the HBM TAM. It continues to grow, it continues to be a very important enabler. HBM deployments continue to be an important enabler of much of the rest of the to be able to reach its potential as well. So it's an important part of the market.

Got it. Maybe following up on that, from a market share perspective, any updated view on how you guys are targeting market share for HBM. In the past, you said you want to get it to within the corporate average, kind of low 20s present market share. So is that still the case? And then I noticed in the prepared remarks, your comments on HBM 4E and the engagements with NVIDIA there. So just any more color on 2027 and HBM 4E the progress that you're making there with customers would be great. .

Sure. Sure. I'll take the first and then maybe Scott can take the HBM 4E since his team is driving that product development. In terms of the HBM market share, about a year ago, we had achieved the milestone that we talked about that our HBM market share reached our broader DRAM market share. And at that time, we said that our goals would then move around based on various different factors. And we haven't really updated exactly our HBM share target other than to say that we do expect our HBM share to be around our broader DRAM market share, but we're not necessarily targeting 1 number or the other. It will move around based on various different factors there. But as I mentioned on the prior question, it's a really important part of the market. It allows us to be very close to the leading edge of the accelerator platforms that many of our customers are designing and then deploying, and it's a key enabler for all the rest of AI to be able to deliver on its promise and its potential.

And then I can just add a little bit about the work with NVIDIA on what really will be the first major custom HBM product out in the market. And we've been working with NVIDIA for over a year on HBM 4E, what's called NV HBM. And we see substantial opportunity there for us in the co-design of that product with obviously a key customer to have this be a product that delivers really substantial value beyond standard HBM 4E. And I think it will be a real impact on industry showing where future system with optimized.

Your next question comes from the line of Atif Malik from Citi.

The first one on the 26 35% of sales by 2030. Are these for both DRAM and NAND? And if you can just split them out?

It, we're not splitting those out, but the SCA agreements do cover both DRAM and NAND. And it is through 2030, but we're not breaking it out specifically. I can tell you that the DRAM volume is a little bit less than that, approximately 35% and the NAND bit volume is a little bit more. And as we think about these going forward, we are -- we have more availability. As we said, that this number could get up to be higher in the future as we continue negotiations for SCA.

Understood. And then on the impact of China competition, firstly, if you can confirm that your China sales exposure is fairly minimum. And if Scott can comment on how should we see kind of China competitors kind of closing the gap on technology? If you can provide any color?

So on the first question, Yes, our exposure to China has been reducing over the last couple of years in the last several quarters, and we expect the exposure will be in the single-digit range in fiscal '27.

Then I could comment a little on the technology side. Currently, our technology leadership is at least 2 nodes ahead of of the China competition. I think it's important to say that our focus is on maintaining technology leadership and having true differentiation in our products in how we compete. We're -- as Manish, I think, mentioned earlier and as it was mentioned in the call, our 1-gamma DRAM technology is already the majority of our bids, and it set up to be the largest node in the company's history. It is dependent on EUV technology. The next-generation One Delta is well underway and we're focused on the ramp of that in the second half of next year. EV technology is critical to all advanced DRAM nodes going forward. And our expertise in that, both from the technology side in partnership with our supplier there as well as mass technology and other things is -- continues to be a key differentiator for Micron.

And then .

Thank you, -- and can I just clarify that the response in terms of DRAM and NAND was actually a bit common just to make sure that, that was clear. I'm not sure I specified. So making sure it's clear. .

Your next question comes from the line of Karl Ackerman from BNP Paribas.

Yes. Thank you, Tara. You are seeing robust demand across much of your portfolio, but this quarter does appear to the second quarter row that mobile and client segment saw bit shipments decline. Are higher memory prices reducing demand in this area of the market. And while this area of the market has been slow to adopt SCA, I'm curious if your growth in SCA is coming from this cohort.

Thanks, Karl. We did see sequential bit decline in our mobile business unit, but we did see revenue growth, obviously, with both higher pricing and favorable mix. I think what's important to note is that the premium segments of the both clients as well as the flagship smartphones do see robust demand for higher content, higher performance solutions. And that's a segment that, of course, we're -- those are the segments that, of course, we're very focused on. And with that wind at their back, we see the PC and mobile industry revenue to be growing even though we do see unit volumes declining. And in terms of SCAs, I'll just comment that we do have SCAs across all of our business units, including the Mobile Client business unit. And we're not specifically breaking them out, but I will tell you that we have SCAs because it's important for us to be maintaining diversified supply to all our end markets.

Got it. Yes. Maybe a question for Scott, if I could. How do you view the competitiveness of your in-house optimized base die on HBM 4E versus peers? Certainly, some of these customers are seeing custom solutions. Does the complexity and economic value flow primarily through the compute customers or the HBM providers?

So maybe just a little clarification first. On HBM 4E, we have co-designed with NVIDIA, but not in-house-based like we use on HBM 4. So I think just to be clear on your question, I just wanted to be sure I was answering it in the right way. On HBM 4E, this codesign is on a foundry process, both for the customized product and for the say get specified product.

So the -- and then I think maybe to add a little color to that. The differentiation just as we have demonstrated in the past on HBM prior HBM products winds up being in the power and the ultimate speed performance and the margin of the product working with our customers. That, in all previous generations hasn't been the same between suppliers in that we think will continue to be a strength of Micron in terms of quality of the product and the capabilities that we're able to put out.

And then in terms of your question on economic value, HBM is a premium product. And as Scott mentioned, as we look at the NV HBM with customization, we do expect that to be a high-value product as well. And we're confident that HBM will continue to contribute and be a strong ROI product for us.

Your next question comes from the line of Jim Schneider from Goldman Sachs.

Congratulations, Manish and Scott. I just wanted to maybe get a sense about -- of the 10 new customer SCAs you signed in the quarter, maybe give us a little bit of color on what customers are asking for? Obviously, they want supply and they want longer-dated supply. But is there any kind of change in the pricing construct either you or they are asking for? Sort of given -- the reason I ask the question is kind of given the expectation about tightness through calendar '27 and '28. Are you maybe a little bit less inclined to call for the fixed ceiling and floor pricing if you think you can capture a little bit more upside over the next couple of years?

Yes, yes. Well, thanks, Jim, and thanks for the shout out. I would say that the framework of the FCAs that we have is similar. But what's different is that the negotiations reflect current market conditions and outlook for market pricing, right? And so the direction of travel has been for higher pricing. And so those are now factored into the discussions that we're having with customers versus the prior ones, which we had talked about that were set at Q2 kind of market conditions. And as I think we gave color that about 3/4 of the SCA have this -- of the SCA revenue has a defined pricing framework in about 1 quarter is open to periodic negotiations or pricing that move with market dynamics.

Of the -- so I guess I would answer that in terms of the overall framework similarity. The majority of the pricing frameworks have floor and ceiling bands, but the newer ones are negotiated with an eye towards the current market conditions and the future market tightness that we see.

Got it. I mean just to be clear, does that mean all the pricing ceilings and floors just reset to the higher bucket conditions you're seeing today? Or is the nature of the pricing condition is actually different, too? And then maybe just to ask, just curious as to whether -- I don't think you've disclosed signing initial hyperscale customers. I'm wondering if those are included in the TAM.

Yes. So I mean there are multiple different frameworks. What I commented on is that the majority of the framework that have pricing are set with floor and ceiling, but there are multiple different frameworks that we're continuing to use. And since the last call in these new 10, we've signed a range of agreements from small to large SCA customers.

We're not commenting specifically, even in the previous question, not specifically breaking out which one just commenting that we now have SCAs across all of our business units, and we have SCAs ranging from small to large, even in the last 10 that we signed. And of course, in the total 26.

Your next question comes from the line of Chris Caso from Wolfe Research.

I guess our first question, if you could address the CapEx and the fab construction CapEx, the construction CapEx as compared to the tool purchases. And what I seem to hear in the comments is that the construction CapEx was increasing faster. And I'm not sure I interpreted that correctly. But understand that there's the clean room space constraints are constraining the ability to bring in tools for this year. But the increase in construction CapEx is obviously interesting because it doesn't result in bit production until at least '29, probably beyond that.

So Chris, I can start and then maybe Mark can add. But I mean, that's exactly right, is that the principal constraint in the industry is on clean room space because we -- the strong growth of AI in this demand vector has come on relatively recently in terms of the time line that it takes to build these clean rooms. So while we're all starting. That's why we're focusing there. And then -- you're right that we did comment that the majority of the construction CapEx increase for fiscal '27 versus our prior plans is for clean rooms that will come online in late '28 and beyond, which shows both how long it takes to build these clean rooms and why we need to get started with the construction investments now. But also shows our confidence in longer-term demand, both through the observation of the demand trends in the near term market trends that we're seeing as well as the structure of the SCAs and the discussions we're having with SCAs and customers now extending those commitments beyond 2030. I mean these SCAs are transformational for us in terms of being able to match supply with future demand and to be able to invest as confident.

Yes. Chris, maybe I can just add that just to make it clear that the majority of the increase is for construction CapEx. Most of that construction increases to accelerate clean room space availability in '28 and beyond. I mean it is a trend that we would expect to continue to beyond '27. And I think you made a very important point that just the spend there doesn't translate into bits and that these fabs will be made -- we put the equipment in the fabs and produce wafers when needed based on our view of the market and these SCAs are a good way for us to keep a pulse on the market and make sure that we get a return on that CapEx.

And I can just a couple just 1 other point. One other point, Chris, is that we will equip the clean rooms and build capacity to the demand trends that we see. So that's just another important point. And we have been executing long-term supply agreements with equipment suppliers to be able to make sure we have access to equipment as needed. But of course, we'll still equip and build the production capacity in those clean rooms in line with demand trends at the time.

Of course. Okay. As a follow-up question, I want to ask a question on the impact of CPU strength. -- both overall bit demand and your view of supply-demand balance. And obviously, that's probably been the biggest incremental surprise since the beginning of the year. You don't have the same trade ratio effects on CPU as you do on HPM. But I guess the question is, how significant is that in contribution to the supply-demand imbalance?

I think definitely, the realization that agent workloads are executed across CPU has been a big driver. There is a large attach rate of both LTE as well as DDR memory and SSD to enable those genic workflows. The -- those agenetic workloads are already starting. You're already starting to see multiple ones, whether in the enterprise or consumers implemented and driving real value. And that's been one of the reasons that server units are growing so strongly, as we noted, in the high teens. And I think the other thing that this shows is that while it's just another vector of logic and logic silicon to grow to be able to take advantage of the AI trends. And so that is driving higher logic silicon and into the overall demand for AI compute and frankly, creating more of a constraint on DRAM and making clear that DRAM is the principal constraint versus logic or power to the data center.

I mentioned the new -- there are many different software implementations for enterprise agentive workflows that we're all seeing. But seeing how quickly Meta's music just in the last couple of weeks since being released is just an example of how quickly the agenetic workloads are realizing -- enabling consumers to realize real value.

Your next question comes from the line of Joseph Moore from Morgan Stanley.

Great, in terms of supply growth decelerating next year, I guess that's a little surprising in the context of the CapEx. And I know you talked about some of the dampening effects of HBM, but I don't think that delta should be changing that much. So I just wonder if you could just kind of explain what the puts and takes are that supply would decelerate given the CapEx that you see.

Sure, Joe. And you're asking for both DRAM and NAND?

Yes, but principally DRAM.

Okay. I mean I think that we gave the color that HBM is growing faster than conventional DRAM. And the trade ratio as you move more of the industry is shifting formation of HBM 3E to HBM 4 and then towards later in the year in '27 HBM 4E and these are increasing trade ratio. So if you just look at both of those two things happening together, that has a dampening effect on the ability for bit growth to be growing. And keep in mind that both the bit growth for us and other industry participants from new technology nodes, both over time as you make more of your transition, you don't have the timing of transitions affect how much big growth you can have as well as the nature of the diminishing returns of each of those newer nodes.

So these are all the factors that are going to be constraining supply. But of course, the principal 1 is clean room space for everyone. And even though there's -- we're going to be having first wafer output from our Idaho facility mid-calendar '27 and others in the industry as well, maybe opening clean rooms, meaningful supply growth takes a few quarters after that. So I think that's really the -- those are all the reasons why we see that DRAM is reducing supply industry shipments next year.

Okay. That's helpful. And then to the extent that if you end up having been conservative on industry supply and there's more supply next year -- can you talk about -- I mean it seems like there's a lot of pent-up demand. There's a lot of appetite to have more supply. We've seen specking out of necessity for some of these AI RAC and things like that? It seems like they'll just respect to higher levels if there's more supply, but am I too optimistic there? Just how do you think about that?

Yes. No, I think that's exactly the point that these are and Sanjay mentioned this on the main call, we definitely think that our customers are choosing to maximize the compute silicon shipments they can make with the available memory supply that they have. And as that does create latent demand for more memory to attach to those, which would then end up delivering higher system performance and improved performance at the end customers. But -- and so I guess that leading demand, I think, is aligned with your concept that where more memory to become available, it would have -- it would easily get put into use in higher content growth in AI workloads, whether attached to accelerators or CPUs.

Your next question comes from the line of Mehdi Hosseini from Susquehanna International Group.

A couple of follow-ons for me. You did highlight your NAND bit shipment in is tracking below industry average. But what should we expect '27 through '28? Would you be able to be of the shipment in line with the industry average of 25%.

Mehdi, we're not commenting out that far, frankly, on either DRAM or NAND. We do expect conditions to be tight on NAND overall even as the industry is expected to grow in the mid-20% range in calendar '27 and '28. Some of the factors that account for our supply growth, we did -- we are utilizing some of the clean room space in Singapore now for our advanced R&D line for future NAND growth. We are also preparing for the ramp of our HBM facility in Singapore next year. And so some of the existing clean room from pilot operations. And so that's some of the reasons why our supply growth grew less than the industry this year. But we feel confident in our technology, and we do expect that our continued ramp to Gen 9 will provide good, very high ROI, cost-effective supply for us as we move forward. And then, of course, we do have the new clean room that we broke ground on earlier this year that will come online in the second half of calendar year '28.

Mehdi, just maybe if I can interject just because of time here. And then if you've got additional -- well, why don't you start with your question and then second question, I'll make a comment after that. .

Sorry, Mark. I didn't mean to interrupt you. But very quickly, I just want to get the team's opinion. When I look at DRAM, especially at the wafer level, the devices are all the same. And I argue that there is a fungibility of DRAM at the wafer level. And then there are some differentiation in the back end. And this fungibility at a wafer level hasn't been seen before because in the past, the cycles were driven by just one product, and there was a significant concentration of customers. So am I right with this assumption that fungibility of the DRAM at the wafer level gives you a better way of managing DRAM costs. Does that make any sense to you?

Maybe start with 1 thing. I think there's a bit of what you say, but I actually probably would have gone the other direction. And if you look at the front end variability or what we do differently on the DRAM processes between optimizing for HBM, optimizing for high-performance EssoChem, LPDRAM and optimizing for DDR6. There's, at this moment in time, probably more different than ever in the history of DRAM. And the same node, the different kinds of products that we have to build on it and a lot more differentiation built into those. I don't know what you would add.

Yes, I mean I would say that it really helps us with in the near term, the fact that we do run the different products in the same manufacturing lines, it does help us adjust mix. We don't have to run products in different fabs. We can run them on the same lines. But as Scott mentioned, each product has its own vector is trying to optimize. Obviously, in HBM, it's bandwidth with the TSVs and that requires unique process steps in DDR and LP, they each have their own as well process steps that are unique. And so I think the most important part of the fungibility is that we do have the ability to flex wherever demand is or even mix adjustments to try and meet our customers' requirements. I think that's probably the -- all within the same fab, that's probably the most important part. But I wouldn't say that it helps necessarily with cost.

Yes. I think, Mehdi, just I think we're at the end of the call. And I am really happy to hear the nature of the questions being focused on technology and the longer-term strong foundation and performance of the business. And I just -- I thought there'd be a question that let me just do a couple of housekeeping things. that I thought would come up earlier. Our first quarter guidance factors in a single-digit sequential bit growth and double-digit for cost for both DRAM and NAND and I wanted to make sure you had that for your modeling. And then also, excluding the incentive comp effects, the year-over-year R&D is going to be more than the $1 billion that we said last quarter, will be over $1 billion in '27 as we have added additional R&D activities. So again, I wanted to just make sure we got that out for your modeling.

This concludes the Q&A and today's call. Thank you for attending. You may now disconnect.

Hello, everyone. Thank you for joining us, and welcome to Micron's Fourth Quarter 2026 Financial Call. [Operator Instructions]

I will now hand the conference over to Satya Kumar, Corporate Vice President, Investor Relations and Treasury. Satya, please go ahead.

Thank you, and welcome to Micron Technologies Fiscal Fourth Quarter 2026 Financial Conference Call. On the call with me today are Sanjay Mehrotra, our Chairman and CEO; and Mark Murphy, our CFO. Today's call is being webcast from our Investor Relations site at investors.micron.com, including audio and slides. In addition, the press release detailing our quarterly results have been posted on the website, along with prepared remarks for this call.

Today's discussion contains forward-looking statements that are subject to risks and uncertainties. These forward-looking statements include statements regarding our future financial and operating performance and our business model as well as trends and expectations in our business, customers, market, industry, products and regulatory and other matters. These statements are based on our current assumptions, and we assume no obligation to update these statements. Please refer to our most recent financial reports on Form 10-K, Forms 10-Q and our other filings with the SEC for more information on the risks and uncertainties that could cause actual results to differ materially from expectations.

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 can be found on our website.

I'll now turn the call over to Sanjay.

Thank you, Satya. Micron delivered an exceptional fiscal Q4 with significant records in revenue, gross margin and EPS, each exceeding the high end of our guidance. Fiscal 2026 was an outstanding year. Revenue was 3.5x last year's record with data center revenue up fourfold. Micron's DRAM revenue for the fiscal year 2026 surpassed $100 billion. I'm thankful for the above and beyond efforts of our employees around the world that made these extraordinary results possible.

In recognition of these efforts and Micron's strong execution, we increased fiscal 2026 incentive compensation for every global team member, reinforcing our performance-driven culture and aligning team member success with long-term shareholder value creation. As strong as fiscal 2026 was, we expect fiscal 2027 to be even better. Industry demand has strengthened since our last earnings call, and we expect memory and storage supply-demand conditions to be much tighter in fiscal 2027 and 2028 than they were in 2026.

AI is becoming super intelligence and memory enhances this intelligence and the competitiveness of our customers' platforms. AI applications across end markets, whether using open source or closed source models are run on a variety of competing customer platforms. These platforms all share one important characteristic. Their value proposition is enhanced by the performance and capacity of memory and storage. Running an AI application on a platform with greater memory capability enables more scalable growth in usage, improves the end user experience and increases the value users get from AI applications. The strategic importance of memory to our customers also provides greater differentiation opportunities for Micron than at any time in our history. As we address AI-driven demand for an increasingly complex set of products across the memory hierarchy, we have opportunities to deliver differentiated performance and quality, time-to-market advantages and geographically diversified supply, including DRAM made in the U.S.

We also have opportunities for richer product mix relative to our competitors with our focus on higher-value solutions. Micron's technology leadership, strong product portfolio, strategic customer agreements and manufacturing excellence position us to capitalize on these opportunities.

Micron is the industry's technology leader. Our 1-gamma DRAM node and G9 NAND nodes are our largest production nodes today and are on track to become the highest volume nodes in Micron's history. Development of our next-generation DRAM and NAND technology nodes is also progressing well, and they are on track to begin volume production in the second half of calendar 2027. We are leveraging these technology nodes and our advanced packaging capabilities to deliver leadership products across the memory hierarchy such as industry-leading HBM, high-capacity SOCAM, high-capacity and high-performance DDR modules and data center SSD products. We are focused on our global manufacturing expansions to help address customer demand growth through the end of this decade and beyond. Micron is investing to provide long-term U.S.-based supply assurance for DDR, LP DRAM and HBM products through our fabs in Virginia, Idaho and New York to support our customers across a variety of markets, including data center, PC, mobile, automotive, aerospace and defense, medical, humanoids, robotics and other industrial and consumer markets.

During the last quarter, we celebrated a concrete poll milestone for our first New York fab with initial wafer output expected in calendar 2030. Our [ ID.1 ] fab is on track to commence wafer output in mid-calendar 2027, and [ ID. 2 ] is on track to commence wafer output in late calendar 2028. In fiscal Q4, we held a groundbreaking ceremony for our DRAM fab expansion in Japan with initial output expected in late calendar 2028 to support technology node transitions.

In Taiwan, we are on track for meaningful product shipments from our [ Tonggu ] facility in mid-calendar 2027. In Singapore, clean room preparation is ahead of plan at our HBM advanced packaging facility with initial output expected in early calendar 2027. Also, construction is on track for our new NAND facility in Singapore to begin output in the second half of calendar 2028. Production from new DRAM and NAND fabrication facilities takes time to ramp and gradually becomes more meaningful starting a few quarters after initial output.

Micron's strategic customer agreements accelerate the transformation of our business. These multiyear take-or-pay agreements sharpen our long-term supply planning and enhance the durability and predictability of our strong financial performance. Further, they provide our customers supply assurance and deepen technology road map collaboration. This, in turn, helps our customers invest more confidently in their business and enables their end consumers to benefit from their products and services.

To date, we have signed 26 SCAs, which we currently estimate to be over 35% of our revenue through 2030. 3/4 of this estimated revenue has a defined pricing framework, a majority of which have pricing bands with floor and ceiling prices. The remaining [ 1/4 ] of this SCA revenue expectation has pricing negotiated periodically based on market prices. Customers want SCA assured supply beyond 2030, and we have now signed SCAs that extend into 2031 as well as 1-year extensions to 2031 for 2 agreements. Any new discussions on SCAs where pricing is involved are negotiated with higher pricing based on prevailing market conditions and outlook. For the 26 signed SCAs and extensions, financial commitments from customers have increased to $32 billion, the vast majority of which are cash deposits. These financial commitments reflect our customers' confidence in their long-term demand for memory and storage.

Turning to our end markets. Please see our earnings press release for highlights across our high-capacity DDR and LP server DRAM, data center SSD, PC, smartphone and physical AI product portfolios. We expect server unit growth in the high teens percentage range in both calendar 2026 and 2027. This strong server unit growth is supported by a modestly lower rate of content growth than prior expectations amid tight memory supply. Growing model parameter size, longer context lens and higher concurrency continue to increase the memory and storage content required to execute AI workloads efficiently. Micron is leveraging our technology leadership and manufacturing excellence to deliver innovative products across the memory hierarchy to data center customers.

In HBM, our revenue for fiscal Q4 2026 grew faster than total company revenue in the quarter as we ramp HBM shipments across a growing number of customers. We have completed agreements for the vast majority of our calendar 2027 HBM bit supply with significant price increases year-over-year, narrowing the gross margin gap with conventional DRAM. We continue to execute well on our ramp of HBM4. We have a strong road map for future HBM products and are proud to be working with NVIDIA on the industry's first custom HBM4E implementation, NVHBM to be adopted on next generation of GPUs and NVLink Fusion platforms.

In NAND, AI context memory storage used for KV cache offload and HDD displacement opportunities are expanding the addressable market for SSDs. Data center SSD revenue in fiscal Q4 was nearly $10 billion, more than 10x the year ago quarter and was over 2/3 of total company NAND revenue. We are on track to deliver the fifth consecutive year of record market share in data center SSD in calendar 2026. This performance is driven by the strength of our NAND technology leadership, end-to-end data center storage portfolio and close collaboration with customers, which has resulted in design wins across the largest data center deployments.

PC and mobile industry revenue remain on track to grow this calendar year, driven by strength at the premium end of the market despite potential double-digit overall unit declines in both markets. OEMs continue to introduce new AI capabilities in flagship PCs and smartphones, driving robust demand for higher-performance devices with increased DRAM and NAND content. Micron is focused on these premium segments and is well positioned to support customers as they expand edge AI capabilities with our industry-leading memory and storage portfolio. Nearly half of MCBU revenue in fiscal Q4 was generated by 1-gamma products as customers accelerate qualifications and adopt our latest technology, which delivers lower power consumption and higher performance.

Autonomous vehicles are the first major deployment of physical AI, which we believe will expand over time to humanoid robots and other intelligent autonomous systems. These increasingly complex systems require substantially higher performance and more power-efficient memory and storage to operate in real time. Memory content in Level 4 and higher autonomous vehicles typically exceeds 200 gigabyte, while storage content reaches multiple terabytes, each more than an order of magnitude greater than in today's Level 2+ and Level 3 semi-autonomous vehicles. Humanoid robots are expected to have comparable memory and storage requirements to autonomous vehicles. With the anticipated increase in both units and memory content, physical AI can become a significant driver of memory and storage demand by the end of this decade. Several physical AI customers are sampling our next-generation products, and we are increasing investments in our technology road map to ensure we are prepared to capitalize on this opportunity.

Now turning to market outlook. We expect memory and storage supply-demand conditions to be much tighter in calendar 2027 and 2028 than they were in 2026. In NAND, for calendar 2026, we expect industry bit shipments to grow in the low 20s percentage range, slightly above our prior expectations. We expect Micron NAND supply to grow less than industry supply growth in calendar 2026. For calendar 2027 and 2028, we expect industry NAND bit shipments to grow approximately in the mid-20s percentage range and the industry to remain supply constrained in both years.

In DRAM, for calendar 2026, we expect industry bit shipments to grow in the mid-20s percentage range. We expect Micron DRAM supply to grow approximately in line with industry supply growth. For calendar 2027 and 2028, we expect industry DRAM bit shipments to grow approximately in the low 20s percentage range and the industry to remain supply constrained in both years. We expect industry HBM bit shipments to grow faster than conventional DRAM through calendar 2028. The structural gap between DRAM supply and demand growth rates is resulting in ongoing supply tightness and clean room additions are required to augment node transition supply growth and help narrow the gap. Even with additional industry DRAM clean room space plans with robust demand trends, including new upside requests from customers, we do not have line of sight to when supply and demand will return to balance.

Given the need for DRAM clean room space and supported by greater visibility from SCAs into our demand through the end of the decade and beyond, we plan to increase our CapEx in fiscal 2027 versus prior plans. A majority of the increase is for construction CapEx, most of which is to help accelerate clean room space availability in late calendar 2028 and beyond. We are also working to optimize production from available clean room space, which is resulting in some pull forward of equipment spending. Mark will provide more details on CapEx.

As we make these clean room space investments, we will remain disciplined in our approach and anticipate ramping equipment capacity appropriately with our demand in the market environment. To further accelerate execution and innovation across the company, last month, we announced leadership appointments. Manish Bhatia has been appointed to President and Chief Operating Officer; and Scott DeBoer has been appointed to President and Chief Technology and Product Officer. Manish leads Micron's business units and global operations with accountability for our operating P&L. In his role as COO, Manish has end-to-end responsibility for demand through supply, enabling faster, more integrated decision-making and stronger alignment across the organization to meet our customers' evolving needs.

Scott leads Micron's innovation, technology and products organization. In his role as Chief Technology and Product Officer, Scott is responsible for advancing Micron's industry-leading memory and storage road maps, accelerating innovation to meet customers' rapidly evolving requirements and overseeing Micron Research Labs, a global flagship research hub dedicated to breakthrough memory and compute technologies.

I'll now hand it over to Mark for our fiscal Q4 financial results and outlook.

Thank you, Sanjay, and good afternoon, everyone. Micron delivered exceptional results to close out the fiscal year with fiscal Q4 revenue, gross margin and EPS all exceeding the high end of our guidance.

For the full year, we achieved record revenue of $133.2 billion, up 256% year-over-year. Fiscal 2026 gross margins expanded to 81.1%, a 40 percentage point improvement from fiscal 2025, and EPS increased 811% year-over-year to $75.52. To date, we have signed 26 SCAs in total, and our remaining performance obligations, or RPO, is approximately $150 billion. All SCAs have take-or-pay contracted volumes, and RPO reflects the contract value for only SCAs that have a determined pricing framework, which can be either a fixed price or subject to a pricing floor and ceiling. RPO is based on committed volumes and minimum pricing and is inherently conservative. As mentioned in our last earnings call, even at floor prices, we expect margins mainly above any prior cycle peak margins. We expect revenue to well exceed the associated RPO over the terms of the agreements.

Consolidated fiscal Q4 revenue was [ $54.2 ] billion, up 31% sequentially and up 379% year-over-year. Fiscal Q4 revenue was sixth consecutive quarterly revenue record. Fiscal Q4 DRAM revenue was a record $39.8 billion, up 343% year-over-year and represented 73% of total revenue. Sequentially, DRAM revenue increased 27%. Bit shipments were up mid-single-digit percentage range. Prices increased high teens percentage range, driven by tight DRAM industry conditions. Fiscal Q4 NAND revenue was a record $14.1 billion up 526% year-over-year and represented 26% of total revenue. Sequentially, NAND revenue increased 42%. Bit shipments increased approximately 10%. Prices increased approximately 30%, driven by tight NAND industry conditions. Consolidated gross margin for fiscal Q4 was 87%, up 210 basis points sequentially. This improvement was driven primarily by higher pricing and strong execution, partially offset by mix.

Now turning to quarterly financial performance by business unit. Cloud Memory business unit revenue was a record $16.3 billion and represented 30% of total company revenue. [ CMBU ] revenue was up 18% sequentially driven by higher pricing and bit shipments. CMBU gross margins were 83%, flat sequentially, driven by higher pricing, offset by higher HBM mix. Core data center business unit revenue was a record $18 billion and represented 33% of total company revenue. [ CBU ] revenue was up 56% sequentially driven by higher pricing and bit shipments. [ CDBU ] gross margins were 90%, up 290 basis points sequentially and driven by higher pricing and favorable mix.

Mobile and Client business unit revenue was a record $13.1 billion and represented 24% of total company revenue. CBU revenue was up 14% sequentially, driven by higher pricing, partially offset by lower bit shipments. MCBU gross margins were 90% up 260 basis points sequentially, driven primarily by higher pricing and favorable mix. Automotive and Embedded Business Unit revenue was a record $6.8 billion and represented 13% of total company revenue. [ AEBU ] revenue was up 47% sequentially, driven by higher pricing and higher bit shipments. [ ABU ] gross margins were 84%, up 470 basis points sequentially driven by higher pricing.

Operating expenses in fiscal Q4 were $2.6 billion, up $1.1 billion quarter-over-quarter. The sequential change was primarily due to the increase in incentive compensation for every global team member, along with our decisions to contribute $300 million to community investments. We generated operating income of $44.6 million in fiscal Q4, resulting in an operating margin of 82.3%, up 110 basis points sequentially and 47 percentage points year-over-year.

Fiscal Q4 taxes were $6.8 billion on an effective tax rate of 15%. Non-GAAP diluted earnings per share in fiscal Q4 was $33.42, up 33% sequentially.

Turning to cash flow and capital expenditures. In fiscal Q4, operating cash flows were $44 billion. Capital expenditures were $10.8 billion, resulting in free cash flow of $33.2 billion. As noted in previous disclosures, customer cash deposits associated with the SCAs are reported within financing activities and, therefore, do not affect our free cash flow. Customer cash deposits received during fiscal Q4 were $12.3 billion. Ending inventory for fiscal Q4 was $10.4 billion, with days of inventory at 129, an increase of 9 days sequentially. The increase in DIO includes the effect of node end-of-life related build ahead and manufacturing-related incentive compensation in fiscal Q4 that was absorbed into inventories. Our inventory levels and supply remain extremely tight and we expect DIO to decline in the coming quarters.

We reached record levels of cash and investments of $73.5 billion at quarter end. Customer cash deposits on our balance sheet at the end of fiscal Q4 were $12.7 billion. SCA cash deposits are unrestricted and will be returned to customers over time towards the latter half of each agreement's term assuming minimum purchase requirements are met.

During fiscal Q4, we reduced debt by approximately $500 million, including a note redemption that reduced senior notes by approximately $300 million. The weighted average maturity on our outstanding debt is approximately 9 years. We closed the quarter with $5.2 billion of debt and a net cash balance of $68.3 billion. During the quarter, we received 2 credit rating agency upgrades and are now rated at BBB+ or equivalent with all 3 major credit rating agencies. Our balance sheet has never been stronger, and we expect it to strengthen further even as we increase investment in technology and needed capacity. As noted previously, we intend to increase our capital return from December 9, 2026, the second anniversary of the signature of our definitive chips agreements. Over time, we expect to return 100% of our excess cash to shareholders.

Now turning to guidance. We expect fiscal Q1 revenue to be a record $61.5 billion, plus or minus $1.5 billion. Gross margin to be approximately 86.25%, operating expenses to be approximately $2.06 billion. Based on a share count of approximately 1.15 billion shares, we expect EPS to be $38.15 per share plus or minus $1. We expect fiscal 2027 to be another record year with sequential revenue growth each quarter. Consistent with projected strong execution and record company financial performance, we expect fiscal 2027 to have higher incentive compensation levels. We anticipate fiscal Q1 to be the floor for gross margins in fiscal 2027. As Sanjay mentioned, we made a decision to increase fiscal 2026 incentive compensation in fiscal Q4. Most of the increase in fiscal 2026 incentive compensation pertaining to manufacturing was absorbed into inventories in fiscal Q4.

As a result, the effects from the sale of these higher cost inventories principally impact fiscal Q1 gross margin. Fiscal Q2 benefits from less of this fiscal Q4 related compensation expense, but this benefit is offset by the impact of higher fiscal 2027 incentive compensation. We expect higher gross margins beyond fiscal Q1 for the remainder of fiscal 2027 with a more moderate rate price increases.

We project operating expenses to increase by approximately $2.5 billion in fiscal 2027, primarily from higher R&D to support an unprecedented set of opportunities in memory and storage and from higher incentive compensation plans. We expect that fiscal Q1 and fiscal 2027 tax rate of around 15.5%.

Micron continues to invest in a disciplined manner across our global footprint to address customer demand. As a reminder, our CapEx is net of anticipated government incentives. In fiscal Q1, we project CapEx of around $11.5 billion and anticipate first half fiscal 2027 CapEx to be approximately $25 billion. We project CapEx to be higher in the second half of fiscal 2027. We expect a meaningfully higher growth rate in construction CapEx as compared to equipment CapEx in fiscal 2027.

Before I close, I would also like to add my thanks to all Micron global team members for their focus on technology and product innovation and disciplined execution that makes these strong results and outlook possible.

I'll now turn it over to Sanjay to close.

Thank you, Mark. Super Intelligence is creating the most compelling opportunity for Micron in its history. Fiscal 2026 was an outstanding year, and we expect fiscal 2027 to be even better. As we celebrate the 48th anniversary of Micron's founding, I would like to acknowledge the nearly 5 decades of innovation disciplined execution and perseverance that have prepared Micron for this moment.

We will now open for questions.

[Operator Instructions] Your first question comes from the line of Timothy Arcuri from UBS Securities LLC. Please go ahead.

Mark, I wanted to ask about capital return. I know you don't want to front run yourself. But can you give us any milepost here? Like what are you thinking of minimum cash balance you have pretty much double the cash that Apple or NVIDIA have. So sort of what's enough cash? Do you think -- do you want to keep $100 [ million ] in cash and you return everything beyond that. Can you give us -- I know you don't want a front runner so you're but can you give us some sense of what the mile toes are?

Sure, Tim. I'm happy to provide some perspective. Yes. The market conditions, Micron's Technology and product position and operational execution, all combined to deliver very strong free cash flow. We had $33 billion as reported here in fiscal Q4. We expect a strong free cash flow growth to continue on the market conditions and disciplined execution. Even while we invest in more R&D and CapEx, as you've heard today.

So with these demand drivers and supply factors and long-term and committed agreements that you heard about today, plus Micron's technology position and execution. We think this free cash flow strength is more durable. So near term, based on our Q1 guide and including the CapEx number that you heard you will see free cash flow is significantly higher than $33 billion we reported in fiscal Q4. On target cash, we expect to be around a target cash level by the end of fiscal Q1. Now over time and with the rate and pace determined by various factors, we plan to return excess primarily done through share repurchase. We intend to increase capital return, as we've talked about before and in the script today from December 9.

And then finally, I'll just add that our current authorization that you can see from our previous filings, for share repurchase stands at $2.2 billion. And you can assume that we will seek additional authorization in the near term on more authorization.

And then just a follow-up. So CapEx, you're not guiding the full year, but it sounds like maybe it's going to be [ 55% ], maybe a little higher this year which seems like it's maybe in the high teens as a percent of revenue for fiscal '27. I know that revenue is a lot higher, so it's going to take time for the CapEx to kind of catch up. But how do we think about capital intensity over the longer term for the business. I know you used to talk about mid-30s. That seems a little high, probably given what's going on now, but is like [ 20% to 25% ] the new norm. Can you kind of walk us through that.

Yes, Tim, I would just comment that you captured correctly that we gave you a first half CapEx number we indicated second half would be higher. Importantly, that mix of capital spend is shifting to more construction. And we gave you some commentary on that versus equipment. And we would expect that trend to continue here next few years. Now as it relates to capital intensity, I think as you point out, the capital intensity is low on historic levels, and that is reflective of the strategic asset that memory and storage has become and the industry is structurally reset and we will continue to work to add capacity in a very disciplined manner, ensuring that we're getting adequate return on that capacity investment going forward.

Your next question comes from the line of C.J. Muse at Cantor Fitzgerald.

I guess first question on gross margins. within the guide for November, can you quantify the impact from the higher ASP inventory? And are there other sort of mix shifts that we should be thinking about that are impacting the sequential?

Yes, C.J., I will -- this is Mark. I will I will take this opportunity to just provide a bit more perspective on fiscal Q4 and fiscal Q1 to help walk you through the puts and takes on margin.

In fiscal Q4, we made the decision to increase incentive compensation. And you can see that most clearly in actually the OpEx number. And that you can see it clearly in the third quarter to fourth quarter OpEx. Now in manufacturing expense, most of these costs, they're absorbed into inventories in Q4. So there was only a small effect in Q4. The higher cost inventories as a result of the increase in incentive comp booked in the fourth quarter, the impact of that, you'll see -- you see in the fiscal Q1 guide. And we also have some start-up costs, which I've talked about previously and some other costs, but incentive comp is a big driver to that gross margin outlook.

And in total, these factors in the first quarter are roughly $1 billion of higher cost in Q1. So that gives you a sense of the margin impact. Now I think it's important to note that we also have significantly increased fiscal '27 incentive compensation. And you'll -- you see the effect of that in first quarter again, in the OpEx number in the fiscal Q1 OpEx guide.

On manufacturing, most of the higher FY '27 incent comp, will start to impact margins in fiscal Q2. Now beyond the normal cost increases from higher volume and depreciation you have this roughly $1 billion of ongoing costs from incentive comp, higher start-up and some other costs in fiscal '27. So while it's a headwind in the first quarter and there is some sustained costs through the year, I just -- it's important to keep in mind how structurally different the business is as far as profitability and return or operating at a much stronger level and have talked about how '27, '28, we expect to be stronger market conditions in '26 and long-term agreements give us and take-or-pay agreements give us visibility beyond that. And I think it's important to also note that these costs -- these -- some of these costs can be considered variable or temporal.

So for '27, we also discussed how we expect Q1 to be the floor and gross margin and we expect higher gross margin for the balance of the year as we have continued price increases and strong operating performance.

Very helpful. I guess as a follow-up, could you speak to, I guess, HBM, you talked about raising pricing to be closer to conventional DRAM. Should we assume that's a Jan 1st increase? And how should we think about the relative growth of that business overall? Is there sort of a percentage that you're comfortable sharing?

So I can take that question, C.J. So as you know, for 2026, our prices for HBM were negotiated with our customers last year. And we mentioned that now or 2027, a large part of the volume is already sold out for 2027 for HBM and the prices are much higher than 2026 prices. And of course, that is helping us narrow the margin gap with the non HBM memory there. So overall, our HBM is on a very good trajectory. We indicated that over for the industry. We expect HBM to be outgrowing the DRAM in terms of the demand growth. So HBM is on a strong trajectory. Our portfolio is very well positioned. Our HBM3, [ HVMI ] products late next year with HBM4E products. So we continue to see strong momentum with our HBM products and well positioned to address the opportunities ahead.

So I hope I answered your question regarding the growth here, strong growth, strong products and with the price increases starting in calendar year 2027, narrowing the margin gap with non-HBM part of the DRAM market.

Your next question comes from the line of Vivek Arya from Bank of America Securities. Please go ahead.

For the first one, Mark, I wanted to go back to cash returns. At this space, Micron could generate over $100 [ billion ], I think, from Q2 to Q4. So even if, let's say, Q1, you're building cash to get to your target balance sheet your cash generation from Q2 onwards should get you to at least $100-plus billion. So if that is the case, why shouldn't we expect that level of cash return in fiscal '27? So I know you're not providing a specific number. But what am I missing in that high-level analysis, right, beyond just going through the outline of sales and margins and CapEx estimates that you're providing?

Yes. Vivek, all I can add is that and hopefully made it clear in the prior response that we have the ability and the intent to increase our capital return and we -- you can expect us to seek to increase our authorization and commence stronger capital return from December 9 in accordance with the agreements we have with on chips.

Okay. For my follow-up, maybe one for Sanjay, when we look, Sanjay, at just the valuation of memory stocks, right, very depressed, would suggest that people feel that next year might be a peak for pricing might be a peak for earnings in the cycle either because the industry is bringing on incremental capacity or there is an additive that maybe some of your customers even in the data center might despec products because either because of the shortages of memory or just because, right, it's such a bigger part of [ pillar ] material.

So I know you don't talk about specific pricing. But as you were to think about conceptually in 2028 for the industry, what is the potential for industry pricing to continue to stay favorable to even potentially increase versus '27 given these headwinds from either incremental capacity coming online or some customers wanting to despec their products to cope with market conditions.

As we noted, calendar year -- in calendar '27 as well as 2028, we see demand exceeding supply. And in fact, we see greater tightness in the industry in '27 and in '28 versus '26. So overall, supply-demand environment is only getting tighter. And of course, even as we work hard to bring up the capacity, as I shared in my prepared remarks, even with any new clean room space coming up in 2028, we see continuing tight supply conditions because, first of all, clean rooms take a long while to build, even after they are built, even after first wafer output, production ramps up only gradually in the clean rooms. That's just the nature of what it takes to bring up production and with HBM going from 3E to a greater mix of 4 and 4E and with the trade ratio that exists, that again, creates headwinds with respect to supply growth node transitions of the future gave less productivity gain per wafer as well. So there are a lot of factors, these key factors that are headwinds to the supply growth.

And even in 2028, even as some of the new clean rooms start ramping up, the supply is tight. And that is in the backdrop of strong demand. While some level of content growth may be moderately lower in some of the servers compared to prior expectations. Overall, unit demand for servers continues to increase '26 as well as '27, we expect high teens in terms of server unit shipments and that sets the stage well for data center growth of shipments for DRAM in 2028 as well. So it's a strong demand environment and customers work with -- they fully understand this extremely tight supply outlook that we have. We have shared with you that we do not see line of sight when supply catches up with demand because the demand trends of larger models, growing context, more current currency, greater agents across enterprise and consumer only continue to drive greater need for memory, greater need for memory content as well as for higher performance memory.

So when customers reduce their content growth versus prior expectations as I refer to in certain platforms, that really is to enable them to ship more units as reflected in the strong server growth in '26 as well as '27, which sets the stage up well for 2028 time frame. So I see healthy demand supply environment here. Going forward for the factors related to demand as well as for supply, and that obviously bodes well for the industry pricing environment as well.

And I'll just add here that our SCAs gave us tremendous visibility with our customers. And we mentioned here that our customers are actually coming to us, asking for more supply, SCA customers are asking for more supply non-SCA customers. I mean we are getting POs. I'll tell you that 2027 more than 75% of our output is already committed for 2027. And a majority of discussions with our customers today are already around 2028. So overall, the industry demand supply environment and the outlook is in a very healthy place here. And we do not have -- in this strong demand environment, we do not have a line of sight to when supply and demand will get in balance. And customers want assurance now even beyond 2030 time frame. I mean, we mentioned that we concluded some extensions to our SCAs as well as the 2021 time frame as well as new SCAs for extending out to 2021 time frame.

Our next question comes from the line of Krish Sankar from TD Cowen.

I had 2 to Sanjay, there's been talk of a 1 large customer [ despecking SBM ]. I'm kind of curious what your view on that is. And also implications given the fact that HBM higher trade ratio, if those wafers get reallocated to DDR with an increased DDR supply quite a bit? And then I have a follow-up.

So as we mentioned, we actually see the overall HBM supply -- I mean, HBM demand outpacing the industry demand in '27 as well as in '28. And we continue to see tight memory conditions in both '27 and '28 time frame. And the latent nature of a need for more memory in the applications continues to be strong because AI platforms in order to deliver their maximum capabilities and potential just need more and faster memory to again address the growing context, concurrency and of course, the larger [indiscernible] as well. So overall, the demand trends are in a very strong place.

And then customers make optimizations, they make these optimizations mainly to be able to ship more units to drive their own more growth as well as to address the end market opportunity of growing scaling up AI and these optimizations, when they occur, they do not take away from the latent nature of need for more memory in the systems. And these optimizations also have diminishing -- diminishing return for any further optimizations and if you contrast optimizations with actually the secular demand and the value proposition of memory in the platforms, you will see that the overall demand trend continues to be very healthy for memory and storage. And again, we see '27 and '28 tighter than '26. And HBM continues to grow faster than conventional DRAM through 2028.

Got it. Thanks for that Sanjay. And Mark, just a quick follow-up. I know clearly, like FY '27 CapEx higher than $50 billion, and it looks like your free cash flow is going to be over $100 billion. And higher growth rate in construction CapEx versus equipment. I'm just wondering, is that because you are constrained in getting semi equipment next year? Or is it not an issue? I'm just trying to wonder if there is a natural cap on CapEx next year because you're constrained in securing equipment.

Krish, it's related to just the strong supply demand and balance that we have and the lead time it takes to get greenfield capacity in place. So that -- we've talked through many fabs that we're building out and coming [ ID1, ID2 ], Japan, Singapore, [ Tong Low ]. And so -- and we've got now the visibility through these strategic customer agreements to build out this greenfield capacity. And then we will equip those fabs as appropriate, given our most current views on demand.

Our next question comes from the line of Harlan Sur at JPMorgan.

Last earnings with 16 SCAs secured, your view was the potential for 50% plus kind of SCA coverage on forward revenues given the pipeline of SCA signed in a new negotiation. Fast forward to this quarter, 26 SCAs are now secured on the coverage. And you and your customers have an updated industry view that is even further supply constrained in calendar '27 and calendar '28. So what do you think your forward revenue coverage will be if all of your current SCA negotiations are completed, I think, Sanjay, you said 70% of revenue is covered in fiscal '27 by customer commits, including SCAs, but would the team actually see 60%, 70% SCA coverage on forward revenues over the next few years when all of your negotiations are completed?

So regarding my comment on '27 that more than 75% of our output is already committed. As I have said, that is, of course, between the SCA customers as well as non-SCA customers. Keep in mind that we have certain large customers, where we, of course, are doing business with them on annual basis as well. So that volume coverage comment, I just wanted to clarify, it is across our customer base, SCAs as well as non-SCAs and given the strong demand trends, we are getting purchase orders for 2027 from non-SCA customers as well early on here.

And regarding your question on where could SCAs ultimately reach to. So that basically is still around 50% of our revenue through the 2030 time frame that we had previously commented on, that comment still remains about the same, that we could reach around 50% of our revenue to be covered by SCA. Of course, that could be less to, depending upon the revenue of all the rest of the business, right? So -- but overall, that objective remains by and large the same. And of course, we are always managing the mix of the business. It is important for us to be able to maintain flexibility in terms of managing our supply across our customer base, across the end market segments, across new customers, too, that may be evolving over the course of the next few years, particularly given the dynamic nature of our industry, the tremendous amount of innovation that's just happening across the board, we want to be in a position to basically remain in that position.

So basically, I think we are really making very good progress in this regard. And the benefit of SCA is to give us visibility into the demand out there in the future, and it really helps us plan our investments and that really is a fundamental change to the nature of this industry when you think about it, that it helps us manage our investments, manage our supply expectations with a long lead time, long time horizon, that is really good for the health of the industry versus the past when you would experience more volatility in our industry.

I appreciate that. And then on the NAND side of the business, relative to your competitors in the Micron team, I think total bit supply share puts you sort of in that #4, #5 sort of global market share regime, right? Yes? The team continues to drive this very, very strong like #2 market leadership position in data center and enterprise [indiscernible] is the strong performance differentiation is required here, right? It's not just you've got this great base G9 technology, but it's also the controller technology firmware customization. You're also a part of the NVIDIA [ SCADA ] initiative for direct [indiscernible] access to storage. And it seems like more and more of the inferencing related KV cash functionality is being offloaded to storage, right? And so with the expansion of memory curing to include more and more flash-based architectures, like has that changed the team's view on your R&D and CapEx investments in your NAND franchise?

So certainly, as you noted, the memory hierarchy from HBM to DRAM to SSD, I mean it continues to be leveraged as context of windows grow and as AI advances, and we are very pleased with our product portfolio position. And as we noted, I mean, our portfolio is in the best position ever in the industry, particularly to capture these huge growing market opportunities with AI. And of course, very proud of our team's strong execution on the side of SSD, which is critical for data center SSDs, which are absolutely critical in the memory hierarchy that is needed to drive the growth of AI.

And you're seeing the results of that 5 consecutive years now of share growth in the data center market, revenue reaching $10 billion in fiscal Q4 for data center SSD, data center SSD coming 2/3 of our total NAND revenue. So we are, of course, continuing to invest in R&D related to the products as well as making the investments on the manufacturing side, our investments related to NAND as part of our overall CapEx certainly are increasing as well. And we are -- as you know, we have talked about the next fab for NAND that we are also building in Singapore, which is currently on track for late calendar 2028 time frame. And our CapEx is supporting our G9 NAND tech transition, of course, driving toward equipment productivity optimization, NAND R&D and other aspects of our Singapore operations as well.

This concludes the Q&A session and today's call. Thank you for attending. You may now disconnect.

Ladies and gentlemen, welcome to KeyBanc Capital Markets Annual Technology Leadership Forum for 27 years. This is where tech's visionaries, investors and industry leaders come together to shape what's next. Please welcome Pat Kratus, Managing Director and Group Head of Technology to the stage.

Thank you so much. I'm going to need you every 8 a.m. to introduce me in the morning. That's a way to wake up. Welcome, everybody, to our tech conference. It means a lot to us to have you make the time and effort to come here in the second week of August. We know that this necessarily isn't on every one of your plans. We know that you're going to get a lot of great value from it. But from my perspective, we know it takes a lot of effort. And so it's our job to make sure that we put on a great program for you.

Also, thank you for giving us the opportunity to work with you. We see a lot of -- I see a lot of familiar faces here, some people who have been coming to this conference for the entirety of its 25-year run when it was in Vail and now here in Deer Valley, and there are some new faces. So for those of you that we actually do formally work with, thank you for giving us that opportunity. And for those of you where our relationships are just building to use a bad sports analogy, just give us a chance to swing the bat. We have some tremendous professionals on the banking research and throughout the investment bank, we would love to show you what kind of value we can bring and provide for you.

I have 3 quick housekeeping items just to make sure everybody is aware of. The first is you should have received an e-mail today to download your app. Sometimes schedules, meeting places do change and the best way to make sure that your meetings are happening as efficiently as possible is to download that app. Also very importantly, that app is going to be one of the ways that we can communicate with you. In the back of the room, I've got Paul [ DeMargo. ] Paul is one of the 2 KeyBanc's security officers that we have here. He's in charge of physical security. Many of you received yesterday an update that there's fires that fire is about, I believe, an hour and 20 minutes away. Paul has been in contact with both the hotel as well as the local Deer Valley and Park City fire departments.

Just to put your mind at ease, in case there is anything that we have to do, Paul and the KeyBanc's physical security team will make sure that we are getting that message out to you. The message we received yesterday was a last to everybody within basically a 2-hour round trip circle. So -- and the third housekeeping item is, I know that it's beautiful out on the patio. The sun is shining. The Vista is spectacular. But we would encourage everybody when you have the chance, please come in to see some of the great panels that are happening here in this room.

So Gartner has forecasted global IT spending up 13.5% this year, driven by AI. Gartner is also projecting that data center spending is going to grow over 55% this year. Very few people and very few companies are in the middle of this more than Micron, which is one of the reasons why I'm so excited to invite my friend, John Vinh up here to help bring Micron up to talk about some of the trends that they're seeing. So with no further ado, John, Sumit?

Great. Good morning, everybody. It's my pleasure to welcome Sumit Sadana, Executive Vice President and Chief Business Officer at Micron. In this role, Sumit is responsible for Micron's business units, P&L, helping drive the company's growth strategy, customer partnerships, product road maps and long-term positioning across memory and storage markets. Sumit joined Micron in 2017, brings more than 3 decades of experience across the semiconductor industry, having had leadership roles at SanDisk, Freescale and IBM.

So as AI drives unprecedented demand for memory bandwidth capacity and storage performance, micron is playing an increasingly important role in enabling next-generation AI infrastructure. We're fortunate to have one of the industry's most respected executives with us today to share his views on the AI markets, memory and Micron's strategic direction. Please join me in welcoming Sumit.

Thank you. Thank you.

So Sumit, you've been a veteran in this industry. You've seen multiple cycles. Maybe just take a minute and talk to us about what's different about this cycle that we're going through versus kind of previous cycles that you've experienced?

Sure. So John, thank you for having me here, and I appreciate all of you being here. I'll just start with some quick opening comments. I'll be making some forward-looking statements. So please look at our SEC disclosures for detailed risk assessments on our disclosures.

Our business is on a terrific trajectory. Our business and financial performance continues to be exceptional. And since our earnings, our aggregate demand signals from our customers have increased even further. Now we have said that we do expect these very tight industry conditions to continue beyond calendar year 2027. Based on these increased signals of demand from our customers, we now expect that 2027 calendar year will be even tighter than 2026 because the growth in demand that is taking shape is faster than the growth in supply for calendar 2027 on an industry basis.

We have been responding to this with significant investments across our global network of manufacturing, especially in key areas like Japan and Taiwan and Singapore and investments in India in back-end manufacturing and so on. And our key investments are obviously focused in the U.S. with an increase from $200 billion of investments to $250 billion over the next many years.

And these investments are also going to be aided by investments in our supply chain, and we made this $500 million investment in GlobalWafers for raw wafers. We're also making -- as part of a $3 billion investment in our supply chain. With all of the efforts that we are doing, we still don't have line of sight as to when the supply is going to be able to meet demand because demand continues to escalate at a very rapid pace over time.

So with those opening comments, I'll just talk to you, John, about the specific question that you asked. Certainly, we have been through many different time horizons and industry conditions that have changed through this time period across many different growth drivers in the industry over the decades. This time feels very, very different for a number of different reasons. First, AI is a transformative capability that has come on the scene. Of course, AI has been researched and developed and progressed on for many years, decades, but the ChatGPT moment and the whole generative AI field has put it on steroids. And I strongly believe we are headed towards artificial general intelligence and beyond that, artificial super intelligence.

And that AGI moment is getting closer and closer in part because AI has gotten so good so quickly that you're seeing these AI improving AI approaches that are accelerating the innovations that our Frontier labs are able to bring to the market. So this recursive self-improvement capability is super exciting and will create a lot of breakthroughs. What this means is that if you step back and look at the demand picture from a longer-term perspective, we are still in the very early stages of how this demand is going to proliferate. For example, most of the interaction today with AI still happens using chat interface.

And if you think about the number of users around the world that are going to grow, who get access to AI, billions of people over time will be using AI who are not using it today. If you look at companies who are trying to leverage AI, they are in the very, very early innings of being able to leverage AI capabilities, whether you look at Fortune 1000 companies or just the global large companies who can benefit tremendously from AI, both in terms of innovation and driving the top line as well as productivity for the bottom line.

And we are such early stages, and you can see the breakthroughs happening with Agentic AI. Now what is very important to understand about Agentic AI is that the interaction of these agents with the hardware infrastructure creates and needs 5 to 30x the tokens that a chat interface needs for a similar type of task.

And if it is a deep reasoning type of an agentic approach, it is even higher. So we are looking at the next few years of such dramatic growth coming from AI. And when you step back and think about it, a lot of this growth is coming in the digital domain, but the supply that we have to create for it happens in the physical domain. And this difference between demand growing in the digital domain and supply growing in the physical domain is a structural challenge because I don't think there is adequate appreciation for how long it takes for that supply vector to catch up.

On top of that, we have this issue of HBM growing very significantly. And HBM is needed in these AI systems because in a lot of the workloads, the processor, whether it's a GPU or an ASIC or a CPU is sitting idle for 50% of the time because it is waiting for data from the DRAM. And that is a huge underutilization of an important asset.

And so you need to really get a much higher performance memory bandwidth and a much higher capacity of memory in the system so that AI can be deployed on an efficient and scalable basis. And that means there is so much demand coming from that. And this HBM increase consequently is creating this 3:1 trade ratio that we have mentioned in the past, where to produce 100 bits of HBM, we have to reduce 300 bits of DDR supply because there is that 3:1 trade ratio between HBM 3E and DDR.

And when you go to HBM4 and HBM4E, by the time you get to 4E, that trade ratio has worsened to closer to 4:1. So all of this growth in HBM pressures the supply that is left for everything else, which means that the wafer supply has to increase dramatically, and that's just not easy to do. It takes a long time because the whole industry got itself into a state where most of that expansion was needed to be done in greenfield expansion, which means you go to an empty space where there's nothing but trees and you have to create an entire massive fab cluster there.

So that's a very challenging endeavor. And while our customers have to build data centers and secure power and real estate, building a data center, no offense to our customers, is considerably easier than building a sophisticated leading-edge technology front-end fab. It's one of the most complex engineering projects in the world from a construction perspective to build such fabs. And so it takes a long time for these fabs to come online and this leading-edge technology that gets deployed takes a long time to ramp.

And so because of all these reasons, we don't have line of sight as to when that supply vector will intersect the demand vector, which continues to escalate with every passing year as our customers have logged up all these contracts for power and real estate and data centers and so on. And they're all telling us that the #1 constraint they have today, and this is not just 1 or 2 customers, but a common theme across our customers is not power or real estate or data center capacity or logic wafers. The #1 constraint they have is DRAM.

So what we have done as a result is we have obviously created all these strategic customer agreements that have also changed how we think about this business. And memory has become a very strategic asset for all of the reasons I have mentioned. So this cycle, if you call it a cycle, but this time is a very, very different time. And we see that we are going to -- we expect to have very robust business and financial performance for a very long time.

Great. A lot to unpack there. I had a brief follow-up from your opening statements regarding aggregate demand has increased for you. Can you just give us some perspective and color on where is that aggregate demand increasing? I assume it's within the data center AI complex, but maybe just some additional color would be helpful.

Sure. Yes. I mean the aggregate demand across all of the segments is at much higher levels than it has been. Our customers are telling us that despite the fact that the prices are at very high levels that they're eager to get more supply because they are not able to meet their own business case requirements from driving their own units and volumes and revenue and so on based on the opportunity that they see in their end markets.

And it's not just focused on the data center, it's across different parts of the market. As we engage with customers on these longer-term agreements and even more near-term supply, their constant feedback is that they are not happy about the extent of volumes we are making available to them.

They are signing up in these agreements, but they feel like they're leaving considerable opportunity on the table, like I said, across market segments. Of course, it's most acute in the data center where quite often, we are not able to meet any more than half of the demand our customers have. And so I think when we look at all of that, it's not just for 2026. I mean, these are like multiyear signals that we're getting from customers.

That's great. I think when I just saw you this morning and catching up with you, you were telling me you've been incredibly busy meeting and talking to customers these days. I've got to imagine that the main thing that they're asking you for is how can I get more capacity. But beyond that, can you just talk about what your customer meetings and conversations are about? And I'm wondering just with this major shift in AI demand, has your relationships with your customers changed?

Yes. I mean, that's a great question because the nature of our engagements with customers is at levels that I have not seen in my entire career. And what I mean by that is because this role of memory has become so important and strategic to our customers. And it's not just about the supply-demand environment.

It's also very importantly, because of the phenomenon I described earlier about the way that memory is determining system performance, whether it is a data center server or it is a consumer product or it is something that goes into an industrial system, as our customers think about how to implement AI capabilities in their products across these various segments, they're persistently hitting up against this wall, which you have heard about this memory wall, where the bandwidth from memory is just not adequate.

The capacity of memory is not adequate. So they are all focused on AI is once-in-a-lifetime opportunity to create differentiation for our customers in their end markets. And they are very keen on my biggest bottleneck is how do I get more out of memory from a performance perspective? How do I use memory to create differentiation in my products? And what that means is we have been able to drive significant inroads with our customers on the engineering partnership front.

So we have visibility now and deep engagements with our customers on the engineering side that stretches out to road maps at our customers that are beyond 2030 time frame. I mean, when we started in the memory industry many years ago, it was rare to even understand what next year's products are going to be at our customers. And now we have, I would argue, even longer-term visibility than most in the semiconductor industry have across the different semiconductor markets.

And we have these deep engagements, and we have become, because of our leading technology and our leading product portfolio, the company of choice for our customers to partner with in order to solve some of their most pressing challenges and in order to innovate with us and co-engineer their products with us. And we have many examples of how we have done that.

I mean if you look at the HBM3E product, we came out with capability that was 30% lower power consumption than anyone else, the next best product. And that is such meaningful capability. You look at how LPDRAM, low-power DRAM was deployed in the data center, Micron innovated that capability first in the industry with NVIDIA.

We took LPDRAM, which was only meant for smartphones and laptops into the data center for the very first time with higher density and much lower power and very robust performance. And for the longest time, we were sole sourced in that with very robust ROI because it was such a differentiated product, and we became multisourced with NVIDIA only because we were just not able to meet and keep up with the demand.

And this is exactly how we expect a lot of other innovative vectors to occur with our customers where we will be first to innovate with customers, get a level of sole source, atmos dual-source. And more and more of our portfolio is going to be ending up in that category. I mean you're seeing our leadership in data center SSDs as another example. First to market with Gen 6, very, very high levels of demand, far above our ability to supply with that product, very high levels of demand for the industry's highest capacity SSDs with our 245 terabyte SSDs.

There are many, many examples of over the last many years, how we have totally transformed the product portfolio, and that's another reason why our customers are coming to us for these deep engineering interactions and innovations. And the fact that we are the only company who's going to be producing memory in the U.S. is another huge differentiator because a lot of our customers are looking for that improved resilience in the supply chain, and Micron is right there as the only company investing $250 billion in the U.S. So it's another huge differentiator. So it's totally changed the complexion of how we work with our customers. And they recognize that the next 10, 20 years, memory is at the center of AI and that relationship has to be different as well.

That's great. Clearly, your customer relationships are clearly deepening here. That's great. Maybe along those lines, I wanted to ask you about your strategic customer agreements. Also known in the industry is long-term agreements that you've put in place. I think you've talked about putting in place SCAs that will potentially cover half of your revenues. And then you also, as part of these agreements, have upfront capital commitments in aggregate of north of $20 billion, which is pretty impressive. I'm wondering if you could just talk about how these agreements came together. Did you kind of proactively approach your customers about putting these agreements in place? Or did they reach out to you just given the shortage situation that we're having?

Yes. Great question, John. So as far as these SCAs are concerned, we started working with our customers many months ago on them, and we pioneered this approach to do these strategic customer agreements. And we call them strategic customer agreements, first and foremost, because they are very, very different than the historical LTAs or long-term agreements.

First of all, those LTAs were somewhat of a misnomer because there was nothing long term about them. They were just 12-month agreements for the next calendar year. Another important difference is that those LTAs had no binding terms in them, right? It was more of a handshake kind of an understanding with customers about ensuring that there is a good level of supply chain planning that we do and documenting it so that they are putting more thought into what kind of supply they are intending to purchase from us.

That was the LTA time. So when we saw some of this AI demand for a multiyear time frame become so urgent for our customers, there was an increased level of anxiety at our customers to secure the supply, and we came up with this proposal and idea of strategic customer agreements. And we have pioneered this concept in our industry, and we have been the first ones to work on it with our customers, and we have also, we believe, signed the most number of SCAs across our industry.

Since our earnings, we have signed up more SCA agreements with our customers. And we really see these agreements as being transformative of the business model that we are used to. For starters, these SCAs cover a long time horizon. Some customers that are smaller like automotive customers have mostly 3-year SCAs. But the SCAs that covered the overwhelming amount of the revenue under SCA is going to be 5-year type of terms through the end of calendar 2030.

And so number one, these provide quite a long-term visibility to us. Second, these SCAs are binding commitments on purchases of these volumes by year, by customer. And these are take-or-pay agreements, and there are no contractual outs for our customers from these agreements. These are very much very, very different terms, very stringent and binding terms on the purchases.

These are backed up by tremendous amounts of upfront cash and cash-like commitments like letters of credit. But the overwhelming amount of the commitment is upfront cash that we are going to have on our balance sheet. So at the time of the earnings, we announced 16 SCA agreements with $22 billion of cash and cash-like commitments, of which $18 billion was just cash alone that we will have on our balance sheet. And of course, as time goes by, we will have more SCAs that we sign. Then these SCAs also have something that our LTAs never had, which is pricing. Some SCAs are going to be floating pricing. So the pricing will be consistent with whatever the market price is at that time. But most of the volume is going to be under these SCAs is going to have a price band.

And so there is a ceiling price. The ceiling price for the SCA that we announced at the time of earnings of 16 SCAs are CQ2 pricing. Since then, as I said, we have signed more SCAs. SCAs that we signed in the future are going to have pricing at the ceiling pricing that is going to be consistent with whatever the market price is at the time the SCAs are signed in the future. So that's how the ceiling price works.

There will be some SCAs that will not have any pricing mechanism and will just be focused on market pricing. The flow pricing, which is a very important part for us, is set at a level that provides a gross margin for us that is well above any prior peak in the cycles of the industry. And so these provide very high ROI that we can leverage to make long-term investments in capital and in our capacity.

With that said, we are going to have continued focus on disciplined investments. So we'll build out all these clean rooms, but we are going to put equipment investments in them consistent with our new -- our view of medium-term demand from our customers. So these SCAs are extremely transformative, but they also go beyond just volumes and pricing mechanisms. They also enable very deep engagements with our customers on the R&D and product road maps as well. So very, very transformative in aggregate.

Thanks, Sumit. Concern that I hear from investors is, obviously, we've gone through past cycles. And I think coming out of COVID, there were a lot of LTAs that were put in place that had legally binding contractual commitments and a lot of those were broken when supply normalizes. Question I get is, what's to prevent customers from breaking these SCAs at some point in the future once supply normalizes as well?

Yes. There are a couple of things I can think about on that front. So first, memory itself has become a very strategic asset, and our customers are viewing memory for the long term in a very different way. So to them, these SCAs are not just a short-term mechanism to get commitments on supply. They know that when they look out the next 10, 15, 20 years, because of AI and the growth driven by AI, the challenges in bringing up adequate supply, but also the fact that I spoke about how memory in a system from a system performance perspective is so critical.

Our customers recognize that they need to change the relationship between the company, the customers and a company like Micron. And so when there is that strategic mindset it changes how they think about the relationship from a long-term perspective. In addition, like I said, these LTAs that we used to have never had any binding terms. These SCAs have tremendous number of binding terms, and there are no contractual way to get out of these SCAs.

On top of that, these SCAs have an evergreen structure to them, which means over time, more years can be added to the back end on these SCAs to keep extending them. And the reason our customers have gone to that kind of a structure with us is they don't expect that this is just a 5-year kind of a thing.

They expect that over time, this will become a good construct to engage with us and because of how AI is coming in right now with a huge amount of growth driven from the data center, but over time, this growth is going to go into a lot of other areas of the market in a very meaningful way, proliferate from the data center to the edge with consumer devices, smartphones, autonomous driving and then huge amount of growth coming from robotics, which is going to be a ginormous growth driver.

Our customers recognize and understand that these spurts of growth are going to happen many times much faster than the supply can grow. So this is not the last time you're going to see a significant deficit of supply in the industry. So my view is that if some customers are wanting to be adventurous on the terms of these SCAs and complying with the terms of these SCAs, what happens when there is the next shortage. They may not be able to get much allocation at that time, right? And so there have been customers who have treated memory very tactically in the past, and those customers are struggling even more right now to get allocation in a tight environment. So I do think that customers recognize the strategic nature of the engagement and intend to think of this as a long-term partnership.

Great. That makes a lot of sense. Sumit, you had referenced this a few times this morning. What are your thoughts on physical AI, right? It's an emerging category. It's getting a lot of attention. You're seeing a lot of semiconductor companies start to invest in it. Where do you think the opportunity is in physical AI?

I think physical AI is in its infancy right now, and it has a tremendous and hugely exciting opportunity ahead. A lot of what has happened in AI has happened in the digital domain. And our customers recognize that the next massive growth vector that could potentially even dwarf the growth that is occurring today could happen in the physical domain when you think about things like robotics and so on, could become massive growth vectors in the future. And if you think about how our customers are approaching this, they have always had a view of robotics becoming, for example, a big growth driver, but there is a lack of adequate data to train these robots on -- from the physical world.

So there's a lot of focus on accelerating the training. And these robots and humanoids are going to have an exponential learning curve based on their ability to accumulate the learnings from each individual unit of robot can be taught something different and ultimately, aggregated learning from all of those can multiply the capability of each individual robot. So there is tremendous excitement on that exponential curve, and these exponential curves grow faster than we can grasp and analyze and project.

And so we very much expect that the physical AI domain is going to -- in its infancy now, it's going to grow rapidly over the next few years. And when it hits its inflection point, it's going to be another vector of growth that we are quite likely going to struggle to supply for years to come.

I thought it was interesting you described it as potentially dwarfing the opportunity that's in front of you today, which is outsized. Why do you feel that this could dwarf that opportunity?

I think if you think about, for example, just humanoids alone, there is massive opportunity there. And each single humanoid robot is expected to have hundreds of gigabytes of DRAM and terabytes of SSD. And that is a massive amount of DRAM.

And the reason you need that DRAM is you need these humanoid robots to have very fast response times and be able to have a level of functionality even if there isn't an access to the cloud and back to the data center. And so the amount of onboard compute capability that is needed for functionality, safety, security, all kinds of issues is going to be driving that level of capacity and driving a level of performance requirements that will enable us to come up with really innovative ideas and solutions for our customers.

And that aggregate amount of capacity multiplied by the extent of deployment that these robots are going to have over the next decade and beyond is going to just be astronomical. So we feel like again, robotics in its infancy now, but will start to grow rapidly later this decade and get into that exponential curve early part of the next decade.

Great. I think one thing that you had mentioned on your earnings call is you said the outlook for the server market has actually increased because you're starting to see some level of [ despecking ] by your customers there because of the increasing prices of memory, right? I think we're seeing RDM densities decrease. Can you maybe just talk to that? Is that potentially something that could be harmful to the industry if you start to see customers start to despec as a trend?

Yes. I mean in our interactions with our customers, it's not really driven by pricing on the server side as much as it is driven by just lack of adequate supply. Really, our customers are struggling to get their hands on adequate DRAM supply because the extent of supply that they feel is available to them is not going to enable them to ship the units that they need to ship to meet the opportunity that is ahead of them.

Consequently, they are focused on balancing how much capacity of DRAM to put in the system with the number of units that they want to ship. And so that is where the modulation of the average capacity in the system is happening, whether it's in servers or some consumer product. And it is very heavily driven by just availability of supply. Despite all of that, when you multiply the units with the average capacity that our customers are planning, the aggregate demand coming from even those new levels of average capacities that our customers are planning is rising at a pace that is just -- I made my earlier comment about aggregate demand increasing despite all of that change that our customers are planning in average capacity. The aggregate demand is increasing despite that.

Now I will also mention that there is this phenomena that I spoke about earlier about system performance getting impacted if there isn't adequate memory capacity. And that kind of changes in memory capacity in the system is going to further reduce the potential utilization that the processor was going to have. And that means that there is tremendous opportunity to improve the system performance simply by increasing the amount of memory in the system. Once the system is configured and qualified with a certain amount of capacity, increasing the capacity of memory in the system is an easier lift for our customers, and they would be able to introduce a higher-performing SKU with a higher level of capacity when they see the additional supply becoming available.

So there is all of this latent demand out there that comes from being able to increase the average capacities and create SKUs that are higher performing and are able to have longer context windows, much better capability from an overall system performance perspective when that additional supply is available. But again, it's not clear on a multiyear time frame when that additional supply will be available. So this is the optimization that our customers are trying to do to try and continue to maximize the units they ship.

Great. Maybe related to that, I think something that you've recently disclosed is that you're starting to moderate your price increases to some extent. Historically, as you know, in memory, when you start to see price increases start to moderate, it typically signifies that you're getting closer to the end of the cycle. But you just recently just said this morning that your aggregate demand in '27 is actually increasing. So that doesn't really make sense to me. I'm wondering if you could reconcile the price actions versus demand getting even stronger for you.

Yes. I mean I think if you look at our financial performance, it is at extraordinarily robust levels, right? You've seen our gross margin performance and the operating margins that we are delivering 81% operating margin in the last quarter is a very robust level of margin. Of course, there is pricing opportunity in the future. We are obviously going to try and optimize what the right level of pricing ought to be based on enabling our customers' long-term demand as well as ensuring a good financial performance and ROI for us.

There are lots of opportunities to continue to grow our revenue and profits over time. And we do expect this constrained environment to last for a very long time. And between the improvements that we'll have in our portfolio mix, the improvements that we'll have in our overall supply in terms of shipments and the opportunities that we have on the pricing front, we will optimize all of those parameters to ensure a very balanced outcome that's in the best long-term interest of our customers and ourselves.

Great. In a few minutes here, we're going to open it up for questions. My last question for you, Sumit, is, if you think about Micron for the next 5 years, what's the one thing you'd want investors to kind of know about Micron?

Yes. I think Micron today is a very different company and has a tremendously stronger business model looking out into the future than it has ever had. And we are executing at the best levels that we have executed as a company across technology, products and portfolio, manufacturing and operations, just a tremendous, tremendous momentum. And AI is a total game changer for memory and has made memory a strategic asset. So we believe that the next 5 years, the next 10 years, we have tremendous tailwinds, a completely changed and robust business model, high levels of ROI expectations and these SCAs that have been transformational, and we expect to be transformational for our business.

Great. I think there are microphones around the room. If you have a question, feel free to raise your hand.

Great. That's been really helpful and insightful. You noted Micron -- one of Micron's key differentiators as being an American producer. How does Micron view the expansion of Samsung's fabs in Texas and SK Hynix entry in the Indianapolis -- Indiana area as a differentiator going forward?

Yes. I mean our understanding has been that Samsung's investments are in logic foundry, not in memory and Hynix's investments are in back-end manufacturing, assembly packaging type of investments. Micron is the only company investing in front-end fab manufacturing in the U.S.

We have Idaho 1, Idaho 2 leading-edge memory fabs coming online middle of next year, first fab, second fab end of 2028. And then we have New York One and a cluster of New York fabs beyond that. We have our investments in Virginia for long life cycle technology and bringing 1-alpha DRAM into Virginia. So when you think about the -- and then, of course, investments in U.S. supply chain as well.

So when you think about all of these investments that we are making, the investments we are making in our communities, the investments in talent and all of the investments in technology here in the U.S., we have a very unique position compared to everyone else. And our customers recognize that, they value that. And we do believe that we will be able to get a premium for our U.S. supply from a pricing perspective.

And that premium is something that we have even baked into our SCAs. So we feel very, very good about our U.S. investments, our positioning as well as the differentiation that it's going to bring to our customers, the resilience it's going to bring to our customers' own supply chains and how it will enable our customers to meet their own goals about how to leverage more U.S.-made content.

I just wanted to ask about the custom HBM opportunity, when it can be material for Micron and what could mean for your market share?

Yes. I mean when we talk about HBM, as you know, we have a hugely differentiated HBM3E product that continues to be in high-volume production. HBM4, very robust capability, which we announced some milestones at the time of our last earnings. HBM4E is when there is going to be a custom SKU of HBM that we work with customers on. And there is plenty of opportunity for differentiation in HBM.

And the one thing that may not be very well understood on HBM is even beyond the custom opportunities that start with HBM4E and continue with future generations of HBM, there is also this phenomena where the whole co-design and qualification of HBM is such a time-consuming R&D-intensive and expensive process that it is not practical for our customers to be able to do business with all 3 HBM suppliers for all projects.

Consequently, many projects and platforms and several customers are going to just end up using 1 or 2 HBM suppliers and a very small number of customers will end up using all 3 HBM suppliers. So HBM, whether it's on a platform basis or on a customer basis, we see is going to become like a 2 supplier market and oftentimes just a single sourced opportunity. And it's not just true for HBM. Like I said earlier, since memory is such a critical asset now, there is this significant push to create differentiation using memory across a range of opportunities.

I gave the LP6 example for data center, but there are -- there is now a proliferation of opportunities to create differentiation in other end markets as well. And we see those opportunities, again, as being either single sourced or dual sourced for a long period of time because it is not easy for our customers to do that kind of co-engineering work with 3 different companies. And they typically will do it with one, then maybe bring in another. And so we think that the market is going to become very much like an ASIC-like market more and more. And we see that more and more of our portfolio over time is going to migrate to those sort of capabilities, which is incredibly powerful from an ROI perspective long term.

Great. I think we have time for one more.

Congratulations, first of all, on both not only the financial performance of Micron, but also the technological advances that you've made, which are pretty impressive. The gap between DRAM on a per storage unit and flash has grown dramatically over the last couple of years. And yet there's still a lot of cold data that ends up in DRAM. Are you seeing any threats in terms of people finding ways to dynamically offload their cold data off of Micron's DRAM over to flash despite its speed disadvantage and then port it back over to DRAM and therefore, reducing the DRAM requirements d needs? Or otherwise, do you see anything else that's happening on the flash side that you might be concerned about?

Yes. I think if you -- and we are doing a lot of work with some of the most leading edge customers and the frontier labs and so on, on architectures for processor and memory and storage and the entire hierarchy of how data is stored and moved across the hierarchy.

And what we are seeing is that the way AI works requires so much DRAM for a more balanced performance that our customers just don't have adequate capacity of DRAM in the system. And when they don't have that adequate capacity of DRAM, then things like KV cache spill on to the flash side of things and go into NAND, and you start to use more of the NAND for those purposes.

AI systems work well when there is ready access to high-performance bandwidth between processor and memory. And so the one gate is the performance and bandwidth from processor to memory, and that's why you see all these HBM-like capabilities and architectures. But then the next gate is just the raw capacity of DRAM in the system.

And our customers find that as they make AI systems do more reasoning and do more intelligent work and get into agentic modes, the context windows lengthen and that requires more DRAM and it ends up spilling over into NAND flash. And so we continue to see an environment where both the performance of DRAM leading to more innovations and the capacity of DRAM is heavily stressed in these next-generation AI systems.

Great. Thanks for your time this morning, Sumit. Thank you. Appreciate it.

Thank you very much.

Ladies and gentlemen, thank you for joining us, and welcome to Micron's Post-Earnings Analyst Call. [Operator Instructions]

I will now hand the conference over to Satya Kumar, Corporate Vice President of Investor Relations and Treasury. Satya, please go ahead.

Thank you, and welcome to Micron Technologies Fiscal Third Quarter 2026 Post-Earnings Analyst Call. On the call with me today are Sumit Sadana, Micron's Chief Officer; Manish Bhatia, VP of Global Operations; and Mark Murphy, our CFO.

As a reminder, the matters we are discussing today include forward-looking statements regarding market demand and supply, market trends and drivers and our expected results and guidance and other matters. These forward-looking statements are subject to risks and uncertainties that may cause actual results to differ materially from statements made today. We refer the documents we have filed with the SEC, including our most recent Form 10-K and upcoming Form 10-Q for a discussion of risks that may affect our results. Although we believe that the expectations reflected in the forward-looking statements are reasonable, we cannot guarantee future results, levels of activity, performance and achievements.

We are under no duty to update any of the forward-looking statements that conform these statements to actual results. We can now open up the call for Q&A.

[Operator Instructions] Your first question comes from the line of Ben Reitzes of Melius Research.

Mark, looking at the numbers here for the next quarter, right? And free cash flow is going to be somewhere around $30 billion plus. I just want to make sure really the buy side and investors understand what you're saying here with regard to cash return. So you're saying 100% will go back to shareholders. I assume the vast majority of that is in buyback. I mean if you go from $30 billion in free cash flow and grow it, you could buy back 10% of the company next calendar year, just basically, if we say something close to this quarter is what you do next year. Are you prepared to do that and buy back at that level? I just want to have you kind of react to the math and the commentary. At a $1.2 trillion market cap, that's where it is. It's 10% of the company. And I just want to make sure that you can react to that.

Sure, Ben, and thanks for the question. We're really pleased with financial trajectory of the business, the combination of memory being so important to so many markets, AI data center, the edge, enabling this -- or helping enable this technology revolution we have underway. And when we've got -- between our technology products and manufacturing performance, we are delivering record cash flow numbers. The last 2 quarters, we've generated as much as much of the company's history. So -- and we expect, as you point out, that, that cash flow growth will increase in the fourth quarter.

Yes, we're going to -- we've paid down quite a bit of debt over the past year. And there's cash will build. And we will maintain levels of cash that we feel comfortable that we can invest through all seasons and business. But as you heard today, we feel good about the durability of the performance of the business given the secular growth -- demand drivers, the need for more and higher performance memory, the structural supply challenges that we've talked about the last couple of years, slower node or node migration yielding less, HBM soaking up more wafers and the need for greenfield for incremental wafer capacity. So -- and then we have these strategic customer agreements, which we announced today, a meaningful number of these. So -- and we expect more.

So we will hold what we believe is appropriate excess cash. And then we've always said that we intend to grow the dividend over time. You saw us do a 30% increase recently. But the principal capital return we have will be share repurchase. I said today in the prepared remarks that we intend to increase our capital return from December 9, which is the second anniversary of our CHIPS agreement signature. And the rate and pace from there will be -- will determine based on a number of factors. But absolutely committed to capital return.

Do you think I can sneak in one more? And just with regard to -- and we're really pleased to see 40% and eventually half of your business on SCAs. But I cover Apple and they've never made a comment that they're -- basically that they're willing to pay full price for a component and pass it through to customers ever publicly. And to me, it was an advertisement that they are open for business for full-price DRAM, let's put it away. Are you tempted to do a lot more in DRAM given these conditions? Do we think that there's an appetite for a much -- maybe a higher mix than expected of DRAM, which obviously would then keep maybe some of the SCAs lower, but for a very good reason.

Yes. So Ben, when you say higher mix of DRAM, you mean versus what?

Well, versus HBM, versus NAND, whatever you're making decisions around, I mean -- and DRAM, commodity DRAM to consumer would certainly not be on SCAs as much as a hyperscaler would, I would think.

Yes. So just a couple of thoughts around that. So as you know, our mix in our business of DRAM versus NAND, DRAM inclusive of HBM tends to be oscillating between the 80% DRAM, 20% NAND, to maybe 75% DRAM, 25% NAND kind of is in that type range. And we are pretty comfortable with that mix of DRAM versus NAND. We intend to obviously focus on and service customers in both of those product categories. And as it relates to HBM, we have made a strategic decision that we have also communicated over time that our goal is to have our HBM share consistent over time with our DRAM share.

And so our intention is that we support our customers on HBM and also support our customers on the non-HBM portion of the DRAM business across all market segments, right? We definitely believe in the strength of diversity. And if you look at the AEBU business and the MCBU business, both of which are non-data center businesses in our business unit structure, that's almost like 40% of our company revenue. So we like that diversity and we continue to focus on ensuring that we are servicing customers and their demand and supporting their growth across all of the market segments, including non-HBM DRAM, HBM as well as NAND.

[Operator Instructions] Your next question comes from the line of Harlan Sur with JPMorgan.

I think it was September of last year when the Micron team said that they were booked out through calendar '26, on HBM3, HBM3E. Obviously, at that time, the team was still in [ Qual ] and [ eval ] on HBM4. But if you fast forward to now and with a stronger, especially XPU ASIC demand profile, is the team already booked up volume and pricing for HBM3E and HBM4 for calendar '27? Obviously, I understand you still need to go through [ qual ] on HBM4E. But are you booked up on current gen HBM3E and 4 for calendar '27?

Harlan, the demand that we have for HBM, our HBM products, HBM3E, HBM4, and of course, even ahead of the HBM4E [ quals ], the asks from our customers for volume, not just in '27, but as you know, due to these SCA agreements, we have been discussing -- one huge advantage of these SCA agreements is we have been discussing these demand requests from customers for multiyear time horizon. So if you look at this multiyear time horizon, even going beyond 2027 into 2028, et cetera, we are able to get very high confidence demand from our customers that is far in excess of our ability to support using our supply.

So the demand continues to be well above our supply even when we do these SCAs for multiple years, these SCAs contain volumes that are less and customers would actually like to sign up for. And in fact, in a lot of these negotiations, we spend a lot of time helping customers understand that this is all we can do in this time frame. And so absolutely, our demand for HBM, not just in '27, but even '28 is well above our ability to supply across all the different HBM flavors. And it's also true -- the same thing is also true, by the way, for non-HBM DRAM as well in the same.

Harlan, it's Mark. Maybe just something to add while we're on HBM. So today, we indicated that we expected market tightness to continue beyond '27. And part of that reason is we did see the HBM TAM increase. We saw that. We had said previously that it would cross $100 billion in '28, you see that now the HBM TAM easily crossing $100 billion in '27.

That's great color. The other thing I wanted to ask is, it's actually been over a year since the team has given us an update on your midterm to long-term view on industry DRAM and NAND bit demand growth. Obviously, much has changed over the past 12 months, inferencing workloads have crossed over training workloads, influencing workloads themselves continue to evolve and become more complex. And then on the server CPU side, your CPU customers are not forecasting like 30%, 40% per year CAGRs given [ Gentex ] like higher CPU intensity. I'm sure the mid- to long-term is also sort of guiding your discussions on these multiyear SCAs. So could you guys just give us an update on your midterm views on DRAM and NAND bit demand CAGR over the next, call it, few years?

So that's a good question. We have provided some updates to you on how we see 2026 bit demand forecasts change versus what we had provided earlier with DRAM forecast up a little bit, NAND relatively similar. The thing -- the reason we are not really providing a lot of forward-looking views on the CAGRS is because for the foreseeable future, the bit demand -- the shipment growth for bits is not really determined by demand anymore. It's actually more determined by the supply because the demand is so much above the industry's ability to supply that the supply growth, in fact, is going to determine how the shipment growth occurs far less so the demand growth because of its relative position versus supply.

So because of that, we are trying to point out what kind of growth trajectory, we believe we give you some kind of data points here and there around how we think this year's supply growth or demand growth is going to be next year. But we are not providing sort of an outlook too much further down because how the supply conditions change, we are continuing to constantly evaluate. But our expectation is that the supply growth will continue to remain sort of what is needed to meet the demand. We don't really see when the supply is going to be able to meet demand. That is not something we're able to project at this time.

Your next question comes from the line of Tom O'Malley with Barclays.

I just wanted to go back to the long-term agreements that you're signing. And just is there any way to kind of walk us through what happens if a customer was to cancel agreement like what financial hooks do you have in it you get to keep all the cash from the agreement? Just any color that you can add there would be helpful.

Sure. So I'll start by saying that these strategic customer agreements or SCAs cannot be canceled. Now there is no provision in this agreement to enable the customer to allow our customer to walk away from this agreement. These are designed to be take-or-pay agreements outside of automotive, generally, these are 5-year agreements. There are annual volume commitments for each of those years. And the take-or-pay means that whether they want to purchase the bits or not, they are obligated to pay for the price times the volume. The price itself for a lot of these large agreements has a price band, there is a price ceiling and a price floor the price gets negotiated every quarter based on market conditions, the price cannot exceed the ceiling, no matter what, cannot go below the floor, no matter what. And consequently, the value of these agreements can be readily determined.

There are also premiums in the agreement for products that may be more sophisticated, higher performance, higher capacities. Newer products like, for example, when we do LP6 versus LP5 or DDR6 versus DDR5, or new version of HBM, then there are provisions for those products to be priced differently looking into the existing products. So we have all of those provisions but the one provision that doesn't exist is any customer's ability to walk away from these agreements. Now beyond the take-or-pay related obligations that exist in these agreements and the financial locations that are price times volume, there is also this upfront cash deposit that these agreements entail, where customers have provided in the form of upfront cash deposit and related financial commitments like a letter of credit, for example, for minority of the total that even for the agreements we have already signed the 16 agreements aggregates to $22 billion plus in terms of total cash and related financial commitments, of which the cash alone is almost $18 billion.

So you can imagine that when we get to our target number of agreements, which will go from the roughly 20% DRAM versus roughly 1/3 of NAND bits all the way to roughly accounting for half of the company's revenue, which is what we are likely to end up at or somewhat more, you can imagine that the cash associated with all of those will significantly increase from the current $22 billion number. So that cash is our customers' commitment to this new business model. And so it's -- that's how the structure of these agreements are.

That's helpful. And just for my follow-up question, just -- I wanted to clarify on the cash deposits. It was my understanding that towards the end of the agreement, you end up returning those back to the customers. So I'm just -- the general question is, any strategic rationale for receiving those upfront and being able to use that at will, like any plans to that cash? And just any way to think through why it ends up getting return back to customers instead of just being recognized as part of the revenue that you end up selling them in the agreements?

Yes. I mean our customers are going to -- as per the terms of the agreement, they're going to be purchasing the volumes over the years. And the cash is sort of contingency and a show of good faith and confidence in this new business model from our customers, meaning in the unlikely event that a customer is unable to purchase or does not purchase the volume at the price as determined by the terms of the agreement, then we do have the right to be able to decrement the cash balance that ultimately will get returned to them as one available remedy, but not the only available remedy. None of this ultimately relief the customers of the liability of having to purchase the volumes over the term of the agreement at the agreed-upon prices. But the cash is just one of the elements of the overall transaction. The cash doesn't get returned all in one shot at the end of the term. It gets returned over a period of time, with the return of the cash weighted towards the second half of the term of the agreement.

Your next question comes from the line of Melissa Weathers of Deutsche Bank.

I wanted to ask on the non-HBM side of DRAM within the data center. So you guys have talked a lot about SOCAMM and using low-power DRAM for data center applications. And especially as we see the mix of server CPUs increase with agentic AI. I was hoping you could give an update on how you guys are seeing the growth in demand for SOCAMM attach? And what kind of trends are you seeing in adoption there?

Sure, Melissa. Agentic AI, as you noted, drives a lot of growth in CPU demand and CPU-based servers, and that is certainly a trend that we are seeing. I see CPU base levers are coming from multiple different suppliers. You're seeing a lot of companies announced products targeted towards the data center for CPUs, you have [ x86-based ] CPUs, you have CPUs from NVIDIA, Qualcomm. I mean, there's lots of different possibilities for types of CPUs that could be used in the data center over time to drive the use of agentic AI.

And we have CPUs that use DDR5 as well as plans from our customers to increase the use of LPDRAM in the data center. When LPDRAM gets used, it will be in the SOCAMM form factor. And as you know, Micron has been a pioneer. We were first in the industry to not just drive the usage of LPDRAM and we were, for the longest time, sole sourced on LPDRAM in the data center. But we were also the pioneers in bringing out new products first to market with the SOCAMM form factor. So we continue to expect that this is going to be an area of differentiation for us with our customers, a recognized leader in this space. We have market leadership in all of these products. We have really strong engagement with customers who intend to use LPDRAM as a way of reducing the power consumption, increasing the performance and even reducing the footprint of memory.

So certainly, these SOCAMMs help do all of that. There are RAS-related complications, reliability, availability and serviceability related complications that have to be worked through for LPDRAM because LPDRAM is not really designed for data centers. And that's where you're bringing in differentiation in helping our customers deal with that. We expect LPDRAM to grow over time as a percent of consumption of DRAM in the data center. -- and we expect to be leaders in that front.

And then maybe 1 for -- I don't know if it's Manish remark, but as we think about Idaho 1, Tongluo, Idaho 2, as you think about these greenfield fabs starting to have wafer outs in next year and the year after that, can you remind us, is there -- like how should we think about the impact of start-up costs or just the incremental impact on cost per bit that we should be flowing through as those greenfield fabs come online.

So I'll let Mark handle the technical question on the start-up cost accounting. But we did say, Melissa, on this call that given the trend -- the industry-wide trend towards higher performance solutions, such as HBM and even within the HBM category, higher trade ratios expected in the future with HBM, which obviously requires more silicon per bit versus traditional DRAM as well as with greenfield build-out, which doesn't get the same leverage on existing capacity as our traditional technology transition models for the industry have both of these trends to higher performance and as well as greenfield investments that take time to ramp scale and are not as efficient.

Both of these are going to be trends that we expect will actually increase DRAM bit costs here in the near term. So that was 1 thing that we did have in the prepared remarks. And then Mark, I think you can -- and that goes for ID1 for Tongluo, those are kind of all elements of that, that greenfield build-out, and that's something that we expect will be an industry-wide phenomenon.

And then Mark, you can comment. I think you've given some commentary before in terms of timing of the start-up costs for facilities, how it will impact the P&L?

Yes. So Melissa, as Manish mentioned, there are a number of factors that will be -- will bias the costs up and DRAM over time here, the trade ratios, as he mentioned for HBM and actually LP, the greenfield facilities. And to your question on startup, we've talked about this before. We began to see start-up costs more meaningfully, begin here in the fourth quarter and then into the first half of next year. And so you'll see '27 at elevated levels, think about $100 million, $200 million per quarter effect versus what we had seen a pretty good run rate. And then what that will be over time would just be a function the various ramp pools of fabs. So as you point out, Tongluo and ID1 are the first to go here.

We would -- we provided more color on this. I think it was maybe at the end of maybe '24 beginning '25 because it was a more material effect to the business at that time. I think I had said at the time, maybe 0.5 point to 1 point plus of margin effect. Today, with the size of the business, this effect is much reduced from before. And obviously, the faster we can get capacity on what we're obviously trying to do for our customers to get much needed supply. The benefit of that incremental bits is going to outweigh this incremental associated costs of start-up.

Your next question comes from the line of Vijay Rakesh with Mizuho.

So just some mention on the 16 SCAs that you announced, the 4 large customers, does it include any HBM? And will they include some of the major CSPs within that? Or can you give us some more color.

Sure. The SCAs that we have signed already do include some hyperscalers and where the purchasing for those hyperscalers requires HBM, and that is part of the overall agreement.

Got it. And then just a quick follow-up on the $22 billion deposit, is the intention that you hold it for the 5-year contract period? Or as the customers buy the product, it's prorated and runs through it? If you could just clarify that.

Yes. It's not a prorated type of a thing, it is a customer cash deposit, and it's not a prepaid revenue or things like that. And it gets returned to customers on a predefined schedule that has been agreed to that X amount will be returned in by quarter over time. And it is back-end loaded in terms of the second half of the agreement term is when the bulk of the return of the customer deposit occurs. And of course, the cash would be returned, assuming it hasn't been decremented for reasons drive terms of the [indiscernible].

Your next question comes from the line of Jim Schneider with Goldman Sachs.

Just stepping back for a moment, as you've had discussions with your customers about their forecast demand needs, let's say, through the end of fiscal '28 or the end of calendar '28. Where do you think you will end up in terms of the percentage of the forecasted demand that they have in terms of your ability to supply? I mean, is it something that's going to be 70%, 90%? And I guess how do you expect to close over time? Clearly, there is a gap between your ability to build facilities and where the demand is. But maybe give us a sense at a point in time, what percentage that represents?

Yes. I mean, there isn't a homogenous percentage number that we can provide because our strategy is different for each segment of the market. Of course, we try to be very diversified, and we are very committed to supporting each of the segments of the market. But as you can imagine, if the automotive industry can only do 50% or 70% of the units, then that would be a catastrophic problem for that portion of the market. So we can't have a one-size-fits-all kind of an approach. And so there are a number of complex factors that go into the assessment of what kind of fulfillment rate would be appropriate for what kind of customer and in what kind of geography. So there are lots of different factors that play into that.

With that said, I would say that the general sentiment amongst customers is that we are very short of their demand. For some customers, we are extremely, extremely short, like some of our supply numbers are a fraction of what they want. And then in other parts of the market, albeit a smaller part of the market, but super important sectors of the economy like automotive and some critical parts like defense, aerospace, some important industrial markets, including medical equipment and so on. Obviously, we try to do our best to minimize the impact.

I'm not saying that those customers get everything they want, but at least we try to minimize the impact that they are going through in this very challenging environment of tightness. So broadly speaking, the overall aggregate supply is substantially below the aggregate demand for both DRAM and NAND. Of course, DRAM is just extremely, extremely constrained is very constrained. All the segments are seeing those challenges.

So Jim, it's Mark. If I could just add to build on Sumit's comments about supply efforts -- we are -- as you heard and you heard Manish talk about the ramp, ID1 and Tongluo, we're doing everything we can to bring on supply. You heard us today between adding construction and tool installs in the fourth quarter here, we're increasing our fiscal '26 CapEx number to around $27 billion. We're also going to increase substantially CapEx next year, and it will be -- more than half of that increase will be construction.

Now we did provide some comments last quarter about the increased FY '26 to '27. Based on our comments, you may have come up with numbers that are sort of in the low to mid-40s, and we will be spending above that level as we look at it today. We're exiting -- we'll do about $10 billion this quarter, and we will step up from there into '27.

That was exactly where my next question was, which is given that run rate sort of implies mid-40s, you said you're going to do above that. But I guess what are the chances you're going to do materially above 50, are we -- something -- is something like 55% or 60% even in the cards?

No. I mean, we -- I mean, if we were -- we're not going to give a CapEx number, but if the number where that order of magnitude, I think we know it to update you more specifically. We're going to run about $10 million this quarter. We'll step up from there. We are going to be higher than we'll be in the mid-40s. And we're going to remain extremely disciplined as we always are. I mean, the ops team has been amazing on figuring out how to sweat these assets that we have as much as possible and then accelerating all these greenfield capacity adds that we have.

I think the -- we've mentioned a couple of times that the majority of the fiscal '27 CapEx, Jim, is for construction, which kind of also gives you some indication of those -- the construction dollars are not going to be producing bits in that time horizon, which is why we also talked about for us in the industry, the kind of greenfield capacity really starts to contribute to in calendar '28. And that supply -- and even with that supply improvement, like we don't see, as Sumit was saying earlier, an intercept for supply with demand.

Your final question comes from the line of Aaron Rakers of Wells Fargo.

Yes. I have one and one follow-up as well. On the SCAs, I know it was asked about HBM, but I'm curious about the NAND flash market. Obviously, that market is getting -- it seems to be further constrained. So as you're engaging with your customers on these SCAs, is there a strategic advantage you're finding of having both NAND and DRAM in your portfolio, competitively in these engagements? Is NAND pervasive across these SCAs? Or any context around that, particularly as it relates to enterprise SSDs?

Sure. Yes. I mean, our enterprise SSD momentum is exceptionally strong, and we provided you some data points on that with a $5 billion quarter in Q3 for enterprises, for data center SSDs outside of the -- inside of the $25 billion overall data center revenue for the quarter. And so absolutely, we do feel really good about the fact that we have incredibly strong portfolio of products, both on the NAND and DRAM side. They, of course, stand on their own feet individually in terms of their capabilities. We have hit record share after a record share of data center SSDs over time due to the strength of that portfolio. And you know the strength of our DRAM portfolio both in terms of HBM and non-HBM products.

Now in terms of the levels of constraints, both DRAM and NAND are very constrained. Of course, when we talk to customers across this long horizon of time, through 2030 calendar year through the end of 2030 calendar year which is the term of a lot of these large SCAs. And then they are definitely interested in getting their hands on NAND, but DRAM is certainly far more constrained and more difficult to supply in the quantities and volumes that our customers need. But like I said, I mean, NAND is very constrained too, but the sense of concern and urgency in the minds of our customers around DRAM is very, very high.

Yes. Very helpful. And then my final question on the competitive landscape, always trying to think about what vector could change some of the dynamics in the backdrop that we're talking very constructively about I'm curious how you've evolved your thoughts around China and the competition from either be it CXMT or YMTC? Have you seen any changes on that front? Or anything you want to share how you view the competitive landscape from that regard?

Sure. I mean, certainly, those 2 companies have grown over the years in terms of their capabilities and share. Most of their output for the overwhelming majority of the appends to be sold within China, we haven't really seen much by way of their product or competition from them outside of China. With that said, we are very focused from a competitive perspective in driving really the highest performing, most complex products in the portfolio.

So when you look at NAND, we are focused on data center SSDs in a very single-minded way. You have seen us also do really well on QLC across client SSDs, but also in data center SSDs. We are the QLC leader in the world. We are leaders in Gen 6, first company to come out with Gen 6 price and we have ramped them in volume. So whether you look at that or you look at the highest capacity to 45 terabyte drives, we are a leader there. You have seen the strength of our DRAM portfolio as well as a thing from HBM to high-capacity dims to LPDRAM leadership in the data center to mobile and client leadership in those markets as well.

So we could go on and on in that, but our focus is to look for these complex difficult to get right type of products, get into deep customer engagements across multiple years on the road map, gain their confidence in terms of being able to have a track record of meeting and beating time to market with the best specs in the industry, have a track record of innovation. You've seen how many nodes in a row of DRAM and NAND have been first to market on how many products are first to market across the board. And of course, I would be remiss to not mention that we have one of the best intellectual property portfolios in the world. Almost 65,000 patents, and we are very aggressive and have a great track record in defending our IP over a number of years, a number of decades, in fact.

So overall, we feel very good about where we are and the structural foundational changes in our business model that the combination of demand the combination of structural supply challenges in the industry and the place that AI is creating for memory and its new found relevance and importance and strategic nature. And now combined with these SCAs are completely transformative for everything.

This concludes today's call. Thank you for attending. You may now disconnect.

Customer cash deposit, and it's not a prepaid revenue or things like

Ladies and gentlemen, thank you for joining us, and welcome to Micron Technology's Fiscal Third Quarter 2026 Financial Conference Call. After today's prepared remarks, we will host a question-and-answer session. Webcast viewers, please note that you will be able to advance the slides as you view at your own pace.

I will now hand the conference over to Satya Kumar, Corporate Vice President of Investor Relations and Treasury. Satya, please go ahead.

Thank you, and welcome to Micron Technology's Fiscal Third Quarter 2026 Financial Conference Call. On the call with me today are Sanjay Mehrotra, our Chairman, President and CEO; and Mark Murphy, our CFO. Today's call is being webcast from our Investor Relations site at investors.micron.com including audio and slides. In addition, the press release detailing our quarterly results has been posted on the website, along with the prepared remarks for this call.

Today's discussion contains forward-looking statements that are subject to risks and uncertainties. These forward-looking statements include statements regarding our future financial and operating performance and our business model as well as trends and expectations in our business, customers, market, industry products and regulatory and other matters. These statements are based on our current assumptions, and we assume no obligation to update these statements. Please refer to our most recent financial reports on Form 10-K, Form 10-Q and other filings with the SEC for more information on the risks and uncertainties that could cause actual results to differ merely from expectations.

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 can be found on our website.

I'll now turn the call over to Sanjay.

Thank you, Satya. Micron delivered an exceptional fiscal Q3, the significant records in revenue, gross margin and EPS, all exceeding the high end of our guidance. Demonstrating Micron's position as a leader, enabling the AI era, our data center revenue exceeded $25 billion in fiscal Q3 on an annualized run rate of over $100 billion. Our data center SSD revenue exceeded $5 billion, more than doubling sequentially. DRAM and NAND industry demand continues to significantly exceed the industry supply. We expect tight conditions to persist beyond calendar 2027 as a result of AI-driven demand across all segments, coupled with structural supply constraints. We are excited to announce that we have now signed 16 strategic customer agreements or SCAs, which we expect will fundamentally transform our business model. .

The memory industry has been structurally transformed by the proliferation of AI. We are only in the early innings of the significant innovation and productivity that can be unleashed in every part of the global economy over time. Data center-driven growth will be increasingly complemented by AI-enabled features in smartphones, high-end PCs and new consumer devices as well as in automotive, industrial applications and robotics. Exciting possibilities enabled by robotics and human line as well as putting autonomous vehicles portend a robust long-term demand environment for memory and storage.

With respect to supply, our customers are recognizing their supply shortages in memory and storage will take considerable time to improve. Even as we expect industry supply to improve gradually in 2028, we currently do not have line of sight as to when memory supply will be able to catch up with increasing demand. Memory industry supply growth is dependent on significant greenfield fab expansions. These greenfield projects are large, complex and time-consuming. Further, the pace is constrained by several factors, including long lead times for fab construction across the world, shortage of workers with critical trade skills, complex regulations, including permitting and the need for enhanced energy infrastructure. Meanwhile, memory process technology which is among the most advanced to develop and manufacture in semiconductors is getting more complex with every new node.

Technology transitions are driving slower bit growth over time, wafer growth needs are significantly increasing cleanroom space and greenfield fab requirements, and HBM's growth and increasing trade ratio with every new generation further pressures non-HBM supply. In NAND, industry suppliers redirecting clean room space from NAND to DRAM, and overall limited cleanroom space constrained NAND bit supply growth. These factors taken together means supply is structurally constrained in its growth and ability to meet industry demand despite our comprehensive efforts to increase supply.

AI systems are powered by GPU, ASIC and CPU designs from an increasingly broad set of suppliers. However, they all share one important characteristic. AI system performance is architecturally dependent on memory subsystem performance and capacity. This has given lives to more complex memory hierarchy that is providing greater differentiation opportunities for Micron than at any time in our history. It has also elevated the role of memory in the AI world to a strategic asset.

Strong long-term demand growth structurally constrained supply growth and memory's strategic importance have caused customers to recognize that their product road maps rely on access to advanced memory technology and dependable and committed long-term memory supply. Micron has been a pioneer in our industry in creating a new class of strategic customer agreements, or SCA, with very robust terms. We are pleased to announce that we have completed 16 SCAs with customers across the data center, consumer and auto market segments. These SCAs accelerate the transformation of our business model, enhanced partnership in technology and innovation and provide customers with contracted supply assurance.

Typically, these agreements have a 5-year term from calendar 2026 through the end of calendar 2030. Automotive agreements generally have a 3-year term. The 16 signed agreements represent roughly 20% of our DRAM volume and 1/3 of our NAND volume over this period. These SCAs include four very large customers and three medium-sized customers. The remaining agreements relate to smaller customers from the automotive industry and represent our commitment to the important sector.

When completed, we expect approximately half or more of our company revenue to be under these SCAs with customers across end markets. Our customer value, our U.S. supply plans and this is reflected in our SCAs. These SCAs are structured as take-or-pay agreements, we are binding commitments to purchase specific volumes over this multiyear term.

The largest agreements generally have a ceiling price for existing products at the current CQ2 market price and a floor price through the term of this agreement. Several SCAs, which account for a modest portion of the SCA-related revenue, include either fixed prices or have no price bands associated with them where pricing will be subject to market conditions. When all planned SCAs are executed, agreements with either fixed prices or price ceilings at or close to current CQ2 market prices are expected to be approximately 40% of our revenue. For SCAs which do contain such bright bands, pricing is designed to stay within this floor to sealing level through the course of the term. This pricing visibility will help our SCA customers across market segments to better manage their business and grow their demand.

For our SCA with price bands, the floor price enables a very robust gross margin for Micron, well above our peak quarterly margins in any past cycle. Fourteen of the 16 SCAs that we have signed have a cumulative revenue at minimum price per hour contracts of approximately $100 billion over the remaining agreement term. They also strengthened our long-term financial performance, margins and free cash flow expectations with higher visibility and improved stability in our business performance.

Under the SCAs we have signed so far, we project to receive cash deposits and related financial commitments of $22 billion. This further demonstrates customer commitment to this new business model. Mark will provide additional details.

Our SCAs with customers across data center to consumer devices to auto and industrial applications create a new paradigm for us to strengthen our customer relationships. They provide committed DRAM, including HBM as appropriate, and NAND supply to our customers over a multiyear time horizon. In a period of significant shortage, this supply visibility is extremely beneficial to our customers. The visibility enables our customers to leverage SCA apply to make progress on their strategic plans, drive growth and enable their end consumers to benefit from their products and services. We are very appreciative of our customers who have worked with us through this period of tight supplies with a strong collaborative spirit to create win-win outcomes for the long term for the entire ecosystem and end consumers.

AI's insatiable appetite for memory bandwidth and capacity with low latency and low power is driving memory architectural choices, memory product mix shift and manufacturing process technology decisions, all of which increase the complexity of memory and storage road map for the industry.

Micron is building on its technology leadership. Our 1-gamma DRAM node and G9 NAND node are both ramping well and on track to become the highest volume nodes in Micron's history. Development of our next-generation DRAM and NAND nodes are also progressing well and are on track to begin volume production in the second half of calendar 2027.

We are leveraging our leadership DRAM and NAND nodes across our product portfolio. HBM4 12-high volume ramp is tracking twice as fast as HBM3E 12-high, and we have already shipped over $1 billion in HBM 4 revenue. We expect to reach mature yields on HBM4 12-high significantly faster than HBN3E 12-high. Please see our earnings press release for other high life across our HBM, high-capacity DDR and healthy server DRAM, data center SSD, PC, smartphone and automotive product portfolios.

We expect future memory demand will continue to skew towards higher performance and higher value products whose complexity carries higher cost per bit. Transitions like LP5 to LP6, DDR5 to DDR6 and newer generations of HBM all come with rising bit costs. This trend, along with the ramp of significant greenfield capacity in the years ahead, is projected to cause the blended DRAM cost per bit to rise from current levels. Our customer SCAs provide for appropriate price premiums for such new products to be negotiated in the future.

Turning to our end markets. AI is driving unprecedented growth in data centers with industry data center DRAM and NAND bit shipments in calendar 2026 expected to more than double from 2 years ago. Agentic AI is a structurally reshaping data center infrastructure, extending beyond accelerator-only racks to include CPU racks for the agents control plane and program execution and storage for rapidly expanding contact store. We now expect calendar 2026 industry server units to grow high teens percent above our prior expectations of low double digits, driven by mid-teens growth in traditional servers and even stronger growth in servers with AI accelerators. We estimate that this increase in our server unit growth expectation is enabled by a modest reduction in average server DRAM content growth as customers focus on maximizing unit shipments amidst a very tight allocation of memory. In NAND, AI context memory storage and HDD displacement opportunities are expanding the addressable market for SSDs.

PC and smartphone industry revenue is expected to grow despite unit volume declines, reflecting resilient demand for the high-end devices at higher prices across end device categories. Agentic AI platforms such as OpenClaw and NemoClaw elevate the value of edge devices, enabling improved economics greater privacy and latency and more efficient orchestration of AI between the cloud and edge. Over time, we expect the value of on-device AI combined with pent-up unit replacement demand to drive memory demand growth in PCs and smartphones.

In automotive, ADAS remains a powerful driver of content growth. L2+ and above vehicles, which feature is progressively increasing levels of autonomy have over 5x the memory and storage content of an average vehicle. The mix of L2+ and above vehicles is more than doubling this year to over 20% and is expected to exceed 40% by 2030. Average auto memory and storage content is expected to further increase as mix shifts towards higher levels of autonomy with progressively higher levels of content.

In robotics, continued advances in simulation, foundation models and integrated hardware and software stacks are accelerating physical AI. This creates a growing content rich opportunity for high bandwidth, low-power memory and storage that powers real-time perception, inference and control. Humanoid robots carry 10 times the amount of memory as an average L2+ vehicle, and we expect a sustained substantial multi-decade memory demand cycle to begin in the latter part of this decade.

Now turning to our market outlook. We now expect supply/demand conditions for both DRAM and NAND to remain tight beyond calendar 2027. In DRAM, we expect industry DRAM bit shipments in calendar 2026 to grow in the low to mid-20s percentage range, slightly above our prior outlook. In NAND, we expect industry NAND bit shipments in calendar 2026 to grow approximately 20%, unchanged from prior expectations. We expect Micron DRAM supply to grow approximately in line with the industry supply growth, while Micron NAND supply grows somewhat less than the industry supply growth in calendar 2026.

Our SCAs provide enhanced visibility on our long-term demand and provide us greater confidence on our CapEx and R&D investments. We are focused on maximizing output from our fabs, including collaboration with our suppliers to accelerate tool acquisition, fab to installation and ramp and tool replacements and upgrades to improve productivity. Recently, we concluded a multiyear EUV supply agreement with ASML, supporting our increased adoption of EUV at the 1-delta note and future generations.

We are also making good progress on expanding our global manufacturing footprint to increase supply over time. This includes our significant investments in U.S. leading-edge DRAM manufacturing with our ID1 and ID2 fabs in Idaho, whose construction is well underway as well as the first of our New York fab cluster where we broke ground in January this year. ID1 is on track for first wafer output in mid-calendar 2027 and ID2 in late calendar 2028. We recently launched first production starts of our 1-alpha DDR4 technology in our Manassas, Virginia, fab, which will add to our capability to support the legacy product needs of our customers in auto, industrial, medical, aerospace and defense markets.

In our newly acquired Tongluo site in Taiwan, we expect to support meaningful product shipments from the existing 300,000-square-feet fab in mid-calendar 2027, about a quarter earlier than our prior expectations. Adding to the existing fab, we have begun construction of a similar-sized second cleanroom at the site. This cleanroom will support EUV equipment. Our construction activities and time lines are on track for our other facilities in Japan and Singapore.

Complementing our advanced packaging capabilities in Taiwan, our Singapore site will become another center of excellence for advanced packaging. We expect this facility will contribute meaningfully to Micron's HBM packaging capacity beginning in the first half of calendar year 2027. As we make these investments, we will remain disciplined in our approach and will be responsive to the market environment to appropriately align our supply plans.

I will now turn it over to Mark for our fiscal Q3 financial results and outlook.

Thank you, Sanjay, and good afternoon, everyone. Micron delivered exceptional fiscal Q3 results with revenue, gross margin and EPS exceeding the high end of our guidance. Our results and today's outlook underscore the increasing value of memory in the AI era and the structural strength of our business.

As mentioned, we have entered into 16 strategic customer agreements. For SCAs with defined price, either fixed or subject to floor and ceiling pricing, in accordance with the revenue accounting standard, we are disclosing remaining performance obligations, RPO, starting this May quarter. RPO at the end of fiscal Q3 was over $5 billion. For the SCAs that we have entered into so far, including ones executed after the end of fiscal Q3, RPO is approximately $100 billion. RPO is determined based on minimum committed volumes and minimum pricing and reflects inherently conservative estimates. RPO is not indicative of the total revenue we expect to recognize in future periods. As such, we expect revenue to well exceed associated RPOs over the term of the agreements.

As Sanjay mentioned, we project to receive cash deposits and related financial commitments of $22 billion under the SCAs we have signed so far. The overwhelming majority of these commitments, approximately $18 billion will be in the form of cash deposits. When all targeted SCAs are completed, we expect to have substantially higher levels of SCA customer deposits and related commitments. These customer deposits will show up on our balance sheet more in fiscal Q4. The cash flows associated with customer deposits appear in financing-related cash flows, and will not affect our free cash flow. This cash will be returned to customers over time towards the latter half of the agreement term.

We are excited with our progress in signing these SCAs, which will strengthen our long-term financial performance and drive enduring robust ROI for the company over time.

Total fiscal Q3 revenue was $41.5 billion, up 74% sequentially and up 346% year-over-year, representing our fifth consecutive quarterly revenue record. The $17.6 billion sequential increase is the largest in our history, eclipsing last quarter's $10.2 billion record.

Fiscal Q3 DRAM revenue was a record $31.3 billion, up 343% year-over-year, and represented 76% of total revenue. Sequentially, DRAM revenue increased 67%. Bit shipments were up low single-digit percentage range. Prices increased in the low 60s percentage range, driven by tight industry conditions and favorable mix.

Fiscal Q3 NAND revenue was a record $9.9 billion, up 361% year-over-year and represented 24% of total revenue. Sequentially, NAND revenue increased 99%. Bit shipments increased in the mid-single-digit percentage range. Prices increased in the mid-80s percentage range, driven by tight NAND industry conditions and a favorable mix.

The consolidated gross margin for fiscal Q3 was 84.9%, up 10 percentage points sequentially. This improvement was driven primarily by higher pricing, and also benefited from continuing strong execution and favorable mix. Fiscal Q3 gross margin more than doubled from a year ago and was a new company record.

Now turning to quarterly financial performance by business unit. Cloud Memory Business Unit revenue was a record $13.8 billion and represented 33% of total company revenue. CMBU revenue was up 78% sequentially driven by higher pricing and bit shipments. CMBU gross margins were 83%, up 9 percentage points sequentially, driven by higher pricing.

Core Data Center Business Unit revenue was a record $11.5 billion, and represented 28% of total company revenue. CDBU revenue was up 103% sequentially driven by higher pricing and a favorable mix. CDBU gross margins were 87%, up 12 percentage points sequentially, driven by higher pricing.

Mobile and Client Business Unit revenue was a record $11.5 billion and represented 28% of total company revenue. MCBU revenue was up 49% sequentially, driven by higher pricing, partially offset by lower bit shipments. MCBU gross margins were 87%, up 9 percentage points sequentially, driven primarily by higher pricing and helped by favorable mix.

Automotive and Embedded Business Unit revenue was a record $4.6 billion, and represented 11% of total company revenue. AEBU revenue was up 71% sequentially, driven by higher pricing and higher bit shipments. AEBU gross margins were 79%, up 11 percentage points sequentially driven by higher pricing and favorable mix.

Operating expenses in fiscal Q3 were $1.5 billion, up $97 million quarter-over-quarter. The sequential increase was due to higher variable compensation expense from the strong performance of the business.

We generated operating income of $33.7 billion in fiscal Q3, resulting in an operating margin of 81.2%, up 12 percentage points sequentially and 54 percentage points year-over-year.

Fiscal Q3 taxes were $5.1 billion on an effective tax rate of 14.9%.

Non-GAAP diluted earnings per share in fiscal Q3 was $25.11, up 106% sequentially.

Turning to cash flow and capital expenditures. In fiscal Q3, operating cash flows were $25.4 billion. Capital expenditures were $7.1 billion, resulting in free cash flow of $18.3 billion. Fiscal Q3 free cash flow was a quarterly record for the company.

Ending inventory for fiscal Q3 was $8.6 billion, with days of inventory at 120. DRAM inventories are very tight and below 120 days.

We reached record levels of cash and investments of $30.2 billion at quarter end. During fiscal Q3, we reduced debt by $4.4 billion, including a cash tender offer that reduced senior notes by $4.3 billion. The weighted average maturity on our outstanding debt is April 2035. We closed the quarter with $5.7 billion of debt and a net cash balance of $24.4 billion. This fiscal year, we received upgrades from all 3 major credit rating agencies, including an upgrade to BBB+ on the strength of our technology and product position, financial outlook and strong balance sheet. Our balance sheet has never been stronger, and we project it to strengthen further even as we increase investment in technology and needed capacity.

Now turning to guidance. We expect fiscal Q4 revenue to be a record $50 billion, plus or minus $1 billion; gross margin to be approximately 86%; and operating expenses to be approximately $1.65 billion. Based on a share count of approximately 1.15 billion shares, we expect EPS to be a record $31 per share, plus or minus $1.

Our fiscal Q4 gross margin outlook reflects a meaningful moderation in the rate of price increases. We project operating expenses to increase by approximately $1 billion in fiscal 2027 as we expand R&D to support an unprecedented set of opportunities in memory and storage. We expect operating expense increases to be weighted to the second half. We expect the fiscal Q4 and fiscal 2026 tax rate of around 15%.

Micron continues to invest in a disciplined manner across our global footprint to address customer demand. As a reminder, our CapEx is net of anticipated government incentives. In fiscal Q4, we project CapEx of around $10 billion, bringing full year fiscal 2026 capital spending to approximately $27 billion. We expect quarterly CapEx in fiscal 2027 to be above fiscal Q4 levels, with more than half the increase year-over-year in fiscal 2027 from construction CapEx as we pull in cleanroom capacity required to address long-term demand.

We forecast free cash flow to increase substantially again in fiscal Q4. From December 9, 2026, the second anniversary of the signature of our definitive CHIPS agreements, we intend to increase our capital return. Over time, we expect to return 100% of our excess cash to shareholders. Any impacts that may occur due to trade or geopolitical developments are not included in our guidance.

I'll now turn it over to Sanjay to close.

Thank you, Mark. AI has elevated the value of memory. Micron is collaborating closely with our customers and suppliers across technology, product, manufacturing and commercial teams in the tight industry environment. Strategic customer agreements are ushering in an exciting era for Micron. We expect these SCAs to significantly enhance the durability and predictability of Micron's strong financial performance, accelerating the transformation of our business model. I'm thankful to Micron's team members worldwide whose relentless focus on execution on all fronts has positioned Micron as a leader in this new AI era, as we continue to advance our mission to accelerate intelligence to enrich life for all.

We will now open for questions.

[Operator Instructions] Your first question comes from the line of Timothy Arcuri with UBS.

Sanjay, so I think we're all trying to figure out how much is locked in, in kind of a floor price scenario over the next 5 years. And there were two things you said. You said that 14 of the 16 SCAs have $100 billion in cumulative revenue. So that sort of says like $20 billion a year at like a low price, which is way below the run rate that you just guided. So that says that not that much would be covered at a floor price, but then you also said that 40% of revenue will be moving inside of these SCAs. So can you maybe double-click on all that and sort of help us in like a floor price scenario? Can you help us think about how much of revenue per year would be guaranteed?

As we indicated that under these SCAs that have been completed so far, at the floor price, the revenue is projected to be $100 billion. But again, as Mark noted in his remarks, I mean, we expect revenue to be much higher than that. Note that at the floor price that our profitability levels and the gross margins and the floor prices are higher than peak margins at any time in the past. And so overall, about 20% of DRAM and about 30% of our NAND volume is covered in these SCAs so far. So that -- close to about 25% of our revenue that you can project over the term of these agreements. So again, RPO at the floor price is to be reported as an accounting measure, but we fully expect that the revenue will be much higher than that.

Got it. And then with respect to just how these layer in, Mark, like how much of the August quarter revenue, for example, will be flowing under SCA? I'm just trying to figure out how to layer that into the model. And when you get to look at full run rate where like by next fiscal Q4, will you be at sort of a full run rate of what's being covered under these SCAs, can you help us sort of like that in?

Yes, Tim. So you'll see a disclosure in the Q, which we'll disclose the next 12 months revenue associated with each set of agreements that have an RPO. So for example, for those that closed within Q3, you'll see an RPO of $5 billion -- over $5 billion. And you'll see the next 12 months associated with that of about $1.8 billion. And yes, that is because those are some of the smaller agreements that Sanjay mentioned automotive agreements.

Now in the fourth quarter, as Sanjay mentioned, you will see an RPO reported on the 14 of the 16 agreements, that is going to be about $100 billion. And there will be an associated next 12 months associated with those that will be disclosed in the K. So you will be able to see roughly how these are feathering in. And keep in mind, this RPO number, it is a minimally contractually enforceable amount for the intersection of volume and price. So you're looking, Tim, at a minimum number, and that's important to keep in mind, and we were clear that it doesn't reflect what we think will happen.

And then also each quarter, this RPO number will change. It will change based on contracts that are added in. It may change on additional volume commitments with the determined price. It will change based on shipments and how that RPO is declines after that performance obligations met. So you'll be getting a lot of additional reporting. This is all under ASC 606. I know it's something that we -- it's not a heavy standard typically in some of our reporting, but this feature of RPO, you will see.

I also want to emphasize, as Sanjay mentioned, that even at the floor price and eventually, we anticipate about 40% of our revenue being under this sort of RPO-related commitments. But even under the floor -- on the floor price, we expect the margins to be significantly above prior peak margins.

Your next question comes from the line of Joseph Moore with Morgan Stanley.

I also wanted to ask about the LTAs. Can you talk about the role of the cash deposits? Should we think of that as being sort of an escrow collateral account where people cancel you would have access to cash? Like if it's not revenue like sort of what is the point of the deposit? And what is the relationship of those deposits with the RPO if there is one?

Yes. Joe, on the deposits, so we mentioned that we have $22 billion of deposits and financial commitments associated with the agreement signed to date as of this call. $18 billion approximately of that is cash deposits. We'll receive those deposits. We received about $400-plus million in the third quarter. We'll receive about another $10 billion in the fourth quarter. And you will -- these will be seen cash deposits -- they'll be seen in financing cash flows that will not affect free cash flow. They will be -- they are held by us during the performance commitments of the agreements. And as those agreements are satisfied, those deposits will be returned over time but heavily weighted to the back half of the agreements. The difference between the $22 billion in the $18 billion, so roughly the $4 billion of others is letters of credit.

Okay. But what is the role -- I mean, what happens to that cash? It seems like they're putting a deposit and then they get the deposit back is -- what is the reason for them to commit that cash? Is that something where there's a take-or-pay that, that cash is related to? It's not a prepayment. Just can you help us understand that?

Yes. Thanks, Joe. It's not a prepayment. It's a separate commitment by the customers and a reflection of the fact that we have a binding agreement and these are take-or-pay agreements. And we hold the cash, and it's a reflection of our shared commitment to perform under these agreements. Now -- this is good for Micron, of course, these agreements and that we get visibility on our demand. It's committed volume that we can be confident about making our investments, large capital investments, closer technology relationship. It's good for the customers because they have supply assurance, they have leading technology. So in our view, it's a win-win. And very, very happy with the nature of the agreements and the impact they have on the business and indication of a transformed business model at Micron.

Your next question comes from the line of C.J. Muse with Cantor Fitzgerald.

Maybe just follow up on Joe's question, Mark. When you think about these cash deposits, do you view that as fungible cash and used for CapEx? And I guess as part of that, when you contemplate capital returns, particularly after December 14 kind of CHIPS at end date, do you -- will you include kind of that cash that you received in your gross cash thoughts and your thoughts around capital returns? Or is that something, given that you will have to return it eventually that would cause you to think steady state, you'll need to hold more gross cash all else equal?

C.J., it's unrestricted.

But does it change your thoughts around gross cash that you feel comfortable holding on your books?

Not in the near term. I think we, of course, are going to have what we view is adequate liquidity to support the operation of the business, that would include over time, returning the deposits as customers and Micron perform on the contracts. And so that, of course, is important. But -- and then we would hold liquidity to satisfy what investments we believe are important for the business. We've got a lot of -- we've got large projects underway to provide supply and also R&D programs. So -- and again, I'll emphasize that the customers, as I mentioned earlier, they will get this return deposit back in the latter half of the agreement.

Perfect. And then maybe as a follow-up on HBM revenues. Could you kind of share how you're thinking about both your market share and perhaps total revenues into calendar '26. And is there an expectation into calendar '27 that you can bridge margins there closer to what you're getting on D5? Or is that a place that will be permanently below that D5 level?

So with respect to HBM, first of all, very, very pleased with our HBM4 product and Micron's shipments already of HBM4 of over $1 billion. HBM market share, we strategically are choosing it to be close to our DRAM share. And this is important because of the trade ratio of HBM, it consumes, as you know, a significant amount of wafers and puts pressure on non-HBM supply in the industry. So targeting our HBM share close to our DRAM share strategically enables us to supply our diversified end market customers across all end markets, data center, consumer, automotive, industrial, the markets that need non-HBM supplies.

Regarding your question on pricing for next year. We are really not commenting on pricing. But certainly, HBM is a product where Micron has a strong leadership position. We have demonstrated tremendous success now with HBM3E 8 high, HBM3E 12 high and now with HBM4 and a strong road map ahead of strong confidence in our ability to execute to that. And it will -- it is a higher price product compared to non-HBM on a per bit basis. And it is a product that is critically important for the entire AI ecosystem from data center to edge. So strategically, it is a very important product for us, and it is a product that does provide strong ROI as well.

Your next question comes from the line of Vivek Arya with Bank of America Securities.

For the first one, Sanjay, you mentioned, I think, four large and three medium-sized customer agreements, and I'm curious how many of them are related to the data center, should we expect more data center-related announcement. And the $100 million, does that align with the large and medium size or does it align with the smaller size customers? I guess I'm just still trying to figure out what is the typical SCA with the data center customer? Like have you given enough bedrums for us to figure out a data center SCA looks like over the next few years?

So our large customers include data center and the large and medium customers that you mentioned and, of course, including our smaller customers, they go across data center, consumer and automotive markets. And we have provided you a color on the large agreements have, of course, generally have ceiling price -- have a price band, which has a floor as well as ceiling and the ceiling is established in the CQ2 price levels. And of course, that you know that CQ2 price levels are reflected in our [ Q3 ] results as well as guidance [ Q4 ] and they provide for unprecedented levels of profitability. And those price bands also provide for floor prices, which -- where the gross margins are well above the peaks at any past cycle in company's history. And the large agreements that we mentioned, these are multiyear agreements, and they provide us tremendous visibility to demand, customer commitments. And they, of course, come with the financial commitments, including cash deposits, that Mark elaborated on further -- earlier.

And for my follow-up, Mark, on gross margins, 86%. Does it kind of hang out here for a while? Is there a ceiling? And then as these SCAs start to kick in, should we assume some kind of normalization to between the mid-80s where you are now versus, I think, the prior peak was in the low 60s. So as your long-term investors build their models for '27, '28, et cetera, should they be assuming a normalized gross margin range somewhere in the mid-70s, right, kind of the range between where you are today versus the prior peaks? If you could just hold our hands on how to think about gross margins beyond this 86% in the near term? And then longer term, what is the right way to think about how these gross margins unfold.

Yes. So Vivek, we're not providing guidance beyond the fourth quarter. But we are at margin levels that as we've talked about before, incremental price yields less in gross margin expansion. So while the -- but having said that, we do see, as we mentioned, we updated our view on market conditions that we expect the market to remain tight beyond 2027. We also have -- we're at a point where memory is very much appreciated for the strategic asset that it is, the value that it brings to improving AI intelligence and more and higher performance memory is needed. And so our continued deployment of bits to data center and edge device higher performance applications is going to be helpful as price moderates and price growth moderates, and we move to optimize the placement of our bits with customers, including those that we do these have done these SCAs with.

And then also, as we've talked about, we will get additional volume starting midyear materially beginning midyear '27 that will grow into '28. And we will have some start-up costs there, but we will get absorption is that -- as those ramps occur. And so over time, we'll get that operating leverage. So I think we feel great about the trajectory of the business. Micron's technology position world-class product portfolio. And you can see we're operating very well, and all those are supportive of continuing to deliver a strong financial performance.

Your final question comes from the line of Krish Sankar with TD Cowen.

I had two of them, Sanjay or Mark. Congrats on the great results. On the floor prices for the LTAs, you said about the prior peak, your prior peak gross margin was somewhere in the low 60%, 62% range. If I try to plug in what a 64-gigabyte server DRAM is, I can get like a $700 price for it compared to $1,500 today, which kind of puts you at like $10 to $12 a gigabyte as the floor and a mid-$20 a gigabyte for the current price. Is that the range we should think about for these LTAs, i.e. low teens to mid-20s dollars a gigabyte is kind of like the range of LTAs for the pricing?

So Krish, we're not going to get into specific pricing discussions, but I just want to note again that I said that the gross margin at the floor will be well beyond the peak that we experienced -- the highest that we experienced in the past cycle, so well beyond those, right? But we are not going to obviously get into the specifics related to the pricing. Bottom line is these SCAs really help provide visibility, strength and durability of demand for us. And they absolutely fundamentally accelerate our financial performance and financial -- the business transformation here.

Got it. Very helpful, Sanjay. And just a quick follow-up. You kind of mentioned how DRAM bits should grow low to mid-20s, probably in the 20% range this year. And clearly, we are undersupplied on both. Is there a way to quantify what happens in 2027? Is there a way to say is the undersupply going to double what it is this year in 2027? Or how do you think about the supply/demand imbalance in 2027?

We see 2027 overall tight. We have said we see tightness continuing beyond 2027. Working hard to bring up supply, but we have shared with you that it takes a long time to bring up the additional capacity that is needed to support the customer demand, the additional wafer capacity. And of course, technology transitions and the less bit gain that they gave per node as well as the HBM trade ratio for tremendous pressure on the overall supply growth as well.

So supply even in 2028, when supply begins to improve gradually, we see that the demand will continue to be on a robust trajectory as well because these AI trends are very long-term trends. AI is still in very, very early innings. The whole token economics needs more memory. Here, system -- AI system performance is really very much limited by memory capacity and memory performance, memory bandwidth. So the demand for memory -- as the compute demand grows and our customers look at a tremendous transformation opportunity that is ahead of them, this -- and continue to make investments like they have never made before to build this infrastructure. The demand trajectory is extremely strong. Memory is at the center of it, and this is a strategic asset and access to memory supply is obviously a critical priority, as you can see in the multiyear agreement that our customers have concluded with us. I mean those agreements reflect the confidence in the growth of the demand. So we are working hard to bring up supply, but we see tightness persisting beyond 2027.

This concludes today's call. Thank you for attending. You may now disconnect.

All right. Good morning, and welcome to JPMorgan's 54th Annual Technology, Media and Communications Conference. My name is Harlan Sur, semiconductor and semiconductor capital equipment analyst for the firm. Very pleased to have the team from Micron Technology here with us this morning. Manish Bhatia, Executive Vice President of Global Operations for Micron, and we also have Samir Patodia, Senior Director of Investor Relations, here with us as well. Manish and Samir are going to kick us off with some opening comments, and then we'll go ahead and kick off the Q&A.

So gentlemen, thank you for joining us today. Samir, let me hand it off to you first.

Thanks, Harlan. So I'll start off with the safe harbor and then pass it to Manish. We'll be making some forward-looking statements today and those statements have risks and uncertainties associated with them. So we refer you to the risk factors disclosed in our filings, including our recent 10-Q and 10-K. Manish?

All right. Thanks, Samir. Standing room in the back, thanks for your interest in Micron. Our financial outlook has strengthened since our last earnings call. We're on track for another substantial record free cash flow in fiscal Q3. Our balance sheet has never been stronger, and that's underscored by multiple credit rating agencies actually from all 3 -- upgrades from all 3 credit rating agencies this year. Demand continues to outpace our ability and the industry's ability to supply due to persistent structural factors. And so we expect tightness for HBM, DRAM and NAND to continue well beyond calendar year 2026.

Perfect. And so that was a great update. And so I think that we can go back and we can ground ourselves in terms of the near to midterm outlook that the team put out back in the mid-March sort of earnings call. And I think at that time, relative to the guidance put out for the May quarter, I think the view was that the strong revenue growth outlook at that time was going to be driven by predominantly pricing, a little bit of volume growth. And so let's start off there. Relative to that, I mean, which of the dynamics has played out a bit stronger than expected or stronger than expected?

Yes. Well, certainly, pricing has played out. But demand is continuing to be very, very strong. And the AI world is transitioning from just human interactions to human-to-agentic interactions and even machine-to-machine interactions. And so -- and that -- those agentic workloads are driving inference and inference is becoming a bigger and bigger part of the demand workloads for our customers.

And we're just seeing memory as continuing to become an even more strategic asset, adding even more value as memory can be equated with intelligence and higher performance memory, higher reliability memory, higher capacity memory, all of those help add to the accuracy of intelligence. And that's true on both the high-bandwidth memory side. It's true on the traditional DRAM side and DDR. And it's also true on the NAND side where context windows are getting longer and NAND flash is a key part of being able to not just store all the new data that's coming up, but improve the accuracy of the results of the models.

On the midpoint of the guidance for the May quarter, $33.5 billion, 81% gross margins, $19 and change of earnings power. Any way to quantify the better strength that you've seen in the May quarter relative to some of the parameters -- the financial parameters that you put out back in the March?

Well, our quarters are going to close in just a couple of weeks, and then we'll have comments on that, obviously, on our next call.

Okay. And so then maybe the way to -- maybe the better way to frame it is, let's think about it over a multi-quarter period of time. Clearly, the environment, as you mentioned, has gotten better for you. And I think that again, going back to the March call, you had talked about demand for this year being supply constrained, but you could still see the industry shipping sort of 20% sort of bit shipment growth in DRAM and low 20% sort of bit shipment growth in NAND. Has that viewpoint changed given the dynamics that you've seen over the past couple of months?

I think what's -- the demand is supply constrained, right, for both right now. And so we're just continuing to work very hard on enhancing supply. But I mentioned persistent structural factors earlier in my comments. What I really mean on the supply side is that technology transitions for both DRAM and NAND are delivering less and less productivity.

That's right.

And they had before, less bit growth improvement per technology node than nodes of the past. And that -- on top of that, for DRAM, we have the structural shift towards HBM, which in order to deliver the high performance of HBM, we have a trade ratio where HBM die sizes are larger, and therefore, you end up with fewer bits per wafer. So it takes more than 3x as many bits per -- more than 3x as many wafers to be able to deliver the same number of bits and that trade ratio continues to grow as you go from HBM3E to HBM4 to HBM4E. That trade ratio continues to grow. And so that all leads to needing more greenfield wafer capacity in DRAM. And greenfield capacity needs more clean room space because you're putting in full lines rather than incremental equipment to get the incremental benefit of the technology transition, you now need more clean room space because you're putting in full lines, and that takes more time.

So we have the strong demand drivers and continuing to strengthen because of the value memory provides to the end AI solution, whether that's training or inferencing. And then you have the structural factors in terms of supply, where the whole industry is still trying to catch up, and we don't see the ability for the industry to catch up to -- supply to catch up to demand for the foreseeable future.

Yes, it is pretty amazing since the second half of last year, and we had Lip-Bu at our -- we were just talking about it, we had Lip-Bu at our keynote yesterday, and we were talking about this significant increase in memory, storage and CPU demand, literally the moment that inferencing -- AI inferencing workloads crossed over training workloads in the second half of the year. It's like the light switch went on and all of a sudden, memory intensity went up, storage intensity went up, CPU intensity went up, right? And that trend has continued, right, because as we track inferencing workloads, I mean, they're continuing to grow like at this exponential pace, and we can understand why the continued strong demand pull from memory. The team talked about key customers only able to secure 50% to 2/3 of their bit demand requirements in the medium term. Is that actually still the case? Or has that gap actually widened?

Well, we talked about the demand drivers getting even stronger and the supply continuing to be very, very difficult to increase. We're all working very, very hard. You had Tim before talking about how closely we work with them and we work that closely with all of our suppliers, and we're all -- we're working very hard to try and catch up, but we don't see that supply being able to catch up to demand for the foreseeable future.

Yes. So on that front, and obviously, Manish, you're Head of Global Operations, let's talk about some of the ways that you will be able to unlock the potential for more bit supply over the coming quarters, right? On your acquired, for example, Taiwan fab in Tongluo, the target was to start production bit shipments sort of second half calendar '27. How should we think about the first few quarters of the initial ramp? We've heard 10 -- starting off 10,000, 20,000 sort of wafer starts per month. Your new Boise fab will be ramping about that same time as well. How much output should we anticipate on the first phase of that build-out? Right now, most of these again are for second half of next calendar year, but give us an update on these 2 initiatives.

Sure. And Tim did -- again, had a great job having Tim go before me because he mentioned that we did make the acquisition of the Tongluo -- the Powerchip Semiconductor Tongluo site earlier this year. That transaction actually closed ahead of schedule, very pleased with the support we got from the Taiwanese government for that. And we're making excellent progress towards that target of having -- converting that fab from the logic output that it had before towards being able to do leading-edge DRAM. And we expect to be able to have production. We're making excellent progress towards production in the second half of calendar '27. And we also announced on our last call that we're going to be building sort of a twin fab immediately next to the existing fab there as well, and we're starting construction on that this summer.

So that site will become a larger leading-edge memory site. It's only 20 minutes from our existing Taichung operations. So it's actually really going to be run like a mega cluster site there. But we're doing a lot even beyond just what we're doing in Taiwan. You mentioned Idaho.

That's right.

Idaho 1 is making good progress. We -- earlier this year, we had announced that we had pulled in our date -- target date for wafer output to -- from second half of 2027 into mid-2027. And then we announced that Idaho 2 construction is underway. Actually, ground preparation is underway right now, and we expect to have that -- wafers out from that late in calendar year 2028. So a little bit more than a year after Idaho 1.

We broke ground earlier this year in New York, and we're actually ahead of schedule there and making good progress and expect to pour concrete later this year there. Really good progress in Singapore on our high-bandwidth memory facility, which we had broken ground on a year ago at the very beginning of calendar year '25. We expect to have production impact from that in calendar year '27 to supplement the HBM operations we have existing in Taiwan. We broke ground on a new NAND fab earlier this year in Singapore as well. So we have a lot going on to try and get supply across DRAM, HBM and NAND.

Yes. And on top of all of that, right, you're executing on all of these new technology migrations at the same time, right? And so to that point...

We're on world-leading technology transitions, right?

Industry-leading technology transitions, yes.

And so on these new process technologies, I mean, 1-gamma for DRAM, G9 for NAND, right? You were targeting both of these technologies to represent the majority of your bit mix by mid this year, very close, obviously. We're almost there. Is the team going to execute to this shipment mix? Can you talk about the progress on yield, manufacturability? And does the insertion of EUV actually relax a little bit of the need for multi-patterning technology? And has that also been a tailwind for yields and defect density?

Yes, yes. So we're absolutely on track for those goals with both our 1-gamma DRAM and our Gen9 NAND to become -- to crossover in terms of bit output by middle of this year. And both of them are ramping very well. Both of them in terms of yields ramped faster than prior nodes. And I always like to highlight, it's not because those technologies somehow got simpler. It's because of the tremendous engineering work, some of the technology development work, the collaboration between technology and manufacturing as well as our use of AI across -- which we've been doing now for many, many years, actually more than 10 years to be able to continually enhance our ability to get yield improvement, have productivity enhancement, improve quality, all those things in all of our manufacturing operations.

And we expect 1-gamma to become the highest volume DRAM node on a total wafer out volume basis that we've ever had of any node in our history. And we did add -- we did comment on the last earnings call that we are very pleased with our EUV progress on the 1-gamma node. It's the first node, which we introduced EUV. And so we're very pleased. We're very pleased with our decision to implement at this point in time when we had more modern, more advanced tools from ASML that are actually giving us better performance, better availability.

So given that, we had said on the last call that we expect to increase our deployment of EUV in 1-delta and future generations. And so to go along with that, we actually have recently concluded a multiyear EUV supply agreement with ASML that supports our technology and capacity plans over multiple generations.

Let's talk about part of the manufacturing excellence, right, that we know about Micron is new technologies, ramping new technologies very quickly. But there's the manufacturability, operational efficiencies, yield side of things, right? And given the strong pricing environment, I mean, there's not been as much focus by the market on operational and manufacturing efficiencies, right? Also, given the significant mix differentials within DRAM, for example, DDR5 versus HBM versus SOCAMM, right? So the cost profiles across all of these are very different. But we want to get an update from you on operational efficiencies. What operational efficiencies has the team executed over the past 12 months? And on a go-forward basis, how should we think about like-for-like cost per bit declines across your DRAM and NAND franchises?

Yes. So again, you had Tim on before. I'm in conversation with all of our core equipment vendors at the C level on every 2-week basis, focusing on trying to address the opportunities we have ahead in terms of greenfield, ensure technology transitions are able to move and move forward as well as enhancing the productivity of our existing fleet. And so we are working across all those areas to be able to meet this surge in demand.

You're right. Yield improvement is part of that. Productivity improvement is part of that. All of those things in this environment, accelerating technology transitions, all those together do help with our output and our cost. But even with that, we still don't see the ability to be able to catch up with the demand that continues to grow for the foreseeable future.

And so while cost is definitely always a focus, it's a religion for us. We're focused on trying to be able to increase supply to be able to meet our customers' growing demand, and we don't see the ability for that to align. Now we now see the tightness in supply to continue well beyond calendar 2026.

Yes. So any way for us to think about as we work forward on our models, any way to think about sort of like-for-like cost per bit declines across -- it's such a more varied product profile. Is there a back-of-the-envelope way to think about sort of...

It is getting more difficult. And you're right, because the product portfolio is getting -- I mean, I talked about HBM and part of DRAM, right? And HBM is obviously much larger die sizes lead to the trade ratio, much more complex packaging. All those things lead to different challenges there. Within the DRAM segment, there's also differences even within the DRAM portfolio between LP and DDR, high-capacity DDR.

And then you have -- on the NAND side, you have a big, big upgradation between QLC that's being used for client applications all the way up to performance TLC that's used for data center for PCIe Gen6. We're actually having a lot of success as well. So it's getting more difficult to think about that. And I think we're just really focused on being able to provide a higher-value product. And of course, cost will always be critical, but we're trying to move our portfolio over time. We've been successful. We continue to move our portfolio towards higher-value solutions.

Micron talked about on the last earnings call, securing their first strategic customer agreement. You guys call it an SCA, right? It's a multiyear agreement with specific commitments. I assume the parameters are duration, volume, pricing back in mid-March. You talked about securing your first SCA customer. That's 5 years in duration. Since then, how many more SCAs have you secured? And how much of your bit shipment target over the next sort of 12 months do these SCAs sort of now cover?

Yes, yes. So you're right. At our last earnings call, we talked about having secured our first strategic customer agreement, and that was a 5-year agreement with a large customer. And we've made meaningful progress on SCAs with other customers who are very interested in having that kind of strategic relationship with Micron. And we'll provide an update on that in the future -- on those in the future.

But I think just in terms of the parameters and kind of the way you framed it, I think one of the things that the SCAs really enable us to do is to improve our planning, right? We are planning the -- I just outlined all these different capacity expansions we have that are going to take multiple years to build up the fabs and then to equip them. And having these longer-term agreements with specific terms help us to align the capacity we're putting in place in terms of both greenfield clean room space and then, of course, how we're equipping the fabs helps us to better align supply and demand out in the future.

When we speak with some of your customers, especially the large GPU/XPU suppliers, they're telling us that the entirety of all of current gen mainstream HBM3E and HBM4 is sold out for next year, right? The only thing that remains under discussion is next-generation HBM4E because this is the next-generation architecture, there's a certain amount of customization, evals, quals are going to take a little bit longer. But just like HBM4 took a little bit longer last year, but is it fair to assume that the team's HBM3E, HBM4 supply outlook for next year is already fully committed volume and pricing?

So we're not commenting on 2027 HBM at this time. But I do know that in the past, we did conclude calendar year '25. Prior to calendar year '25, we concluded calendar year '26 in the kind of CQ4 time frame of calendar year '25. And at that time, that included both HBM3E and HBM4, and the timing that enabled the mix to be aligned between the different technologies for our customers. And so we're not going to make -- break that out specifically right now, but just know that that's been our history and track record in terms of how we've been able to align the mix of new technologies ahead of the year -- the last 2 years.

Yes. Okay. That makes sense. The new HBM architecture ramping this year is HBM4. You started shipping HBM volumes into the upcoming Vera Rubin GPU platform starting in March of this year. How has the volume ramp progressed so far? It seems like some of your peers have had some issues on performance requirements. Any issues for the Micron team as you've continued to scale up volume shipments of HBM4?

So we announced earlier this year that we began shipments and shipments are underway. And our HBM4 production ramp has actually ramped twice as fast as HBM3E 12-high did last year, right, which was the ramp we were very, very pleased with, and we're going at twice that rate in terms of production output. And our yield is also improving faster there. So -- which we had said we expected the HBM4 yield to come up faster. I know it's a more complex product. We said that we expected it to come up faster.

And based on the learnings that we had from HBM3E 12-high, we've -- so we've realized that as well. And we're just excited about our HBM4 product. It's on our 1-beta platform, which has been a really strong workhorse for us. And it has an in-house optimized base die that together is giving us excellent -- we think it's an excellent product.

And I think you did say in your opening commentary that part of the better outlook, better demand profile that you're experiencing this quarter was contributed by HBM. Is that correct?

I didn't give any specifics on the overall outlook.

Right. I think you said better HBM and better overall DRAM portfolio, but includes HBM and non-HBM.

I didn't sort of say specifically for the quarter, but in general, our portfolio has improved over the last few years as we've enhanced with HBM, as we've enhanced with data center SSDs. I think our overall portfolio has strengthened.

Sticking with HBM, HBM4E is the next inflection. It will ramp in calendar '27 on your next-generation 1-gamma process. And importantly, you've talked about HBM4E customization options offering further differentiation opportunities, deeper engagements with your customers. The base logic die offers customization, and that means that the GPU and XPU customers have to engage with Micron 12 to 18 months earlier in their design cycle, sharing IP, co-optimizing the design process flow and so on. So are you at a point where customers have decided on their memory partners to execute the base die design and, therefore, your HBM4E architecture? And if so, like what's been the breadth of your customer design wins?

Yes. So HBM4E development is well underway, and we expect it to ramp in calendar '27. The first product that will ramp on that, by the way, is going to be a JEDEC part, right? So that will be a JEDEC standard part that will see the first HBM4E part to ramp.

So no customization is what you're saying?

No. Actually the first product will be a JEDEC standard part. We are working with customers on customization, but the first product will be JEDEC. And you're right, it will have our 1-gamma DRAM, which, as I mentioned before, we're really confident. And by then, it will be a terrific node for the high performance that HBM4E will provide. And we're using TSMC for the logic die as well. And we are expecting that to ramp in calendar year '27.

Would that -- are you expecting to use TSMC both for the JEDEC-based base die as well as the customized base dies as well? Or is that...

Okay. Yes. That's right, the JEDEC-based part as well.

Given the customer-specific design of the base die for HBM4E, like how does the team think about the potential margin profile compared to a standardized base die design, right, that is leveraged across multiple different customers, right? Will the Micron team price HBM for -- I know the first one is JEDEC compliant. But beyond that, once you start going custom, will the Micron team price HBM4E that reflects the customization for every customer? Or will the team leverage other -- and will the team leverage other design resources like some of the custom ASIC companies like a Broadcom or Marvell or some of the other guys that do have a track record of quick turnaround customization capabilities for advanced logic SoCs?

Yes. And I think what you're seeing with the interest in HBM, the interest in ever-increasing performance and the potential for customization is just how valuable that performance is to the end accelerator companies and their customers, right, who are actually driving the AI road maps. I think that's definitely -- you're seeing that -- how the value just continue to increase. And what that really comes from is the complexity -- what that drives is the complexity in engineering, whether that's on the design side, process side, the core process side, which is absolutely critical, whether that's 1-beta, 1-gamma and then advanced packaging, right?

And I think the Micron team has done a tremendous job over HBM3E and now HBM4 of being able to drive innovation and proven track record of being able to execute on all these areas, whether that's driving innovations in design, having really, really strong core DRAM development and then novel advanced packaging to be able to execute on these products and provide value. I think the potential for customization is the same sort of same next gen [ evolution ] of that with even more value being able to be provided. And so as we think about the path that, that can take, we think that, that could end up -- our customers will be willing to pay for that.

Let's turn to your flash memory storage franchise, right? On data center SSD in particular, this is an area where Micron has seen significant share improvements over the past several years, right? Your data center SSD share has gone from 5% to 7% in '22 to 10% to 12% in '23, exiting last year at 15%, right? So very, very strong share gains and performance. And it comes at a time, right, where we're talking about this inferencing of inflection, right, in terms of AI compute workloads. And all of a sudden, whether that's KV cache offload and so on, right? There's all of a sudden this very, very strong demand pull for enterprise SSD.

So again, team exited last year with 15% global market share. You're now the third largest global market share leader in enterprise SSD. You've got your 9550 PCIe Gen5 SSD is out there doing well. You were first to market with your Gen6 9650 platform based on your G9 technology. And actually, that is the part of NVIDIA's STX reference platform.

You've always had good base NAND process technology. But the other big differentiator is, as we know, is you've got to have great controller design. And even more importantly, you have to have a great software and firmware stack. And so does the team expect -- so how did the team, over the years, acquire these sort of capabilities? And more importantly, like does the team expect continued share gains going forward?

Yes. So Harlan, every year when I come, you always ask me this, and I appreciate that you've been able to, for years, been seeing the progress that we've been making, and you've always called it out and highlighted it and definitely appreciate. You recognizing that for us because it has been a deliberate part of our strategy over a decade or so to be able to enhance our own internal ASIC capabilities, our own internal firmware development and not just target for me-too kinds of products, but leadership in terms of our ability, like you mentioned PCIe Gen6 to be qualified, lead qualified on NVIDIA's STX platform. And I think that, along with share gains across all the other areas, as you said, going from 5% to 10% to mid-teens really, really shows our track record of deliberate strategy as well as strong engineering execution, coupled with strong technology road map on the NAND side and manufacturing capabilities on the NAND side.

But I think the more interesting thing is really what you mentioned about what NAND now means and the high-performance SSDs mean to all of these AI workloads, right? And the agentic AI paradigm shift, you talked about the key value cache, right? It all leads more agentic workloads and the importance of the key value cache is just all about extending the context window. And the context window has grown 30x a year, right? To give you a sense of how much -- and the context window growing is about accuracy and accuracy is about value, right? So NAND is not just about being able to store more data that's generated by various different workloads, whatever they may be. It's also about improving the value and improving the learning rate, and that's becoming a more important part of the AI technology stack.

As we see new tiers of storage and memory emerge as the inferencing workloads continue to evolve, right, we are seeing new memory and storage architectures that may be required, right? One of those is something like high-bandwidth flash. We're hearing more about high-bandwidth flash. But is the team exploring new architectures like high-bandwidth flash? As you work with your cloud and hyperscale customers looking at the evolution of their workloads, are you working very closely with your customers? And how do you see this sort of evolution unfolding?

Yes. Yes, so high-bandwidth flash is an interesting technology. It has some positive attributes, clearly providing larger capacity and taking advantage of advanced packaging to be able to try and address some of the AI opportunities that, of course, NAND cell by itself is -- it does have some limitations versus DRAM. And so those are things that engineering needs to work through.

But I think the bigger picture here is that there's demand for innovation across the memory hierarchy, right? And I think that's just a positive across every single tier. There's more opportunity and innovation. And I think for a company like Micron that has such a strong portfolio and a history of being able to execute in terms of research and development as well as understanding all different types of memories, I think we're very well positioned to be able to be a leader in whichever trend ends up becoming an area that adds the most value to AI technology stack.

Yes. And to that point, you talked about key value caching. And the first implementation of that was we've seen new storage architectures, right, like NVIDIA's STX architecture, of which you guys are a part of it. But we've seen other hyperscaler companies taking advantage of it by using this new CXL technology, right, memory expansion and memory pooling technology to not leverage SSD, but to leverage more DRAM. So to create these DRAM pools, right? And then leveraging the CXL controller and pooling technology to essentially store a lot of that sort of KV cache sort of implementation.

So there, you kind of benefit both ways, right? Whether your customers choose to -- on these higher like high-volume like KV cache offloading capabilities, it looks like it's morphed into, yes, we can do it with DRAM offload acceleration or we can choose to do it with enterprise SSD. But is that kind of the innovation that you're seeing in the industry?

Yes. Absolutely. I think what you're seeing is that there's no one standard workload or different cloud providers have -- and different transformer companies each have their own workload challenges that they're trying to tackle. And so they'll come up with architectures that match their -- the long-term workloads that they'll see. And so you can imagine that you're going to see innovation up and down that memory hierarchy.

And again, I think we're going to be very well positioned given the breadth of our technology portfolio and the engagement we have with all of these customers, right? I mean by Micron having moved to leadership now 4 generations in DRAM, 3 generations in NAND with leadership -- of leadership, that's enabled us to be helping define those architectures with our customers. We're not just sitting behind. JEDEC spec on every single one of these. We're now defining -- working with all these different customers on their workloads, helping them think through what can optimize their tokenomics, what can optimize their end applications as they see AI unfolding, whether that's in training or whether that's in inferencing. And so I think that's the exciting part is that memory has -- it's a strategic asset now to help everyone optimize those -- their end -- the intelligence that they're driving. And that's why there's never been a better time to be in the memory industry, and there's never been a better time to be at Micron actually.

Given the tightness in NAND, I assume that the team is also working with customers on SCAs for NAND as well. Have you secured any SCAs on NAND and SSDs?

Yes. We're making meaningful progress on SCAs, and we'll have more comments in the future on this.

Perfect. Well, we are just about out of time, Manish. Samir, I appreciate the participation, as always. And 2026 is shaping up to be another very strong growth year for the Micron team. So hope to be -- hope to continue to monitor the team's execution on that. Thank you very much.

Thanks, Harlan.

Thanks, Harlan.

Ladies and gentlemen, thank you for joining us, and welcome to Micron's post earnings analyst call. [Operator Instructions] I will now hand the conference over to Satya Kumar, Investor Relations. Satya, please go ahead.

Yes. Thank you, and welcome to Micron Technologies Fiscal Second Quarter 2026 Post Earnings Analyst Call. On the call with me today are Sumit Sadana, Micron's Chief Business Officer; Manish Bhatia, EVP of Global Operations; and Mark Murphy, our CFO. As a reminder, the markers we're discussing today include forward-looking statements regarding market demand and supply, market trends and drivers and our expected results and guidance and other matters. These forward-looking statements are subject to risks and uncertainties that may cause actual results to differ materially from statements made today. We refer to the documents we have filed with the SEC, including our most recent Form 10-K and upcoming 10-Q for a discussion of risks that may affect our results. Although we believe that the expectations reflected in the forward-looking statements are reasonable, we cannot guarantee future results, levels of activity, performance and achievements. We are under no duty to update any of the forward-looking statements to confirm these statements to actual results. Operator, we can now open the call up for Q&A.

[Operator Instructions] Your first question comes from the line of Melissa Weathers from Deutsche Bank.

I wanted to touch on the NAND side of things. So you guys are one of the only players to openly talk about greenfield capacity adds so far. And then there are some industry participants saying that the growth in manned bids going forward can be served by node upgrades alone without the need for capacity add. So can you talk about the decision to add greenfield capacity there? And what kind of trends you're seeing that gives you confidence to add that capacity?

Melissa, this is Manish. I can -- let me talk a little bit about the decision and then maybe Sumit can talk about the NAND demand trends. So we have said that in the past that while DRAM required greenfield wafer capacity, meaning new wafer start capacity growth to be able to meet the long-term demand trends that we saw NAND was able to meet that with technology transitions. And I think we still see that NAND technology transitions are able to provide strong bit growth going forward. And so our decision here while yes, reflecting confidence in the market demand outlook, which Sumit will talk about, was also driven by our continued space consumption for those technology transitions and for the next technology transitions that we'll be be having in the future as well as our decision to locate more of our NAND R&D in the -- in Singapore, where it's closer to our manufacturing. And so those 2 drivers of additional clean room space combined with our outlook on the market, we're really what was behind our -- and our confidence, frankly, in our product portfolio, which I'm sure Sumit will talk about we're behind our decision to be able to break ground for our new NAND fab.

And just on that, it's not a greenfield site, right? I mean you mentioned greenfield in your question, and this is the existing site where we are adding this additional clean rooms that Manish said. So in terms of the demand side of the picture, I mean, we see very robust demand for NAND driven by growth in the data center. AI servers are using huge amounts of SSDs, high-capacity SSDs as well as high-performance and this is working really well in our favor because our portfolio is doing exceptionally well. We are the first company in the world to have Gen 6 SSD in the market. And this has been something that works well with NVIDIA systems. We have seen tremendous demand for it that we are not able to completely even come close to meeting. And then even high-capacity SSDs, we have seen tremendous growth and tremendous demand and we have come out with a slew of new products that we are looking really excitedly to continue to grow our share in the data center SSD space from record levels that reached at the end of last calendar year, calendar 2025. We have been growing to record shares in the data center every year for the last 4 years. So we are very excited about that and our supply is nowhere close to being able to meet the demand that we see for the foreseeable future. And so this expansion that Manish mentioned is something we will put to good use to continue to grow our business with obviously focus on disciplined investing when it comes to CapEx going forward.

And I should just add that even though we're breaking ground now, that capacity -- that clean room isn't expected to provide new boosts to our capacity until the second half of 2028. So we do think that not just for us but across the industry, in NAND, premium space over the medium term is still going to be a challenge for the industry, particularly as many in the industry have redirected some NAND cleaner base towards DRAM.

Perfect. And then back to the DRAM side, I wanted to ask pretty candidly on the pricing that we're all seeing. Next year, we have line of sight to a couple of DRAM projects coming online, including you guys bringing on a couple of those new fabs. So how are you internally modeling the impact of that increase in supply going forward? Should we are you guys thinking pricing could come down next year or flatten out? Or do sell? Just any color on how you guys are internally modeling the impact of all those new fabs coming online next year and the year after?

Maybe I'll answer the supply side or what the industry, and I think for the projects that we've talked about, which are the new Idaho show in the DRAM side, Idaho Shell as well as the -- our ramp in the Power Chip too facility that we've now acquired. We've said that the supply impact from those will be towards our fiscal '28 in terms of being able to meet revenue shipments. And I think that's sort of largely the case for the industry in terms of new new large cleanroom projects coming online, they will be really impacting later in 2017 and really into 28 before you see meaningful supply, which is one of the reasons we have reiterated that we see the tight supply conditions to continue beyond '26.

Yes. I think we obviously don't provide any kind of price modeling or anything else type of discussion in that forward-looking way. But we have mentioned that the demand forecast that we get from our customers for 2026, 2027, continue to escalate and despite efforts that we are making to increase the amount of supply we can bring online and to a modest extent, in '26, but some more than is not really making that much of a meaningful dent in the gap. There is a lot of demand growth. driven by AI and also across different segments of the market, not just in the data center. And we expect tight conditions for the foreseeable future and certainly beyond 2026 and not giving any further guidance beyond that.

I would just -- the last thing I would add, says that as Sumit mentioned, demand far exceeds supply as Manish mentioned, these fabs talked about for us are coming on basically meaningfully in our fiscal '28. We always have the ability to modulate tool installs. So it's and that would be an option as well.

Your next question comes from the line of Aaron Rakers from Wells Fargo.

Yes. Can you hear me?

Yes.

Perfect for this call. Two questions, one and a quick follow-up. I guess, Mark, maybe just to help model wise, I'm curious if you could maybe frame the current quarter guidance between DRAM and NAND, what maybe assumptions you're making in this tight environment around your ability to ship this will that grow sequentially and then kind of separation between the 2 buckets between DRAM and NAND?

I can provide you. We don't break it out. I can provide you a little bit of color. I mean, let me just -- maybe you heard the 2Q report outs. So both DRAM and NAND pricing was up strongly, NAND, even more than DRAM both also grew volumes sequentially NAND, less than DRAM. In Q3, we would expect price to again be the largest factor but we've given you enough on 26 industry bit shipments and that those industry bit shipments are constrained by supply. And then our supply, we expect to grow in line with the industry if you do all that, you can sort of determine that there's -- yes, you can assume some modest growth -- volume growth in third quarter for both DRAM and NAND.

Okay. That's perfect. And then as a follow-up, maybe more architecturally in this environment. I'm curious, there's a tremendous amount of demand that you guys are seeing in servers. I think you've guided low teens unit growth this year. But one of the architecture things that I'm starting to hear a little bit more about is the idea of CXL and memory pooling and whether or not some of these hyperscalers could drive better efficiency of their memory architecture. I'm just curious at a high level what your guys' thoughts are on that. Obviously, there's a lot of other factors, margins and just content growth in general, but any thoughts on CXL and having an effect on DRAM?

Yes, Aaron, CXL is likely to be experimented with -- at some of our customers, and they are different companies are in different phases of either assessing it or figuring out what they want to do with it, if anything. And we are certainly going to have our memory be available in those CXL configurations. Now my feeling is that the demand is in such a robust place in terms of gap between supply and demand being so significant that any and every available opportunity that can be used and deployed at scale is going to be something that customers will likely try if it is architecturally feasible and can be made to work with their software and their systems. There is a lot of work to be done to bring these solutions to a large deployment scale, and it's not easy and it's not something that can be pervasively used. There are lots of technical limitations as well. But I'm sure those experimentations will continue and some limited deployments will be done in order to test it out and see how it performs.

Your next question comes from the line of Jim Schneider from Goldman Sachs.

Can you hear us?

Do you have a question? Operator, you want to go to the next.

It seems like we may be having some technical difficulties. [Operator Instructions] In the meantime, we will move on to Atif Malik from Citi.

On the call, you mentioned that the non HBM margins are higher than HBM. So my question is really on allocation, how are you guys thinking about allocation? And the reason I'm asking that question is because one of your Korean peers has been reported to be raising pricing by 100%. And I know you had a pricing mid-60 for DRAM and I understand there's always delta across different memory makers. So are you leaving money on the table by not pivoting more towards non-HBM, just your understanding on HBM versus non-HBM allocation?

Yes. I mean we definitely go through different quarters and these relationships between different products can change over time. There is no doubt that HBM pricing was set for a large portion of the shipments in 2026 calendar year late in calendar '25, as is generally the case with HBM where the pricing is determined some time before the start of the calendar year. And that kind of a model is a good model to have. It provides good stability and visibility into the business. The pricing that we negotiate as very robust ROI and profitability, and we feel good about that business. Of course, the upside that we have been able to create from our operations team and supply chain teams on the HPM, we have been able to sell those for even more robust levels of pricing as those upside volumes have materialized. With that said, we don't view a lot of these HBM versus non HBM allocations to be that tactically focused.

These are strategic things that we have to do in terms of providing our customers with match sets of products so they can build AI systems. You cannot ship into AI system that doesn't have enough HBM in it and vice versa. So there is a natural level of balance that is needed for the market to be appropriately having match sets of products to be able to ship the scale. So with that in mind, certainly, the HBM market margins are good, and the non HBM margins, of course, have become even higher. But then this is not just data center issue, the non- HBM margins outside the data center, meaning the DRAM margins outside the data center are also exceptionally robust. So we don't tend to like just jerk around the allocations to different customers in different segments just based on where the pricing is. We have a goal of working with our customers to meet their business needs, and we do that with an intent of helping them meet their business goals as well.

And as my follow-up, NVIDIA talked about could be like 25% of the ultrafast inferencing market. And my understanding is currently these LPU chips have embedded SRAM at one of the Korean makers. And in the future, if these chips move to a foundry, will the SRAM be embedded on that chip? Or can the SRAM be stand-alone and somehow bonded to the processor.

There are definitely on-chip SRAM approaches that are currently in use. And there is -- I've seen some talk about bonded SRAM, et cetera. I would not want to comment on what directions our customers would go in terms of how they work with the SRAM over time. My main focus -- our main focus here is that we look at all of these systems as being very well balanced in the way they are evolving. So these SRAM-based systems complement in small users, the larger systems that are being utilized and deployed at scale like the Vera Rubin, for example and other similar systems that are based on ASIC accelerators. And we see these as continuing to move the ball forward in terms of making the systems more balanced, more efficient. And the DRAM usage in these systems continues to grow over time and has gotten to levels which, of course, we don't have adequate supply for -- but over time, we continue to focus on the growth of these average capacities, the growth of the high-performance steers and even the growth of DRAM in these AI systems, all of which are really big positives for us on the DRAM side of the business.

Your next question comes from the line of Vijay Rakesh from Mizuho.

Just a quick question on the NAND side. I know you mentioned seeing a pretty strong uptick. Just how to size it with the cash demand what's the mix of the data center as you say, in the NAND, if you look out for '26 versus last year, if you could push a way to kind of is that pickup in demand from follow.

When we made our prior forecast of the extent of growth we are likely to see in the data center space, it did include a view that KV Cash would be a meaningful driver. And as that has become more in focus in terms of how the -- how our customers have been wanting to deploy their SSDs. Definitely, it has increased our view of the extent of demand coming from KV cash-related applications. and that has continued to cause our view of the total market opportunity to continue to grow. I'll just remind us that something we have said the last time also, which is that we have also seen a significant uptick in demand for data center SSDs coming from shortages in HDDs. And we continue to see those shortages for the foreseeable future. That has been another driver -- so when we put all of these together, the NAND market is significantly undersupplied to the demand in the data center and that demand continues to escalate in part driven by KB cash, but also driven by just the insatiable appetite that these AI servers have to have fast storage capability available as these systems get deployed more and more. And so the outlook is really strong. And as we have mentioned earlier, our portfolio is incredibly well positioned to continue to gain share in that space. including JV cash apps.

Got it. And just a follow-up on that. As you look at the CapEx, obviously, last couple of years, CapEx on NAND has been lighter -- but as you look at the mix of CapEx for '26, '27 with the CapEx numbers that went up, how would you look at the mix of DRAM to NAND CapEx given both DRAM demand is up, but you're also seeing a spike in case.

Yes, maybe I'll start. So Vijay, the CapEx is still going to be dominated by DRAM and HBM additions yes, this FY '26, we increased our outlook on CapEx to over $25 billion, which was up from the $20 billion we did on the last earnings call. the updated investment reflects this investment in the, which we communicated in February conference and increases in U.S. expansion again, it's DRAM and HBM done, including the increase. Today, we also provided more detail on construction, which we've said in the past was becoming a more material part of the build out, obviously, because we need greenfield capacity. On the December call, we said expect that FY '25, '26 construction would double. And so we expect FY '26 construction to be mid- to high single-digit billions, net when I talk CapEx, we're talking net CapEx. Now as we look out to '27 we did say that we project approximately a $10 billion incremental construction cost and also for equipment spend to increase. Now that '27 spend NAND will begin to increase, but it will still be a much smaller portion of the spend compared to DRAM.

Your next question comes from the line of Karl Ackerman from BNP Paribas.

Can you hear me?

Yes.

This is Sam Feldman on for Karl Ackerman. So we've seen continuous HBM content uplift with each new generation of XPUs with the growing trade ratio of HBM, keeping the DRAM market tight and increasing memory requirements in the form of server DRAM, LPDDR and SRAM Leasing HBM will continue to see such large content uplift with each new generation processors? Or do you expect the content for exit eventually plateau?

Well, we are not going to make long-term projections of where these average capacities will go because those are things that are customers are going to decide as their architectures evolve. What I will say is that if you look at the direction in which AI is trending, and the types of things that are creating value in the AI domain for customers, they go towards more reasoning capability and more longer context windows and the ability to do more with agents and multiagent orchestration. All of these things are really requiring more DRAM capacity and more DRAM bandwidth. And when it comes to delivering on that kind of bandwidth and being able to really optimize the system for all of the different stages of prefill and decode and different aspects of training as well as inference, you really come to the conclusion that these accelerators, whether it's GP user ASICs, do require increasing amount of HBM and increasing amount of DDR5 or LP5 capacity with time.

And that's the trend that we have seen thus far, and you're also seeing in the announced architectures that have been publicized. And so we feel certainly that the trend is -- has been clear and not only has the trend been clear, but that the way our customers, customers meaning the end customers derive value out of the AI system and AI applications is very much connected and consistent with the trend of needing more DRAM capacity and bandwidth, which HBM delivers. So effectively at very good efficient power consumption levels. And so that's the reason we have said in the past that memory is becoming a strategic asset in the AI era is precisely one of the important drivers of that is precisely this trend that you really can't have a high-performing AI architecture or hardware infrastructure without all of those capabilities that DRAM and HBM brings to the table. So that's something we feel pretty good about in terms of where the market is at.

Your next question comes from the line of Chris Caso from Wolfe Research.

I'm wondering if you care to update your view of long-term bit growth for both DRAM and NAND. And I know you mentioned in the call, low 20s in DRAM, 20% for NAND, but those are obviously supply-constrained numbers. And maybe can you speak to the extent to which you think long-term bit growth is increasing as a result of all the things that we've been talking about.

Yes. I mean we haven't provided new long-term bit growth number. I think your -- you have seen that in the past, we have spoken about high teens, mid- to high teens type of ranges for DRAM. And yet you have seen that last year, this year, our forecast has been more robust than those levels. And we continue to feel like we are in an extended space of robust industry demand that obviously due to HBM being part of these numbers, with its trade ratio is just stressing the entire industries and certainly our capabilities to be able to meet those demand numbers. So you're right. I mean, these numbers at least in the foreseeable future are all supply limited numbers rather than the actual level -- true level of demand. So yes, I mean that's sort of the environment we are in. We do expect that next year, again, we will have a fairly robust level of growth in calendar but yes, we are not providing a long-term number beyond that commentary.

Got it. As a follow-up, obviously, there's been a lot of discussion about clean room constraints. And as you folks have pointed out, you're starting greenfield in Singapore, and it's not available until the end of 2028. I guess, you probably can't speak for the industry, but for at least for Micron, at what point do you think you can get caught up with having enough clean room capacity that gives you some headroom for what the customers need. And then obviously, then it's going to take time to move the tools into that. But it goes to the sustainability of of what's going on right now in the extent to which the clean room constraints are going to drive that sustainability?

Well, I think I think, Chris, the first part of the question is really around what Sumit had described in terms of long-term demand, and we're continuing to evaluate all the different demand drivers and signals, including what all the announcements from GTC this week. In terms of the availability of unbased for us and the industry, certainly, we have -- we see this relative constraint for all the major DRAM players to be there through this year and into next year with major meaningful improvement to clean room space availability only out into '28. But of course, as we think through our projects that we've announced, go beyond that into '28, '29, '30, as I think -- as we talked about those projects, the time line for us to get to the point, as you mentioned, will be a function of the demand and we will be nimble to be able to adjust equipment orders and equipment installations in order to stay aligned with whatever the demand is as it played out over that time pricing.

I'll just add to what Manish said, which is that as we look at the demand side of the picture, and we have been engaged with our customers, several of them to be talking about these 5-year multiyear SCA agreement. And in the context of that, of course, we are assessing their longer-term demand, their 5-year demand, for example. And we are assessing those against our own supply capabilities and within that time frame, we are also seeing the emergence and growth of some really exciting new demand vectors, including things like robotics, which we expect to become a very major demand driver. And so when we put all of those things in the equation, we don't have a high confidence view yet as to when the supplier will be able to catch up with demand because the escalation of demand from these various vectors is just very phenomenal. And so to answer the question as to when do we have a high confidence view that the supply will be able to catch up with demand. We don't really have a high confidence view as to when that would happen.

Your next question comes from the line of Srini Pajjuri from Royal Bank of Canada.

Mark, I want to ask the previous question slightly differently on CapEx. You talked about CapEx being for the construction being high single digits and 25%, growing by $10 billion that suggests that roughly next year, half and half between construction and equipment, for the projects that you already announced, do you think 27% is the peak year for construction CapEx if so, what is the normalized, I guess, mix between construction at through the cycles that you kind of think about?

Yes. Srini, we're not going to give any more breakdown that we've provided. It will be lumpy. Just to be clear, the $10 billion was in reference to '26 to '$27. It wasn't quite -- you may have said that, but I just want to make sure it's clear. And we're, of course, going to modulate spend as we see demand and to maintain stable bit share. As Sumit mentioned, it's not clear if we're investing as we can, including this recent acquisition of the fab to put capacity in, but it takes time and a lot of effort, and we can't get meaningful bits until our fiscal '28. So beyond '27, of course, we're going to be very disciplined and could be -- CapEx could go down after '27, but we're not making that call at this point. We're just investing through this next several year period that we can try and get the supply we need for our customers.

Okay. Then maybe a follow-up for the next few quarters, there were questions on gross margins. Obviously, you don't want to comment on pricing, but could you give us an idea how to think about any puts and takes on the cost side and also any mix dynamics that we should be aware of? And also, given the higher CapEx, how should we kind of think about depreciation over the next few quarters and any start-up costs that you anticipate?

Yes. So good questions. maybe tackle first just on costs. We continue to execute really well on cost reductions each can comment some more but as we've talked about FY '26, we have the benefit of the node transitions want am n particular on DRAM and G9 on NAND that are driving our bank growth. So we get a lot of efficiency there and cost downs and the on gamma, for example, replacing 1 alpha capacity. So that's good. And the spend control has been outstanding. So as busy as we are as good as the numbers are, the discipline in the business is very good, including managing by the way, all this geopolitical and the team has really been on top of that, and there's no impact to our operations. So that's good. I think to your question on start-up costs as we bring on ID1 and then there's going to be somewhat toned I made some comments, maybe it was a year, 1.5 years ago talking about start-up costs.

And those comments are still applicable and at the time, I said, I think, between 1 point and 2 points of cost. Now revenue was much lower at the time. So if we dollarize that, it's probably million, $100 million to $200 million per quarter starting in the next quarter or so. And then continuing on through '27 and then it would come down off of that. So again, at these revenue margin levels, it's a much smaller impact 50 basis points or less. And then I think as the -- what was the third question?

Depreciation.

Yes. Depreciation just depends on when production wafers are out and then a useful life of those those assets when they're put in service. And then keep in mind, it's because this is greenfield, this is very long depreciation life. So that's important to keep in mind.

So just a couple of things I'll add, Mark asked you had some comments on cost. I think you covered a lot of them when going really well. Srini both going well, and we had positive comments on those. The only other one I'll add is HBM. I think our HBM3E high has continued to execute well as we've gone through the last -- gotten to high volume over the last couple of quarters, and we actually see HVMI, even though we're in the early stages of the ramp, having an even faster yield ramp than cost structure, we feel very good about both HBM 3 as well as HBM providing improvements for us over previous periods. Then the other comment I'll just have, I think Mark referred to geopolitical managing, I think, the costs relative to geopolitical issues, I think you meant about the Middle East and some of the media reports regarding various input disruptions, but we don't see any supply risks and very, very minimal impact to cost at this time.

Yes. Maybe a just maybe one housekeeping issue since we're covering sort of those issues now. On OpEx, we indicated that our OpEx would would be ticking up on due to R&D costs. So in the fourth quarter, we would expect OpEx to be closer to $1.6 billion, just given that part of it is the extra week, but part of it is just, again, this increase in R&D spend, which we think is completely appropriate to drive the technology, the increased value of memory, Micron's position and then just customer demand on specific projects. And then we would expect in '27 for that OpEx number to be over $1.6 billion, probably kind of a 1.7% run rate number and stabilize from there.

This concludes today's call. Thank you for attending. You may now disconnect.

Ladies and gentlemen, thank you for joining us, and welcome to Micron Technologies Fiscal Second Quarter 2026 Financial Conference Call. After today's prepared remarks, we will host a question-and-answer session.

I will now hand the conference over to Satya Kumar, Investor Relations. Satya, please go ahead.

Thank you, and welcome to Micron Technologies Fiscal Second Quarter 2026 Financial Conference Call. On the call with me today are Sanjay Mehrotra, our Chairman, President and CEO; and Mark Murphy, our CFO. Today's call is being webcast from our Investor Relations site at investors.micron.com including audio and slides. In addition, the press release detailing our quarterly results has been posted on the website along with the prepared remarks for this call.

Today's discussion contains forward-looking statements that are subject to risks and uncertainties. These forward-looking statements include statements regarding our future financial and operating performance, as well as trends and expectations in our business, customers, market, industry products and regulatory and other matters. These statements are based on our current assumptions, and we assume no obligation to update these statements.

Please refer to our most recent financial report on Forms 10-K, Forms 10-Q and our other filings with the SEC for more information on the risks and uncertainties that could cause actual results to differ materially from expectations. 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 can be found on our website.

I'll now turn the call over to Sanjay.

Thank you, Satya. Micron delivered an exceptional fiscal Q2 with stellar records in revenue, gross margin, EPS and free cash flow. Quarterly revenue nearly tripled versus 1 year ago, and revenue for DRAM, NAND, HBM and each business unit reached new highs. Our fiscal Q3 single quarter revenue guidance exceeds the full year revenue for every year in our company's history through fiscal 2024.

For fiscal Q3, we anticipate exceptional records across revenue, gross margin, EPS and free cash flow. Reflecting confidence in the sustained strength of our business, I'm pleased to announce that our Board has approved a 30% increase in our quarterly dividend.

The step-up in our results and the outlook are the outcome of an increase in memory demand driven by AI, structural supply constraints and Micron's strong execution across the board. Our memory and storage solutions are at the heart of this AI revolution. Memory makes AI smarter and more capable, enabling longer context windows, deeper reasoning chains and multi-agent orchestration.

As AI evolves, we expect compute architectures to become more memory intensive. This is why we strongly believe that Micron is one of the biggest beneficiaries and enablers of AI. AI hasn't just increased demand for memory, it has fundamentally recast memory as a defining strategic asset in the AI era. We continue to work with customers on strategic customer agreements, or SCAs, that are different from prior LTAs and have specific commitments over a multiyear time horizon for improved visibility and stability in our business model.

These SCAs also provide customers greater certainty to plan their businesses while reinforcing long-term engagement across our broad product portfolio. We are excited to have signed our first 5-year SCA. We are making excellent progress ramping our industry-leading 1-gamma DRAM and G9 NAND technology nodes. We expect 1-gamma to become the highest volume node in Micron's history. Our 1-gamma node was already the fastest ramp to mature yields is ramping volumes faster than all prior nodes in our history and is on track to become a majority of our DRAM bit mix by mid-calendar 2026.

We plan to increase EUV adoption at the 1-delta DRAM node, utilizing the latest generation EUV tools. These more advanced EUV tools will help us optimize both cleanroom space efficiency and patterning when scaling to 1-delta and beyond. In NAND, our G9 node also remains on track to constitute a majority of bits by mid-calendar 2026. We also achieved a record mix of QLC bits in the quarter.

Looking ahead, we expect co-location of R&D and high-volume manufacturing at our Boise and our Singapore sites to speed up time to market for our leading edge products. We see an unprecedented set of opportunities for memory and storage to enable the AI era across market segments and expect to meaningfully increase our R&D investments in fiscal 2027.

Micron's technology leadership, product excellence and manufacturing execution is being recognized in quality scores from our customers. I am pleased to report that a clear majority of our customers rank Micron #1 in quality.

Turning to our end markets. AI demand is driving DRAM and NAND data center bits TAM to exceed 50% of the industry TAM for the first time in calendar 2026. Traditional server demand is robust, driven by a combination of demand from workloads initiated by agentic AI as well as broad-based server refresh. AI server demand continues to be strong. Both AI and traditional server demand are constrained by lack of adequate DRAM and NAND supply. We expect server units to grow in the low-teens percentage range in calendar 2026, driven by growth in both AI and traditional servers. We expect server DRAM content to continue to grow in calendar 2026 with the introduction of new platforms.

At NVIDIA's GTC, we announced that Micron has begun volume shipment of its HBM4 36 gigabyte 12-high in the first quarter of calendar year 2026 and is designed for the NVIDIA Vera Rubin. With our HBM4 production ramp and volume shipments underway, we expect to reach mature yields faster than HBM3E. We have also sampled our HBM4 16-high product, which provides 48 gigabyte of HBM capacity in each HBM cube, a 33% increase in HBM capacity compared to HBM4 12-high.

Development of HBM4E, our next-generation HBM product, is well underway, and we expect to ramp volume in calendar 2027. Our HBM4E will leverage Micron's production-proven, industry-leading 1-gamma DRAM technology node and is set to deliver another step-function improvement in performance, enabling a whole new generation of AI compute platforms across the industry. Additionally, HBM4E customization options offer us further differentiation opportunities and even deeper R&D engagement with customers.

Micron pioneered the development of LP DRAM for the data center, which consumes one-third the power of DDR DRAM server modules. Building on this leadership, we sampled the industry's first 256 gigabyte LP SOCAMM2 product, which is built using our 1-gamma node and enables a massive 2 terabyte of capacity per CPU, quadrupling the content from just a year ago. We see expanding use of LP DRAM in the data center in the years ahead, and we are excited to maintain an industry-leading innovative product roadmap in this market.

Rapid growth in AI inference is driving the emergence of new architectures optimized for the token economics of specific workloads. Micron's broad portfolio of HBM, LP, DDR and SSD is a critical enabler across these architectures. At GTC, the recent announcement of NVIDIA Groq 3 LPX implements up to 12 terabyte of DDR5 in a rack scale architecture.

We are seeing an acceleration in NAND bit demand in the data center due to AI use cases such as vector database and KV cache offload, and due to growing share of SSDs in capacity storage tiers. Micron's data center SSD product portfolio, enabled by our technology leadership and vertical integration, covers the spectrum from highest performance to highest capacity. We are now in high-volume production of our G9 NAND-based PCIe Gen6 high-performance data center SSDs. Our 122 terabyte high-capacity SSD is seeing strong adoption and delivers 16x the sequential read throughput per watt of a capacity-matched HDD configuration. Our strategy and execution are delivering results. Our data center SSD market share increased for the fourth consecutive calendar year in 2025, to a new record. In fiscal Q2, data center

NAND revenues more than doubled sequentially, reaching a substantial new record, and we expect further growth in the quarter ahead. Micron's data center SSD portfolio is industry-leading, and we have secured a robust set of design wins across our customer base. We are now seeing NAND demand significantly in excess of our available supply for the foreseeable future.

In calendar 2026, a number of factors, including DRAM and NAND supply constraints, could cause PC and smartphone units to decline in the low double digits percentage range. Over time, we expect the value of on-device AI to drive strong memory content growth in PCs and smartphones.

In PCs, there has been exciting innovation recently with agentic AI applications such as OpenClaw, where AI agents can perform tasks independently on the host PC and also initiate workloads in the cloud. PCs with on-device agentic AI capabilities have recommended memory specifications of at least 32 gigabyte, twice as much as the average PC. Additionally, the fast-growing new category of personal AI workstations, such as NVIDIA DGX Spark and AMD Ryzen AI Halo, come in 128 gigabyte configurations, ideal for using large language models on device. Likewise, in smartphones, OEMs have recently announced new flagship devices such as Samsung Galaxy S26 and Google Pixel 10 with agentic AI integrated into their mobile operating systems. The mix of flagship smartphones shipping with 12 gigabyte or more of DRAM increased to nearly 80% in calendar Q4, up from under 20% a year ago.

Micron is well-positioned to accelerate the opportunities in these markets with our industry-leading portfolio of products. In PC, Micron completed qualifications for LPCAMM2 at a major OEM. At CES, we launched the industry's first Gen5 QLC client SSD based on G9 NAND. Micron's LPDDR5X is now designed into leading personal AI workstations, expanding our addressable market, with high volumes shipped to key customers. In smartphones, Micron continues to receive strong interest and feedback from OEM and ecosystem partners on our 1-gamma based LPDDR6 samples. We built momentum with additional qualifications and mass production of our 10.7 gigabit per second 1-gamma LPDDR5X 16 gigabit product.

We saw continued pricing improvement across automotive, industrial and embedded markets. Total AEBU revenue reached a record, with automotive and industrial revenue together exceeding $2 billion in the quarter. In automotive, OEMs are deploying Level 2+ ADAS across their fleets at an accelerating pace. The average car today has less than L2 ADAS capability, containing approximately 16 gigabyte of DRAM, while vehicles with L4 autonomy require over 300 gigabyte. As more advanced ADAS and smart cabin adoption scales, we expect robust long-term growth in automotive memory demand. We have shipped samples of the industry's first automotive grade 1-gamma LPDDR5 DRAM, and in NAND, we were first in the industry with a G9-based UFS 4.1 automotive solution, further reinforcing our technology leadership in this market.

Rapid improvements in AI are supercharging the capabilities of robots. We believe we are on the cusp of a 20-year growth vector in robotics and expect robotics to become one of the largest product categories in the technology world. Humanoid robots will be AI-enabled and will be powered by a compute platform that rivals that of a high-end L4-capable automobile, thus requiring significant memory and storage capacity. We expect this exciting new category of growth to further underpin the long-term favorable dynamics that shape our industry environment. Micron is very well-positioned to leverage this opportunity in close partnership with our customers, enabled by our industry-leading technology, product solutions and operational capabilities.

Now turning to our market outlook. We expect both DRAM and NAND industry bit demand in calendar 2026 to be constrained by supply. We continue to expect supply-demand conditions for both DRAM and NAND to remain tight beyond calendar 2026.

We expect industry DRAM bit shipments in calendar 2026 to grow in the low-20s percentage range, slightly above our prior outlook. In DRAM, cleanroom constraints and long construction lead times, higher HBM trade ratio, higher HBM growth rates and declining bits per wafer growth from node migrations constrain bit supply growth.

We expect industry NAND bit shipments in calendar 2026 to grow approximately 20%. In NAND, some industry suppliers redirecting cleanroom space for DRAM and overall limited cleanroom space constrain bit supply growth. We expect Micron DRAM and NAND supply to grow approximately in line with the industry in calendar 2026.

Micron is working to address the unprecedented gap between supply and demand, and we achieved several important milestones in expanding our global manufacturing footprint this past quarter. In DRAM, earlier this week, we announced the successful closing of the acquisition of the Tongluo site from Powerchip Semiconductor, completing the transaction ahead of schedule. We expect this site to support meaningful product shipments from the existing fab beginning in fiscal 2028. Adding to the existing fab, we plan to begin construction of a similar-sized second cleanroom at this site by the end of fiscal 2026. We continue to expect initial wafer output at our first Idaho fab in mid-calendar 2027, and ground preparation has begun for our second Idaho fab. We broke ground on our first fab at the New York site, and initial ground preparation activities are ahead of plan. In Japan, we are making good progress on ground preparation for our cleanroom expansion to enable future technology transitions in our Hiroshima site.

In NAND, the combination of a higher demand outlook and our decision to co-locate R&D cleanroom in our manufacturing fab underpin our decision to break ground for a new NAND fab at our Singapore site. We expect initial wafer output from this fab in the second half of calendar 2028.

In assembly and test, we commenced commercial shipments from our new facility in India. This state-of-the-art facility will be among the largest single-floor assembly and test cleanrooms in the world. Our Singapore advanced packaging facility for HBM is on track to contribute meaningfully to Micron's HBM supply in calendar year 2027.

We expect fiscal 2026 CapEx to be above $25 billion. From our last earnings call estimate, the majority of the increase is driven by cleanroom facility-related CapEx, of which the largest factor is Tongluo, followed by construction spend increase in our U.S. fab projects. We project our fiscal 2027 CapEx to step up meaningfully to support HBM- and DRAM-related investments. We expect construction-related CapEx to increase by over $10 billion year-over-year in fiscal 2027 as we build out our global manufacturing sites to address long-term demand opportunities. In addition, we expect higher equipment spend year-over-year in fiscal 2027. As we make these investments, we will continue to be responsive to the market environment and our customer demand to appropriately align our supply plans.

I will now turn it over to Mark for our fiscal Q2 financial results and outlook.

Thank you, Sanjay, and good afternoon everyone. Micron delivered strong financial results for the fiscal second quarter, with revenue, gross margin and EPS, all exceeding the high end of our guidance. In fiscal Q2, we generated record free cash flow, reduced our debt and closed the quarter with the highest net cash position in our history.

Total fiscal Q2 revenue was $23.9 billion, up 75% sequentially and up 196% year-over-year, representing our fourth consecutive quarterly revenue record. The $10.2 billion sequential increase is the largest in our history.

Fiscal Q2 DRAM revenue was a record $18.8 billion, up 207% year-over-year, and represented 79% of total revenue. Sequentially, DRAM revenue increased 74%. Bit shipments were up mid-single digits. Prices increased in the mid-60s percentage range, driven by tight industry conditions and included favorable mix.

Fiscal Q2 NAND revenue was a record $5 billion, up 169% year-over-year, and represented 21% of Micron's total revenue. Sequentially, NAND revenue increased 82%. NAND bit shipments increased in the low-single-digit percentage range. Prices increased in the high-70s percentage range driven by tight NAND industry conditions and included favorable mix.

The consolidated gross margin for fiscal Q2 was 75%, up 18 percentage points sequentially. This improvement was driven primarily by higher pricing and also included favorable mix and cost performance. Fiscal Q2 gross margin nearly doubled from a year ago and was a company record.

Now, turning to quarterly financial performance by business unit. Cloud Memory Business Unit revenue was a record $7.7 billion and represented 32% of total company revenue. CMBU revenue was up 47% sequentially, driven by an increase in prices and favorable mix. CMBU gross margins were 74%, higher by 9 percentage points sequentially, driven by higher pricing and cost execution.

Core Data Center Business Unit revenue was a record $5.7 billion and represented 24% of total company revenue. CDBU gross margins were 74%, up 23 percentage points sequentially, driven by higher pricing and favorable mix.

Mobile and Client Business Unit revenue was a record $7.7 billion and represented 32% of total company revenue. MCBU revenue was up 81% sequentially, driven by higher pricing, partially offset by lower bit shipments. MCBU gross margins were 79%, up 25 percentage points sequentially, driven primarily by higher pricing and favorable mix.

Automotive and Embedded Business Unit revenue was a record $2.7 billion and represented 11% of total company revenue. AEBU revenue was up 57% sequentially, driven by higher pricing, partially offset by lower bit shipments. AEBU gross margins were 68%, up 23 percentage points sequentially, driven primarily by higher pricing.

Operating expenses in fiscal Q2 were $1.4 billion, up $87 million quarter-over-quarter. The sequential increase was driven by higher R&D expenses. We generated operating income of $16.5 billion in fiscal Q2, resulting in an operating margin of 69%, up 22 percentage points sequentially and 44 percentage points year-over-year.

Fiscal Q2 taxes were $2.5 billion on an effective tax rate of 15.1%. Non-GAAP diluted earnings per share in fiscal Q2 was $12.20, with 155% sequential growth and 682% growth versus the year-ago quarter.

Turning to cash flow and capital expenditures, in fiscal Q2, operating cash flows were $11.9 billion. Capital expenditures were $5 billion, resulting in free cash flow of $6.9 billion. Fiscal Q2 free cash flow was a quarterly record for the company, exceeding our prior record in fiscal Q1 2026 by 77%. Ending inventory for fiscal Q2 was $8.3 billion, up $62 million sequentially with days of inventory at 123. DRAM inventory days remain especially tight and below 120 days.

We reached record levels of cash and investments of $16.7 billion at quarter-end and had liquidity over $20 billion when including our untapped credit facility. In fiscal Q2, we repurchased $350 million of shares as permitted by the terms of the CHIPS agreement. During the quarter, we also reduced debt by $1.6 billion, including redemption of senior notes maturing in 2029 and 2030. The weighted average maturity on our outstanding debt is August 2034. We closed the quarter with $10.1 billion of debt and a net cash balance of $6.5 billion.

Reinvesting in the profitable growth of our business across R&D, CapEx and other strategic investments remains our top priority for capital allocation. We are committed to maintaining a strong balance sheet, have reduced our total debt by over $5 billion in the last 3 quarters and are at our strongest net cash position ever. Reflecting the sustained strength of our technology leadership and cash generation, as Sanjay mentioned, the Board has approved a 30% increase in our quarterly dividend to $0.15 per share.

Now turning to our guidance. We expect fiscal Q3 revenue to be a record $33.5 billion, plus or minus $750 million; gross margin to be approximately 81%; and operating expenses to be approximately $1.4 billion. Based on a share count of approximately 1.15 billion shares, we expect EPS to be a record $19.15 per share, plus or minus $0.40. We expect higher price, lower cost and favorable mix to all contribute to gross margin expansion in Q3.

As mentioned last quarter, Micron's fiscal Q4 2026 OpEx will also reflect the effect of an additional work week in this 53-week fiscal year. We expect to increase our fiscal 2027 OpEx as we ramp R&D investments in support of an unprecedented set of long-term opportunities in memory and storage. We expect a fiscal Q3 and fiscal year 2026 tax rate of around 15.1%. Micron continues to invest in a disciplined manner across our global footprint. To address customer demand, as mentioned earlier, we now project our capital spending in fiscal 2026 to be above $25 billion. In fiscal Q3, we project CapEx of approximately $7 billion while delivering significantly higher free cash flow on stronger operating cash flow. Due to the need for cleanroom capacity, we expect our construction spend growth rate to outpace equipment spend growth in both fiscal 2026 and fiscal 2027. Any impacts that may occur due to trade or geopolitical developments are not included in our guidance.

I'll now turn it over to Sanjay to close.

Thank you, Mark. Decades of investment in innovation and execution has established Micron as the technology leader in memory and storage and as one of the semiconductor industry's biggest beneficiaries and enablers of AI. As the only U.S.-based manufacturer of advanced memory products, Micron is uniquely positioned to capitalize on the unprecedented opportunities ahead. I want to thank our team members worldwide whose execution made this outstanding quarter possible. As strong as these results are, I am even more excited about what's ahead for Micron.

We will now open for questions.

[Operator Instructions] Your first question comes from the line of Krish Sankar from TD Cowen.

Mark, the 81% gross margin guide is very impressive. Just kind of curious how to think about the sustainability of gross margins, especially as you bring more HBM4 into the mix. If you can give some color on how to think about gross margins in the August quarter and beyond, that'd be very helpful. Then I have a follow-up for Sanjay.

Krish, this is Mark. We provide a strong guide up of 600 basis points sequentially into the third quarter. We're not going to provide the fourth quarter gross margin guidance. However, we have indicated that market conditions we expect to remain tight beyond '26, clearly beyond the fourth quarter. What you're seeing reflected in our gross margin is the benefits of AI driving a multi-year investment cycle, most of which is ahead of us. And AI requires more high performance, more memory and more high-performance memory. And that's reflected in the margins. Also we've talked about supply factors, and those are going to continue beyond '26.

The 81% contemplates a growth in HBM4, but we expect, as I mentioned, market conditions to be strong. Now, keep in mind that as at these gross margin levels, incremental increase in price is going to have less of an effect on gross margin. But, beyond that, we're not providing a fourth quarter gross margin.

Got it. Thanks for that, Mark. And then a quick question for Sanjay on the SCA. Congrats on your first 5-year SCA. How different is it from an LTA? Is this a multi-year volume and price commitment, or does the price get negotiated every year? And also, how to think about cancellation terms on the SCA in case the cycle slows down during the timeframe?

Thank you for recognizing us for the first SCA that we have completed here. And as you noted, SCA is multi-year agreement, and we noted that in our remarks as well. LTAs have tended to be typically 1-year agreement. And of course, in this environment of extremely tight supply outlook in the foreseeable timeframe as well, of course, our customers are very motivated in order for their own planning purposes and for their better predictability to have these structural strategic agreements with us. And of course, these agreements are really meant to bring stability and greater visibility into our business model as well. We have completed 1 SCA, so we are not going to be getting into the specifics here or these agreements. I'm sure you can appreciate that these SCAs are confidential in nature. But of course, these SCAs are meant to achieve the objectives for the customers in terms of their ability to plan and be able to count on supply commitments that are in the agreements, but also for us to be able to count on specific commitments that are there from the customers. And these are meant to go across the periods when the industry is very tight versus other parts of the industry environment as well. So that's why they're long-term agreements, and they have robust terms in them for us as well as for our customers.

Your next question comes from the line of Joseph Moore from Morgan Stanley.

Allocation questions by end market. Obviously, the AI is the area that has the most urgency, but do you worry about demand destruction for things like PCs and smartphones? Are you trying to balance big customer, small customer? Just how are you thinking about that allocation process?

I mean, clearly supply is extremely tight and supply is tight across all end markets. Demand trends are strong across the end markets. While price-sensitive markets such as the consumer examples that you gave, may have some demand that is getting impacted due to the higher prices, but overall demand in those markets as well stay pretty strong. And our goal and strategy always is to be a diversified supplier to our various end markets. I think that is very important for us. Of course data center is becoming a bigger and bigger part of the industry TAM. So of course, bigger portion of the supply goes there, and that's the main driver of growth for the industry as well as for Micron itself.

But other parts of the markets are important to us, such as PC, such as smartphone, automotive, of course, industrial. And we want to maintain that well-diversified mix for our end markets. And I would just like to point out that overall, whether in data center or in the consumer parts of the market, such as smartphone or PCs, the AI trend is continuing to drive greater and greater requirement for memory content. Of course, customers are working in this tight supply environment to manage the mix of their products. But overall, we are very much working with customers across our end markets.

Great. And I think in the past you've sort of said that some customers are getting 70% of what they're asking for. Is that still kind of the ballpark of what you're dealing with? Are there customers higher or lower than that than they were 3 months ago?

Yes. What we have said is in the last earnings call that some of our key customers are -- we are able to fulfill only 50% to two-thirds of their demand in the medium term. And yes, that still remains the case.

Your next question comes from the line of Timothy Arcuri.

Sanjay, I also wanted to ask about the SCAs. Can you give -- I think we're all trying to sort of think on the -- to the other side of the cycle and hope that these SCAs provide some mechanism that will kind of limit your gross margin on the downside to a certain number. So I know you don't want to give too many details, but is there -- is it fair to say that there is a mechanism in these SCAs that would limit your gross margin on the downside when things do finally roll back over?

So certainly not getting into the specifics of these SCAs for the obvious reasons of confidentiality of these agreements, we have done -- successfully completed 1 SCA. We are in discussions with multiple other customers. If and when we complete these agreements and as appropriate, we will of course share further details with you. But what I want to highlight is that these SCAs are multi-year, and they have specific commitments in them. And these are robust agreements and of course these are meant to absolutely give us the visibility and stability toward our business model. Beyond that, really I cannot get into any specifics at this point.

And then Mark, I mean, I just had a question about cash. So you're going to generate, I don't know, $35 billion, $40 billion in free cash flow this fiscal year. You're going to probably have more than $50 billion in cash by the end of the calendar year. So what do you do with this? Are you planning to like set aside a bunch of it to sort of buy back a bunch of stock on the other side? And I guess with respect to that, you do have restrictions on the repo from the money you took from the CHIPS Act. Is there any way to get that reworked?

Yes, Tim, so we're thrilled with the performance of the business and the improvement in the balance sheet, and in the second quarter with record net cash and record free cash flow, beating the previous quarter's record by 77%. Our third quarter guide, when you take those numbers and consider the CapEx we gave, we could see cash flow double, roughly sequentially. Yes, we're going to continue to build on the balance sheet strength and improve our net cash position. We're continuing to delever and pay down debt. Noteworthy that we received 2 credit upgrades in the quarter. So we're now a solid BBB. So we're getting stronger, while as you can see, we've talked about increasing our CapEx investment and increasing our R&D investment.

Now to your specific question on balance sheet priority or capital allocation. Balance sheet is always going to be a priority along with organic investment in our business to advance technology and to put in capacity for value add bits, which we certainly see now. And generating return on capital at this point, over 30%, headed towards 50%. We're going to remain disciplined there though. And then you saw today, we're pleased to announce a dividend increase of 30%, reflects the confidence we have in our business outlook, stability of the business and cash returns in the future. And then as you said, we will have -- we believe significant capacity then for returning cash to shareholders through repurchase, combination of offsetting dilution from stock comp, and then opportunistically repurchasing.

Your next question comes from the line of C.J. Muse from Cantor Fitzgerald.

Wanted to follow up again, SCA question. So you've had an evolution here, LTAs binding, now SCAs. I'm curious if you could kind of discuss the breadth of the different customers that you're speaking with. Is it only hyperscale, or are there others that are interested in? And I know you don't want to go into specific details on the contract. But just to follow up on the last question, is there any CapEx forward requirements tied to of these agreements? Are there pricing tied to an ROIC on those investments? Any help there would be helpful.

So we'll share with you that the SCA that we just have signed is with a large customer. Of course, these agreements are very much focused on allowing us to invest with confidence in our future supply plans, and of course, also just having specific terms that enable us to have overall better visibility into the future demand and as I said earlier, enable stability around the business model as well. And C.J., beyond that, really, we are not commenting on the SCAs other than we'll say that, of course, as I mentioned earlier, these SCA discussions are proceeding with multiple customers, and yes, these are across multiple markets as well.

Very helpful. And then I guess there's a quick follow-up for HBM. I think you guided last quarter 40% growth CAGR, which would suggest roughly $50 billion in revenues for the market this year. Has that number changed? And are you seeing any sort of preference for perhaps moving base to D5 over HBM by any industry players given today the higher margins that we see there?

So, yes, it is correct that the margins for non-HBM today are higher than HBM margins. Demand for HBM of course continues to be strong. We have not updated the numbers that we had provided last in terms of the outlook for the HBM TAM. Of course, the demand for DDR5, LP, and HBM, all of them continue to be strong in the data centers. We, of course, continue to manage the mix of the business as the data center AI demand continues to grow. And as I said earlier, we, of course, outside of data center, we are very much focused on making sure that we are maintaining relevant share in our key other market segments as well.

So we are overall in this environment of strong AI demand trends across data center to the edge, we are very much focused on continuing to manage our portfolio, and we see strong growth opportunity for the full portfolio of Micron's in the data center. And I'll just point out there that portfolio is about HBM, it's about LP, it's about SOCAMM, it's about DDR5, as well as our data center SSDs, which have made tremendous strides in terms of our market share in data center SSDs over the course of last few years.

Your next question comes from the line of Harlan Sur from JPMorgan.

Congratulations on the solid results and strong quarterly execution. Maybe, Sanjay, to carry on from where you left off on your commentary on SSD. In the November quarter of last year, I estimated that your enterprise SSD business was almost half of your total flash business, right? I think it was up, like, 60% sequentially. Obviously, a very favorable mix shift from a margin perspective for the Micron team. And as you mentioned, you remain a top, strong top 3 global supplier of eSSD. Off of that strong number, it looks like your eSSD business doubled sequentially in the February quarter, still 50% of the NAND mix. Looking forward, with the G9 node continuing to ramp, your next generation eSSDs, performance optimized, capacity optimized, mainstream, all on G9. Does this give the team a runway to continue to drive sequential growth in eSSD through the remainder of this calendar year and into next year? And then I just wanted to get your thoughts on this new proposed memory tier of high bandwidth flash. Is this an area where the Micron team might start to focus some R&D resources?

So regarding your question on data center SSDs, of course, this is an area of strong growth ahead. And NAND supply stays strong -- I mean, NAND supply is very tight, and the demand for NAND stays strong. The data center SSDs are a big driver of NAND growth here as well. And Micron is well-positioned with our portfolio of SSDs really going across the requirements in terms of capacity as well as performance across the various customers using TLC as well as QLC, with respect to our data center mix. So very well positioned with this.

And as part of our strategy of continuing to shift our portfolio, our revenue mix toward higher profit pools of the industry and on higher value parts of the market, we, of course, will continue to address opportunities for growing our SSD business. We feel really good about the trajectory that we have been on with data center SSD and the trajectory that is planned ahead for us as well.

And regarding your question on HBF, high bandwidth flash, here. So of course, high bandwidth flash has some positive attributes, such as capacity, but it has the limitations that NAND has such as write speed as well as power and retention. Therefore, there'll be potentially some workloads here, where this may be a possible solution, but it is really early and what is needed, of course, is engagement with the customers here in terms of really understanding the business value proposition of HBM. But we of course continue to study this.

No, I appreciate that. And then how much of these multi-year SCA agreements is due to the sort of inherent requirements for earlier and longer term engagement with your GPU, XPU chip customers just due to the customization of their next generation HBM architectures, right? Especially around the base die, and given the 12- to 18-month design cycle times for these custom base dies, the sharing of IP between you and your chip customers and optimizing the base die to your process flow, right? It does imply that they have to engage with you much earlier in the design phase for their GPUs and XPUs. Is this another factor driving these multi-year SCAs?

Again, we are not really getting into the specifics, not getting into the specific type of customers here as well. But what I will definitely tell you is that, yes, these SCAs really bring us closer to the customer in terms of customer as well as our partnership. And that partnership, of course, extends into bringing us closer in terms of R&D collaboration and roadmap planning, both ours as well as for customers. That's definitely one of the benefits of these SCAs as well.

Your next question comes from the line of Tom O'Malley from Barclays.

Really nice results. So GTC and OFC this week, I think there's a lot of conversations around the LPU architecture and the increased use of SRAM. Could you talk about your view on the memory market longer term as you see more workloads relying on other types of memory outside of the HBM that you're already using? And then just as a broader question, with so much of the demand coming in these longer term agreements being associated with data center and just a few in number of customers that can actually acquire and build these products, how are you benchmarking when you're adding capacity? Do you have an internal forecast for accelerators? Are you talking customer by customer and building bottoms up forecasts? Just so that you know in year 3, year 4, year 5 that you're offering enough supply to the industry and not getting into a situation in which we're at oversupply.

So, first of all, on your question on the SRAM and LPU based architectures, I would just like to point out that, of course, these kind of architectures make the AI infrastructure more efficient. And any architecture that make AI infrastructure more efficient, they're good for all AI. Basically, they help the pie grow faster. Keep in mind that this LPU architecture works in conjunction with Vera Rubin. Vera Rubin, which utilizes tremendous amount of HBM as well as DRAM and the NVIDIA Groq LPX, this LPU-based architecture, actually a rack uses 12 terabyte of DRAM in it as well. So all of this actually is addressing the workloads in a more efficient manner. This helps with, of course, the token economics, the token speed, the scale up of AI across inference, helps with the power and every bit that helps overall is good for further scaling up and acceleration of AI demand as well.

So we look at this as complementing what already exists with respect to HBM and DRAM, and, of course, continuing to grow the pie and accelerate deployment of AI. Just keep in mind that today in the enterprises, the AI deployment as a percentage is still very, very low. There is across all verticals, across all industries, across economies, there's a lot of opportunity ahead. So we are excited about all of these opportunities for our full portfolio of HBM, LP, DRAM, SOCAMM and SSD in addressing these future market requirements.

And ultimately, all of this just points to how strategic of an asset memory is for AI because without more memory, without faster memory, AI just cannot scale up. AI just cannot deliver the capabilities whether it's in training or in an inference. Just look at from last year to this year the DRAM requirement in the advanced AI accelerators has now doubled. Really, and this is what -- I mean, these are some of the factors that are contributing to the supply shortage. And of course, these trends of greater deployment of AI apply on the edge devices, smartphone, and PCs as well. So we are excited about the opportunities ahead and really, absolutely continue to see strong opportunities for our full portfolio ahead.

Your next question comes from the line of Vivek Arya from Bank of America Securities.

Sanjay, on HBM4, do you expect your share to be in this target 20%, 25% range right off the bat? Or do you think you will kind of build towards it over time? Just conceptually, how do you see the puts and takes in terms of whether you can actually expand your HBM share in this upcoming Vera Rubin generation?

We have shared before that in CQ3 of last year, we reached our HBM target, which we have targeted for calendar '26, and to bring our HBM share in line with DRAM share. We had also said that going forward, we are going to manage our HBM as part of the mix of our total portfolio and are not going to break out the share, quarter by quarter here. But what I can tell you is that we feel very good about our HBM product positioning. Feel very good about overall HBM product. Of course, the market is there for both HBM4 as well as HBM3E in calendar 2026, and we will be supplying both of these products and feel good about our overall position here and our ability to fully manage the mix of the business.

For my follow-up, Mark, I wanted to revisit this 81% gross margin guidance. I appreciate you're not giving a specific forward view, but what has happened in kind of prior historical peaks where Micron's margins, I think, peaked in the low 60s, I believe. What is the difference between the prior situations versus now? What have those kind of historical precedents indicated to you about how the trajectory of gross margins can be over the next several quarters? How do customers -- do customers start to react differently when they see these level of gross margins and what is a very, very important input into their AI silicon?

And before Mark answers that question, can I just point out that I accidentally said that we targeted to reach our HBM share in Q3 '26. Of course, I said it wrong by mistake. I meant that we had targeted to get to our HBM share in '25, and we achieved our HBM share in line with our DRAM share in Q3 of '25. Beyond that Q3 '25, we had said we are not going to be providing any further mix of HBM share. So I just wanted to correct what I accidentally said '26 instead of '25.

Vivek, I would say that keep in mind that the industry is supply constrained, so -- and conditions will remain very tight, and that's beyond '26. So that certainly supports the near term, medium term pricing and we've discussed how we're working with customers to allocate best we can to their businesses and work with them on adding capacity, work with them on supply assurance, work with them on new products and so forth. I think your question about reverting to some historical mean, I think maybe that's the thing that should be revisited is, we have a situation where AI is a transformational secular driver.

As Sanjay mentioned, AI requires more and higher performance memory. And this memory helps with the -- driving the token cost down. It helps lower the energy cost per token. It increases the number of tokens. It increases intelligence overall of AI, which drives harder problem sets and agent use, which drives more tokens and needs more memory. So it's become -- the margins are reflecting recognition that memory is a lot more valuable and an efficient way to monetize AI, and that's from data center to the edge. So, and then on top of that, we've been clear for a year or more that there are supply constraints that exist on a number of fronts that will take time. There are low inventory levels. There's declining bits per wafer on node advances, HBM trade ratio, and that's increasing. And then any new capacity really needs to be greenfield, which is a physical constraint, which takes a lot of time.

So these are both durable factors, both the value of memory and the structural challenge of bringing on supply. And we're working that, both those issues. We're investing in capacity, and we're also increasing R&D to continue to advance the technology and improve the value of memory. And we believe these will help with margins over time, and I think customers are recognizing that and entering into these agreements.

This concludes today's call. Thank you for attending. You may now disconnect.

Okay. We're back. So thanks for joining us, everyone, at Wolfe's Auto, Auto Tech and Semiconductor Conference. I'm Chris Caso, Wolfe's semiconductor analyst. So thanks for joining us. We're pleased to have with us this morning, Micron. From Micron, we've got Mark Murphy, the CFO; and Satya Kumar, Head of Investor Relations for Micron.

Gentlemen, thanks for joining us today. Good to have you as usual.

Chris, thank you for having us.

Absolutely. So I guess before we'll start with Q&A, there's a lot to talk about because we're really an extraordinary time in the memory market right now. But why don't I pass to you, Mark, if there's any sort of opening remarks you'd like to make kind of set the table for us here.

Sure. So I'll start 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 filings, including our recent 10-Q and the 10-K at the end of last year. So our business is on an extraordinary trajectory. Since our last earnings call, our business has -- our financial outlook has strengthened further. Demand is significantly higher than our ability to supply and the industry's ability to supply. We continue to expect supply/demand to be tight beyond '26. We are doing everything we can to plan and invest appropriately for our customers' needs over time. We're making good progress on multiyear agreements with specific commitments. We are preparing clean room space that will come online over time, and that will give us the ability to grow our bit supply in line with market demand.

During the quarter, we announced 2 new groundbreakings, one in New York for DRAM, one in Singapore for NAND. The new NAND fab will be -- have first wafer outs in second half '28. We also announced the Tongluo site acquisition in Taiwan, which we expect to close in the calendar second quarter. And that Tongluo fab will be used to support DRAM production. Micron is executing extremely well across technology, products and manufacturing.

And let me, at this time, address some recent inaccurate reporting by some on our HBM4 position. We have been in high-volume production on HBM4. We've commenced customer shipments of HBM4, and we see shipment volumes ramping successfully this calendar Q1. This is a quarter earlier than we mentioned during our December earnings call. Our HBM capacity is ramping well, and we have sold out our calendar year '26 HBM supply as we highlighted a few months ago. Our HBM yield is on track. Our HBM4 yield is on track. Our HBM4 product delivers over 11 gigabits per second speeds, and we're highly confident in our HBM4 product performance and quality and reliability.

You've given us lots to talk about. So that makes my job a lot easier. So I appreciate that.

So maybe we could start with the comment about the business has strengthened further since the start of the quarter. And really, the guidance kind of coming into this quarter was quite extraordinary. I think it's up 37% sequentially with about 11% in gross margins. So I guess maybe the first question that while demand has strengthened, we know you're supply constrained. And so how should we think about that with respect to fundamentals for both the quarter and for the out quarter given the supply constraints?

So we have a very positive set up for the business. And Micron is in the best competitive position it's ever been in at a perfect time. And it's positive in both the demand setting and it's positive on the supply backdrop. Of course, on the demand side, we have this AI-driven demand. And as AI works to -- as companies work to improve the intelligence of AI systems, these are requiring more and better memory. As the models get larger, as the context windows get longer, as the reasoning becomes more intense, these are all driving higher performance memory and storage.

And so we're seeing this proliferation in the amount of tokens generated, the demands on the processor, the rearchitecture of memory systems in order to support that. And we're seeing high-performance memory being added throughout the architecture. That demand has been very strong. We saw that happen through '25, where server demand increased through the year from what was going to be in the single digits to mid-teens. And we also saw related a traditional server growth increase through last year as workloads begin to affect even traditional server. So we see growth, that strong growth continuing.

And then I think just the latest CapEx investments that you've seen from the hyperscalers, I think for '26 now, expecting close to $800 billion when just a few years ago, it was under $200 billion. So we have a lot of very positive things, and we can talk through the fireside here on the different memory products and the performance attributes of those, but the demand picture is very strong. And that is -- it's been data center driven, but we are seeing it proliferate to the edge. And we've seen the content increase there, but eventually with more autonomous-related activities, more capabilities in smartphone and PC, we're going to see, we believe, that will further strengthen demand.

The supply side is very important as well in contributing to this positive financial performance. We worked down inventories through '25. We have -- we just simply do not have enough supply to meet demand and by a substantial margin, Sanjay said on the earnings call that some key customers were only able to meet 50% to 2/3 of their demand. So that gives you a sense that it's large customers, it's smaller customers, it's broad-based across markets that we are unable to provide supply. So we are actively working that around the clock, whatever we can do to bring on supply. We're certainly sweating the assets best we can to incremental bits.

Our inventories are lean, especially in DRAM. So incremental bits are coming through the ramp of our 1-gamma node. And that will provide us the supply in calendar '26. We have incremental greenfield capacity coming online in mid-'27 with Idaho 1. This Tongluo acquisition, we will be able to provide some supply near the end of '27, going into '28. And then we have Idaho 2, which will be beyond that. And so the point is that supply is coming on, but it takes a long time to bring it on because it's greenfield.

Traditional node transition is still the most efficient way for us to bring on supply, but those nodes are less efficient than it used to be. And there's simply not enough ability to produce enough supply out of node transition.

Adding to the supply challenge, as we know, is the increase in high-bandwidth memory, which is more silicon intensive. And that trade ratio, which we talked about 3:1 historically for HBM3, that is increasing over time as we go to 4 and 4E and 5, that trade ratio simply expands. So that further pressures supply.

So with that backdrop, we see continued strong demand. We see supply-demand tightness beyond '26. We're committed to our investment plans to invest in a disciplined way to supply customers what they need over the long term, and we are working feverishly to get that in place. But it's resulting in a favorable price environment, which, as I mentioned, our -- since our earnings, our financial outlook has strengthened further, and that's been driven by price.

We're not quantifying that at this time, however.

We are not.

The other thing you mentioned in your remarks also was progress on supply agreements with the customers. So maybe you could expand on that a little bit. And again, it's extraordinary times in an industry that historically has negotiated a quarter in advance. So maybe the question is, what is that progress? And what are your customers asking you? And how is that different from before?

Yes. So we've mentioned for a few quarters now that customers have approached us for these multiyear agreements with specific commitments. And I think it reflects a natural progression of the business that memory and storage have become more important to AI systems and more important broadly to PC devices and smartphones, and we expect this trend to only continue.

And the value we bring is real. I mean, if you look at HBM3E, for example, our product was 30% lower power than the competition. And HBM itself is a very efficient way to -- power-efficient way to deliver data to the processor. Just on generations of HBM3E to HBM4, we've more than doubled the terabits per second bandwidth in just that generation. So that provides a lot of value.

Furthermore, on another example of value, and we'll talk more about LPDRAM, but we introduced LPDRAM to server configurations to be able to move some memory functionality to LPDRAM and LPDRAM offers 60% lower power than traditional DDR. So again, we provide innovation and great value to the customers. And on the NAND side, there's on a read-to-watt basis, we have 15x better performance than HDD. So there's an opportunity that we are bringing value to customer systems. And I think our technology position, our leadership in technology, our leadership in products enables us to -- customers appreciate that, and we have deep engineering relationship.

So you combine that added or increased value of memory and storage with this pressing concern of supply assurance, then customers want to make sure that they have supply and that they have supply over the next 3 to 5 and longer periods of time. So we have -- so we're working with customers. It's a positive for them because they work with a technology leader. They gain supply assurance. They know that they can work with us and get reliable supply and high-quality supply. We're best in market there.

And maybe lastly, I'll mention that we now have U.S. supply that will be coming online. That has become of interest to customers. So when you add all that up, we think it's a win-win. Customers get the best technology, the best products, but we believe we're the best operator in the business, and they get supply assurance. And then we also -- we get a firmer commitment and a clarity as we invest the large amounts of capital and R&D required to support their needs and advance the industry.

And how far out are customers willing to give you visibility now? I mean, is it a year or is it more than a year? And how firm are the commitments? Because these LTAs have been in place for a while, but there were fewer commitments -- fewer hard commitments in those LTAs, right?

Yes. I mean they're longer than the LTAs we've done, but we've not provided specifics and negotiations are underway. We'll provide those specifics at a later date. I would say that the time horizon of the customers as it relates to memory and storage has extended. And it's extended because the product road maps are hugely important for them and us, and these are multiyear efforts requiring very deep customer engagement, engineering collaboration, codesign efforts. And that time horizon on just the memory product side has extended versus what you would maybe think it was -- what it was 10, 20 years ago. And so that's a positive and provides visibility.

Furthermore, I would say even our customers, certainly as it relates to capital investment, their time horizon has extended, right? They're investing more CapEx in their own businesses than they ever have. And that -- you have a certain time dimension, you think when you're putting assets on the ground and that time dimension is longer than many of them typically have operated as far as physical assets. And so that has -- they have to marry that -- those capacity plans up with our ability to supply product over time as they fill those facilities and light their technology. So I think there is a positive things that both support longer-term agreements, and we bring value in a way that customers are wanting tighter arrangements.

Okay. Maybe moving on, and I think your opening comments were pretty clear about what you felt with regard to your technology on HBM4. So really no changes versus your past comments. And I guess I infer from that, that your view in the past was that your share of HBM this year would be about equal to where your commodity DRAM share is. And I presume that's still your view.

We've not given an update. We -- as you mentioned, we said that when we launched initially in HBM for our high-volume products, we said that we would ramp our HBM3 high to the above high into getting HBM share in line with our conventional DRAM share. And we achieved that. And now it's a product like others that we are always making decisions on -- portfolio-based decisions on what our customers need, what our footprint will support, where the value is in the space.

I mean, fortunately, with our technology and product portfolio, we have the flexibility that we can be agile in directing our bits to places we find value. But we're thrilled with our HBM performance. We believe best-performing product in the business, very deep customer engagement, a very rapid cycle time for each generation that we think makes being a technology leader all the more important.

And I would encourage you, we first started -- we mentioned HBM4 sampling back in the September call. And then we had Scott DeBoer at a November conference, he provided more detail on HBM4, including the technology aspects, the metallization characteristics, the design attributes, the CMOS that we do in-house. And he talked about all these things working to be first to market in this over 11 gigabits per second and -- or first to achieve. And we talked further about our confidence in the December call.

So again, as we mentioned today, we're in high-volume manufacturing of HBM4. We're -- commenced shipments to customers are ramping in calendar Q1. Again, that's an improvement versus what we said in the December call, which was calendar Q2. So we're thrilled with where we are in the future road map for that product is very strong.

Understood. Maybe I'll pivot to some of the financials as well. And obviously, gross margins have been very strong, both as a result of HBM, AI and pricing that's going on in the broader market. How do we think about margin progression from here from these levels? And I think one of the things you mentioned in the call was that pricing improvement at these margin levels doesn't have the same incremental benefit. So maybe perhaps level set us with where our margin expectations should be given this environment.

We guided to 68%. So -- and we've indicated today that our financial outlook has improved since our earnings. And we did talk about how just the math and that as you increase price at this level, you get less of a margin effect than at lower margin level. We do believe that margins will expand from our second quarter here in the third quarter. And we believe that the market environment, as we mentioned, we expect the market to remain tight beyond '26. And so that clearly supports the pricing environment.

And then the -- of course, we're ramping our 1-gamma node for bit supply. And so our cost performance has been good. And we're operating really well. I mean we didn't really talk about it much on the call, but our spend control is very good. And so our cost performance between that and the -- just the volumes absorption has been good. And then on mix, that's also been a positive story for us. We have the flexibility in our premium technology and our portfolio to have mix be a contributor over time. And so we're working our best to optimize the business.

Okay. Maybe I'll pivot over to NAND. And it's a smaller part of your business, but maybe you could talk about how your outlook for NAND may have changed since this time last year. Obviously, pricing is up a lot, but where does that fit into Micron's strategy? And is this a situation where you're leaning into NAND a little bit more? Is it -- tell us where you are?

NAND is an important business for us, always has been. And -- but as you know, last year, sort of from late '22 on, the industry had ample supply of NAND. And we had brought utilization down in the fabs or in the fab. We had slowed node migrations and in an effort to get supply-demand balance. We concurrently had over the years, including through the downturn, really worked to improve and sharpen our portfolio. We brought controller capability in-house. We extended our leadership on technology, 232-layer at the time, process technology. And we just work to build the best business we could in that environment. And today, we enjoy really a preeminent position in the NAND market, where we have the best portfolio of products serving the most attractive part of the market, which is the data center SSD market.

And we have ranging from our performance SSDs where we do TLC to our capacity drives where we have world-class QLC. We've gained share in those markets. And then the last quarter, we mentioned that we had achieved over $1 billion run rate to get that business.

Again, it's maybe a similar story as broader Micron. Micron is in the best position it's ever been in, in its history at a perfect time. this AI revolution has afforded us the opportunity to serve very important customers, driving very important technology forward and the financial result for us has been positive.

Now likewise, on the NAND business, while it was not performing financially as we would have wanted due to supply-demand balance in the market, that has tightened up pretty quickly. And as you -- it became particularly clear during CES where the processor loads are so high in AI systems that there are very active efforts to reconfigure the memory and storage to allow that processor to be most efficient. And so you're seeing a tiering of memory and storage, that storage has emerged as an important part of the architect you have the hottest data as an HBM serving the GPU or ASIC very aggressively. You can offload some of that through LPDRAM, which we innovated and bringing that into servers, and there's an efficiency there. And that's warm data.

And then you've got cooler data, which now if it doesn't need to be used immediately or with a latency that is acceptable, it can reside in a drive. And so that's occurring, and it's a function of this as the context windows get longer as the reasoning becomes more intensive, this offload to KV cache has become more important is bigger, and that is helping drive storage.

So as a result, we do see the need for additional greenfield space on NAND. And hence, we announced the fab in Singapore. And that will be '28, we said first wafer out in that second half '28. So it's a ways out. But until then, we will supply the market as needed through node transitions. And in this case, it's a G9 node transition. But we're in a great position on NAND. And maybe, Satya, if you want to talk about LPDRAM specifically since we're talking about the data.

Chris, I know the question was on NAND, but Mark touched on an important point, the hot, warm, cold, KV cache tiering. And we actually published a paper -- white paper last week on LPDDR, which sits in the warm tier. So if you search LPDDR Micron, you'll find that white paper. And what it talks about is the importance of content increase in that tier. So it talks about how you're seeing -- you're going to see systems that will be coming out this year, which will have triple the amount of LPDDR compared to what we had last year. And how adding that additional DDR content -- LPDDR content cuts the time to first token by 98% and improves inference significantly. So do take a look at that, good information there.

Yes. We can include a link to that in the note that we write up after this discussion. So we've got about 6 minutes left. We could take a question from the audience if there are any, not to put anybody in pressure. I'm happy to keep going. But nobody is raising their hand, that's the usual response.

Maybe I could follow on with one of the most frequent investor questions here is the sustainability of what we're seeing right now. And we've all been through a number of cycles. Boy, this one sure feels different than the other cycles. But what's giving you confidence in the sustainability of what we're seeing and for how long?

So we talked about it quite a bit at the beginning of the chat here. What gives us confidence is that both the demand and the supply factors are very positive, and they are improving actually. So on the demand side, I don't want to repeat everything earlier, but it's clear that as this AI technology advances and the token generation is increasing to produce better results to become more intelligent. That is demanding an efficient way to do that, an important way to do that is through more memory and storage, more as in more of it and higher performance memory and storage.

That is becoming clear, and that's from the data center to the edge and the edge will become increasingly important here as you think about robotics and autonomous activities. It's been largely training and then focused on training and reasoning and inferencing. And what will be the next, I think this year, there'll be a lot of talk of agentic activity. So where the enterprise systems are being used by agents, and that's going to increase the activity more. So again, this all points to a strong demand picture that is broad-based that plays to our strength on technology and product leadership and manufacturing excellence. So demand side is very good.

On the supply side, there is -- we are doing everything we can to add supply to meet our customers' needs. That is very important to us that we serve our customers the best way we, technology, products and supply. So we are doing that, but there is no easy or fast way to get that done. We're working very hard toiling to get incremental bits out on our existing capacity.

The node transitions, which are less efficient than they used to be, the node transitions we are working this year will provide our supply, as we talked about on 1-gamma. 1-gamma will be the majority of our bits in the second half of our calendar '26. HBM is soaking up more of that supply. So it's challenging that. So it's got to be greenfield that's added. And we've given you the road map of our greenfield plans.

I did want to mention, I forgot to mention earlier, the -- so we have ID1, which will produce bits in second half or mid-'27 at volume. And then this Tongluo acquisition that we just completed in Taiwan and expect to close in the calendar second quarter. And it was in the press release, that was $1.8 billion. So that $1.8 billion and the associated tool install and other things, that will be added CapEx for this year. So we will have -- we guided $20 billion on the last call for CapEx in fiscal '26 and this Tongluo additional investment will add to that number.

But again, that is capacity that's going to produce later in '27 and then our other expansion plans are well beyond that. And so there are very positive demand factors. There are very positive supply factors that I think allow the look that the industry will be in good condition.

Now we always are investing very carefully. We're reassessing the market continuously. We're looking at what competitive additions there are in the market. And to the extent there's any change in our view on demand, be it positive or negative, we will adjust. And with this -- with the increasingly greenfield capacity versus just node transitions, I think that allows some standoff to modulate builds to make sure you're meeting the long-term demand in the most responsible way.

So in the near term, working very aggressively to get supply to our customers, doing everything we can, working with them on longer-term agreements, which helps us get line of sight to longer term and then always investing in a very disciplined way in our business and having the best technology and products in the space.

Okay. It looks like that's the time we have. So we'll wrap it there. But Mark, Satya, thanks for joining us today.

Thank you, Chris.

Thank you, Chris.

Thank you for standing by, and welcome to Micron's Post-Earnings Analyst Call. [Operator Instructions]

I would now like to hand the call over to Satya Kumar, Investor Relations. Please go ahead.

Thank you, and welcome to Micron Technology's Fiscal First Quarter 2026 Post-Earnings Analyst Call. On the call with me today are Sumit Sadana, Micron's Chief Business Officer; Manish Bhatia, EVP of Global Operations; and Mark Murphy, our CFO.

As a reminder, the matters we're discussing today include forward-looking statements regarding market demand and supply, market trends and drivers, and our expected results and guidance and other matters. 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 documents we have filed with the SEC, including our most recent Form 10-Q and our upcoming Form 10-Qs for a discussion of risks that may affect our results. Although we believe that the expectations reflected in the forward-looking statements are reasonable, we cannot guarantee future results, levels of activity, performance and achievements. We are under no duty to update any of the forward-looking statements to confirm these statements to actual results.

We can now open the call up for Q&A.

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

I guess my first question is, Mark, you talked a lot about being constrained and unable to meet some of the demand out there. I know in the quarter you talked about kind of a slight increase in bit shipments of DRAM this quarter. I guess as part of that, did the non-HBM-DRAM bits grow sequentially in this quarter? And how is your guide factored in between bit shipments between NAND and flash in this current quarter?

So I think, Aaron, one thing I wanted to mention was we are doing all we can to increase bit supply now. And we were able to provide additional bits to do 20% bit shipment growth in our fiscal '26. And we're doing that through efforts within an existing footprint and just efficiencies in the fabs. We're doing that with node transitions as we've said, 160 -- Aaron, I'm sorry, 1-gamma and G9 to get additional bits for DRAM and NAND. And we're pulling in what we can on construction that will help us largely in '27, but we are able to provide growth what we think will be in line with the market in fiscal '26 -- or calendar '26.

I think we did have a very modest bit shipment growth sequentially in the first quarter. We expect a little bit more in the second quarter, but we -- again, a quarter primarily driven by price as it relates to revenue growth.

Yes. That's helpful, Mark. And then as a real quick follow-up. As we look forward, I know you mentioned the 14-week period for fiscal 4Q, but how would you have us think about the OpEx trajectory from here over the next couple of quarters?

Yes. I mean we've guided the second quarter. It will be sort of flattish in the third quarter relative to the second, that will be up in the fourth quarter. And up and plus the additional week.

Our next question comes from the line of Vijay Rakesh of Mizuho.

Great quarter. So just a quick question, Sumit, Mark and Manish. On the -- when you look at the -- DRAM pricing is up pretty nicely, 20% sequentially and NAND was up pretty nicely, too. I was just wondering when you look at the demand side, how do you decide to allocate capacity versus conventional DRAM versus HBM because it looks like the conventional DRAM profitability is improving very significantly. So just wondering how that reasoning -- how you look at that as you look through '26, '27? I have a follow-up.

Yes. So good question. I mean, we are in a very interesting environment where the aggregate demand for both DRAM and NAND is substantially higher than the ability to supply to it, not just from a Micron perspective, but even at an aggregate industry level. So we are not really able to meet the demand for customers across any segment. So all segments are short in terms of what they need from us versus what we are able to supply. And we are also short on the HBM side, non-HBM side, all parts of the market. There is a mismatch between supply and demand, and that continues for the foreseeable future because the -- over the course of the last few months, there has been a very significant and pervasive step-up in demand across the data center customer base. And so it has created a set of challenges in being able to meet the demand from our customers on a very broad level.

So in terms of allocation, I can't get into too much specifics. But suffice to say that we are working very hard to find the threshold level of supply for all of our customers to minimize the impact to their businesses due to lack of adequate supply. And we are working hard to ensure adequate level of diversification in our business across segments, across customers, and focus on our strategic customers when it comes to doing some of these longer-term strategic deals that can improve foundationally our business model. So those are some of the things that we are attempting to do. And in the course of all of this, our mix over time will tilt more towards the data center, but we're also trying to remain focused on having diversification across all the other segments of the market as well.

Got it. And then on the eSSD side, you're seeing a very strong pickup with your QLC, Gen 9 SSDs as you mentioned, 122 terabyte, 245 terabyte. Is there a way to kind of look at what is the attach rate that you're seeing now with the SSDs on AI servers versus last year, especially as some of these inferencing vectors with reasoning and answer and all that starts to pick up? How do you see that attach rate growing year-on-year?

The attach rate is certainly growing, and these AI servers do need more high-capacity SSDs. They also need a significant amount of high-performance SSDs. This is where our Gen 6 SSDs have come into the picture. So we have been the first company to supply qualified Gen 6 SSDs to the market. So that is gaining rapid market traction and contributing to our improved share position in the data center SSD space. And then these high-capacity SSDs that you mentioned, Micron has the leadership position in QLC workloads. And so our overall QLC bit mix is higher than the rest of the industry. So that's an additional capability that we have.

And if you look at workloads like KV cache, they do require a lot of higher capacity SSDs. Now one additional thing that has happened is that beyond the underlying growth trend of AI servers using more of these SSDs, we've also had -- our customers not have adequate amount of HDDs as well. And because of that, there has been a lot of demand coming towards SSDs and consequently, the ability to supply NAND to our customers has also been oversubscribed due to a combination of all of these trends.

I do want to note that on SSDs that if you remember, our business was over $1 billion in the first quarter of '25, and in the first quarter '26, the business did clear that. And as Sumit mentioned, as sort of the industry is able to get on top of the supply chain-related issues we expect. And given the demand in that business, we expect that growth to accelerate through the year.

Our next question comes from the line of Vivek Arya of Bank of America Securities.

I actually had 2 questions. First, back on the gross margin, I think you are at 68%. Is there reasonable way to think about what can be the path forward? Like is it 70%? Is it 75%? I know you have not given a specific number. But Mark, is there a way to logically think about where your margins can go in the cycle because this is well above the prior peak? So is there a certain business model you have in mind? And the reason I ask the question is because you have close to 70% gross margin, you're right, your customers, NVIDIA, AMD are also talking about very high gross margins or expanding gross margins for their business. So I'm just curious to know how you think conceptually about the path forward for your gross margins.

Yes. They're -- we're at record-setting levels and that's positive. We guided to that. And there are a lot of favorable factors in that, that are driving that. First is the demand driven by this generational change in tech with AI. And that's a multiyear build-out done by well-capitalized companies. So we think that's sustainable. And fortunately, AI needs more and better memory to perform better. So the underlying secular driver for memory is positive.

We're just in a fantastic position with leadership technology, best products in memory and storage, and then we're operating really well, yielding quickly, cost performance is good, capital discipline is good. And then you've got these general structural supply constraints. So when combined with the demand from AI, data center to the edge actually. We've got this supply shortfall and you see the price movement associated with that.

And this is a -- not in supply gap, that's not easy to address in the near term given we can do with node transition, but you've got greenfield capacity that needs to be put in place. You've got increasing HBM and the capital intensity associated with that or silicon intensity associated with that. And these are putting on earth, the most -- some of the largest and most complex factories on the planet. So we need partnership to do that, and we're going to do it carefully.

So with all that, we do expect that margin can go up from second quarter. It will go up on the basis of, we believe market conditions that should remain positive beyond '26. We believe it's possible based on our ability to deploy bits to premium products in which we're best positioned. And we believe that our cost performance is such that it will help us sustain or expand these margins.

Now just the math of being at these levels, for equivalent increase in price, we're going to get less gross margin expansion. So there can be -- as we mentioned on the call, on the primary earnings call, there's -- the increase from here would be more gradual than you've seen in fourth to first or first to second quarter guide, but we do believe that margins can go up.

And for my follow-up, every time the customer, right, whether it's NVIDIA or Broadcom or AMD, every time they report, they seem to be upsiding numbers for next year. So when you say you're sold out for HBM, does it mean that you don't have an ability to upside your production? Or does it mean that you're sold out based on the current level of forecast and you have kind of unit and ASP certainty around that? I'm just trying to understand what scenarios let them continue to upside the numbers, unless it's just a share shift between you and your peers. Like if they continue to upside the numbers, can you help them do that? How much more flex is there in your production for HBM?

Yes. So I think the way to read what we are saying is that we have visibility to our supply. And of course, we're always trying to improve that. But we have a good understanding of what our supply is. The current situation, when we say sold out for HBM, for example, the HBM supply for 2026, we have reached agreements on volume and price with our customers. Of course, if there are upsides to supply, we are able to -- very quickly be able to place that upside supply at our customers at good terms. But when it comes to the overall aggregate business, the demand on us is so much higher than the supply, that even small increases in supply are not going to be able to make a dent in that demand. So we are more than sold out. I mean, we have a significant amount of unmet demand in our models. And this is just consistent with an environment where the demand is substantially higher than supply for the foreseeable future.

And Vivek, I'll just add on HBM and DRAM because the whole market is short. And HBM production in fab output is shared with DRAM production, right, all in our 1-beta node. So this constraint is kind of across the entirety of our market, as Sumit mentioned, and as we said before, which is incredibly constrained. So we are working hard to try to increase supply for DRAM, HBM and NAND. Yields are -- yields have come up really well for both our 1-gamma and our Gen 9, and those are going to be, as we gave color, the majority of the bit growth for next year is coming from those transitions. And we're also going to have the majority of those -- of our DRAM bits on 1-gamma in the second half of next year. And Gen 9, as we go through the year, will become the highest volume node for us in NAND.

So technology transitions are moving aggressively. We're optimizing and looking to maximize production where we can. And I just want to point out that we've also said that we expect HBM4 yield ramp to be faster than our HBM3E 12-high yield ramp was, which was already faster than our HBM3E 8-high yield ramps. So we're gaining confidence in our HBM capability at scale. So all those are things that we're continuing to work on, but just the demand environment is such that it's really way oversupply, whatever we can or what the rest of the industry can supply.

Comes from the line of Kevin Cassidy of Rosenblatt Securities.

My question, along those lines, if you're saying the HBM4 yield is better than HBM3. What -- how does this work with the process node with the qualification time taking so much longer with HBM? Does this change your cadence for process technology? Do they stay in production longer?

I'm not quite sure of that. I think both our HBM3E and HBM4 are on our 1-beta process technology. They both use the similar assembly architecture and process that we've had now from HBM3E 8-high, 12-high and now at HBM4. So the reason I said that we expect our HBM4 yields to ramp faster than HBM3E 12-high did. Once we go to production in the early part of calendar year '26 and start shipments in calendar second quarter is because of being able to leverage the learnings that we have, both on 1-beta for HBM as well as on the packaging and test side because we have common architectures between those products.

My question was more on a bigger picture. Will 1-gamma get qualified for HBM4? Or will it always be 1-beta?

HBM4 is on 1-beta, and we haven't given specific guidance on how future generations will be, but HBM4 will be on 1-gamma. And the cadence of platform -- I'm sorry, 1-beta. And the cadence of our customers' platforms continues to be rapid every 12 months. So we're just going to be focused on delivering new memory -- new high bandwidth memory solutions for our customers on an approximately 12-month cadence. So that's not slowing down.

Okay. Maybe another question. Just how quickly can you change your wafer allocation from HBM to low power to high performance, I guess how far into the process technology -- process can you change what the final product is?

So they are shared process between the 2 -- between the various different designs that we have on a given DRAM technology node, but the design itself is unique. So when you start one design, whether it's a low power design, a DDR design or an HBM design, obviously, that's going to be the time frame. So the real -- the answer to your question is that, basically, it's one process cycle time that we can change between any of these different products.

I think, Kevin, I just want to make sure it's clear that we have extensive planning that we do around all the products and the product road maps and what products go on what node at what time. And there are a lot of factors that go into that. Cost is not the only factor. And in fact, in HBM, it's such a complex product that's stability and turning quickly with the customer is an important factor. So there's some value in 1-beta and as stable and strong a node as that is. And then 1-gamma is available at the right time that we can use that node. Otherwise, we have plenty of other bit requirements that 1-gamma, we're ramping that as quickly as we can to get bit supply.

I mean, generally speaking, we tell our customers that it takes us 5 months if they want to make changes, that use different die in the fab, 3 months, 3.5 months to get to the front end and 6 to 8 weeks depending on the complexity of the product on the back end. So roughly speaking, 5-plus months of lead time. And customers understand that, they know that these things take a long time.

Our next question comes from the line of Steven Fox of Fox Advisors LLC.

I just had one question. On the CapEx spending that you talked about on the call. Are you giving any sort of sense for the breakdown between facilities, equipment, construction, all of that, say, from -- and how it might change between '26 and '27, that mix?

Yes. We've not indicated anything about '27 other than that it will be -- '27 total CapEx will be up versus '26. I did on the call, I believe, indicate that '25 to '26, the plans to -- our fiscal plans to roughly double the construction CapEx. And so that gives you sort of an indication on direction of travel. Now we've got -- '27 will have Idaho 1 finishing, we've got Idaho 2. We've got Japan, we've got a lot of construction sites. So those are going to contribute to that sequential '26 to '27 CapEx increase. And Manish, on HBM and Singapore...

We do have a considerable amount of construction, I was going to say, Singapore as well, where we've got that, and we're going to be equipping our India assembly site, which is just now going from pilot production to production ramp in the beginning of calendar '26. So yes, there's going to be a mix of growth in terms of construction CapEx to fund the mid-term and long term equipment CapEx into existing clean room that we have right now. And then, of course, assembly investments for both HBM as well as conventional packaging that we'll be making as well.

Yes. I'll just maybe add to this point. Tim asked about capital intensity on the earlier call. We had CapEx as a percent of sales in the first quarter was below 35% that we talked about. And then on the second quarter, basically, where our guide is to the -- both on the revenue and the CapEx guidance we gave you, you could tell it's in the mid-20s on CapEx as a percent of sales.

And we've talked about strengthening free cash flow. So we're not only increasing investment in order to get the bit supply, we are generating a lot more cash. We had near 30% free cash flow margin in the first quarter, and we talked about free cash flow increasing through the year. We paid down debt, $2.7 billion of debt in the first quarter, bought back $300 million of shares and went to net cash, all in the quarter, and it was a record setting free cash flow in the first quarter.

Our next question comes from the line of Jim Schneider of Goldman Sachs.

First one would be just in terms of the capacity allocation question that came up earlier. Maybe to put it in a different way, can you maybe -- I know you're sort of a key supplier in many embedded and other kind of mission-critical applications, automotive and the like. Can you give us a sense about like a floor level, a minimum floor level that some of those sort of legacy or embedded applications would reach or not below in terms of production kind of given your importance in that market? Or conversely, can you maybe talk a little bit about sort of what is the maximum level of data center capacity or data center exposure do you feel comfortable having?

Yes. I mean I think there are a couple of things to keep in mind. First of all, the overall TAM is also shifting more towards data center, that's just how the dynamics of the industry are playing out. So certainly, our portfolio and our mix is shifting towards data center. Our whole goal over time is to keep mixing our products within each segment and across segments towards higher ROI type of homes for our bits and for our output. And that just means that for some time now, the portfolio mix has been an important tailwind for Micron and continues to be a driver of our longer-term financial performance because we have, over time, shifted more and more of our revenue towards higher-margin products and taking a bigger share of the industry profit pool based on that initiative. So that will continue.

With that said, you're right. I mean, we do have large share in automotive and industrial and several mission-critical applications. So we are making investments in our Manassas, Virginia fab in the U.S. to modernize it, bring 1-alpha technology, DRAM technology into that fab. And so that is going to be an important initiative to continue the support of long life cycle legacy products and we intend to continue to support them. Of course, we also support a lot of these products and customers and segments from our high-volume fabs. And certainly, the demand environment and the structural challenges on supply for us as well as for the industry just make it very difficult to meet all of the demand for customers across different segments.

So a lot of work that we are doing is to try and assess the threshold level of supply that each of our customers need, encourage them to also figure out other sources of supply in this environment where we may not be able to meet all of the demand. Of course, we also continue to make sure that our commitments towards our customers are kept. We are certainly very big on that. So it's a delicate balancing act. It's a difficult environment to meet all of our customers' needs. We're working very, very hard to do the best we can on that front across the different segments of the market, including the ones that you mentioned.

And then just a quick clarification. Relative to the Hiroshima clean room expansion you referenced on the call, can you talk about when that kind of would turn into volume production capability?

Sure, Jim. What we said is that we are working to develop and deploy next-generation DRAM technologies there with the support of METI and our Boise R&D team working with our Hiroshima R&D team. So it will be future generations of technology we'll be deploying and the clean room space will support those future technologies. So it will be timed to be able to support those -- the space needed for those new nodes.

But not in 2026, to be fair.

No, not in '26.

Our next question comes from the line of Joseph Moore of Morgan Stanley.

Along the same lines, you had talked about the PC market could be limited by DRAM availability. How do you make those kinds of allocation decisions? Is it kind of just the highest gross margin is the focus? And how do you make sure that the mid-tier PC companies kind of have enough that there customers when you need them down the road? Just how are you kind of thinking about balancing the allocation across these different markets?

With a lot of difficulty. Certainly, there are competing requirements. We are not just focused on optimizing gross margin. Of course, when we think about pricing in different markets, we try to ensure that the pricing reflects the value of the product that we are selling. And so a lot of the margin performance of different segments starts to get very close to each other in situations like this.

With that said, we have had customers of different sizes across different parts of the world for a number of years, decades, sometimes, and so we are very mindful of our responsibility to them and we try to do our best to manage the allocation in an appropriate way to ensure diversity across segments, ensure that we are supporting our strategic customers and doing the best to maximize our supply towards these customers in times like this, as well as to minimize any impact that could occur to our customers' business due to lack of adequate DRAM or NAND as the case maybe.

That's helpful. And then just my follow-up. You mentioned a few minutes ago paying down some debt. I mean it seems like you're going to generate a really large amount of cash here in the next few quarters if your view that gross margins continue to improve comes true. Just what are the priorities for that cash over time, just between dividend, buybacks or other uses of cash that you might think about?

Yes, Joe, the priority for our cash generation is always to reinvest in the business. So we're going to make sure that we're investing in this case the capacity we need to support the market and ensure that we're getting clear line of sight to return on that. And that includes, in this case, now looking to strike commercial arrangements with customers that reflect the value of our products and our assurance on supply. So that's a priority. And also with that, maintaining the technology leadership that we enjoy today and that we intend to maintain.

The balance sheet also will always be a priority, and that's moving in the right direction. It was already strong and getting stronger every day, in fact, today, on this call, we're record liquidity. And we recently got put on positive watch by one of the agencies. And we effectively have no net leverage now. In fact, we're a positive cash, net cash and our gross leverage is very low. Now we do have, on an absolute basis, more debt than we've had historically. So we would expect to bring that down some.

And then we would -- as we've said before, intend to grow the dividend over time. And then repurchase shares with excess liquidity that we believe we have. Now we repurchased $300 million in the first quarter. If you recall, I mentioned a couple of quarters ago, I believe that when we signed the definitive agreement for the chips program, our repurchases have some limitations for the first 2 years. We've ended that first year. We're in that limitation period in the second year. We are able to buy back some stocks this year, a little bit more than what we bought back in the first quarter. But then we're -- this time next year, we're largely unconstrained. I mean there are some requirements on CapEx and R&D, which will -- we believe we'll easily be able to surpass those. So I think that's the rough cut of the capital allocation.

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

I just have one. Could you remind us as a mix of your CapEx this year, how much was construction CapEx? Or if you haven't given that, were trends on a normalized basis?

Yes, Tom. So we haven't, but we did say -- I think Mark did say that from fiscal year '25 to '26, we're doubling our construction CapEx. That gives you some sense. We got...

Yes, just looking for the base.

Higher than -- Yes. Yes.

Thank you. And ladies and gentlemen, we have reached the end of our time. So that does conclude today's conference call. Thank you for participating. You may now disconnect.

Thank you for standing by, and welcome to Micron's First Quarter 2026 Financial Call. [Operator Instructions] As a reminder, today's program is being recorded. And now I'd like to introduce your host for today's program, Satya Kumar, Investor Relations. Please go ahead, sir.

Thank you, and welcome to Micron Technology's Fiscal First Quarter 2026 Financial Conference Call. On the call with me today are Sanjay Mehrotra, our Chairman, President and CEO; and Mark Murphy, our CFO. Today's call is being webcast from our Investor Relations site at investors.micron.com including audio and slides. In addition, the press release detailing our quarterly results has been posted on the website, along with the prepared remarks for this call.

Today's discussion contains forward-looking statements that are subject to risks and uncertainties. These forward-looking statements include statements regarding our future financial and operating performance as well as trends and expectations in our business, contractual terms, market, industry, products and regulatory and other matters. These statements are based on our current assumptions, and we assume no obligation to update these statements. Please refer to our most recent financial reports on Forms 10-K, Forms 10-Q and our other filings with the SEC for more information on the risks and uncertainties that could cause actual results to differ materially from expectations.

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 can be found on our website. I'll now turn the call over to Sanjay.

Thank you, Satya. Micron had an outstanding start to fiscal 2026, delivering fiscal Q1 revenue, gross margin and EPS well above the high end of our guidance. This financial performance was driven by our strong execution across end markets and products in a tight supply environment.

We achieved a number of records in fiscal Q1. Total company revenue, DRAM and NAND revenue as well as HBM and data center revenue and revenue in each of our business units also reached new records. We have completed agreements on price and volume for our entire calendar 2026 HBM supply including Micron's industry-leading HBM4. We forecast an HBM TAM CAGR of approximately 40% through calendar 2028 from approximately $35 billion in 2025 to around $100 billion in 2028. This $100 billion HBM TAM milestone is now projected to arrive 2 years earlier than in our prior outlook. Remarkably, this 2028 HBM TAM projection is larger than the size of the entire DRAM market in calendar 2024. We are excited about our customized HBM4E customer engagements which offer further differentiation opportunities to us, and we continue to make excellent progress on our HBM road map.

Memory is now essential to AI's cognitive functions, fundamentally altering its role from a system component to a strategic asset that dictates product performance from data center to the edge. This structural shift means that system capabilities heavily rely on advanced memory for real-time contextual processing, which is vital for achieving autonomous and intelligent behaviors in AI data centers as well as in applications ranging from self-driving cars to advanced medical diagnostics.

With our technology leadership, differentiated product portfolio, strong operational execution and solid balance sheet, Micron is in the best competitive position in its history and is one of the semiconductor industry's biggest enablers of AI. We anticipate substantial new records in revenue, gross margin, EPS and free cash flow for both the second quarter and the full fiscal year 2026 and we expect our business performance to continue to strengthen through the year.

Sustained and strong industry demand, along with supply constraints are contributing to tight market conditions and we expect these conditions to persist beyond calendar 2026. We are making progress with customers in our discussions for multiyear contracts with specific commitments.

Simultaneously, we are focused on maximizing our production output from our current footprint, ramping our industry-leading technology nodes and investing in new cleanroom space to add to our supply capability.

Micron's technology leadership is foundational to our strong competitive position. Micron has led the industry for 4 consecutive technology nodes in DRAM and 3 nodes in NAND with progressively faster yield ramps in every node. Our 1-gamma DRAM node is ramping well. 1-gamma will be the primary driver of our DRAM bit growth in calendar 2026. And will be the majority of our bit output in the second half of the calendar year. Looking beyond 1-gamma, development is underway for 1-delta and 1-epsilon nodes which will feature innovations that we expect to extend our differentiation and technology leadership.

In NAND, we are ramping our G9 node with robust yield ramps across both data center and client SSDs. Our QLC NAND mix, including G9 QLC, reached a record high during the quarter. Technology transitions to G9 will be the primary driver of our NAND bit growth in calendar 2026, and we expect it to become our largest NAND node later in fiscal 2026.

I'm pleased to report that calendar 2025 is a record year for Micron in terms of both internal and customer quality measures, positioning us well to deliver for our customers as the memory industry's quality leader. As our products are increasingly integrated into higher-value applications, our leadership in quality is becoming a more important differentiator.

Turning to our end markets. As the world's leading technology companies advance toward artificial general intelligence, and transform the global economy, our customers are committing to an extraordinary multiyear data center buildout. This growth in AI data center capacity is driving a significant increase in demand for high performance and high capacity memory and storage. Server unit demand has strengthened significantly, and we now expect calendar 2025 server unit growth in the high teens percentage range higher than our last earnings call outlook of 10%. We expect server demand strength to continue in 2026. Server memory and storage content and performance requirements continue to increase generation to generation.

Micron has a differentiated portfolio of high-value data center solutions to address these requirements, including our HBM, high-capacity server memory solutions and data center SSDs. Micron's HBM4 with industry-leading speed over 11 gigabits per second is on track to ramp with high yields in the second calendar quarter of 2026, consistent with our customers' product ramp plans. Our HBM4 uses advanced CMOS and advanced metallization process technologies on the base logic die and DRAM core dies which are designed and manufactured in-house. This, along with our unique HBM design, packaging and test capability enables Micron's industry-leading performance and low-power leadership.

Micron pioneered the adoption of LPDRAM in the data center. Micron's low-power DRAM server modules consume 1/3 the power of DDR DRAM server modules. Building on this leadership, we have sampled our 192 gigabyte LP SOCAMM2 product which enables a 50% increase in capacity per module and a rack scale LPDRAM density of over 50 terabytes.

Our data center NAND portfolio revenue exceeded $1 billion in fiscal Q1, and we are seeing strong momentum across our data center SSD portfolio enabled by our leadership NAND technology. In the performance SSD category, Micron has introduced the world's first PCIe Gen6 SSD, leveraging our G9 NAND. We are seeing rapidly increasing qualification commitments for this product, including at hyperscalers. In mainstream storage, our SSDs based on G9 NAND are already seeing robust demand in the first quarter of calendar 2026.

In capacity storage, our QLC based 122 and 245 terabytes, G9 SSDs are entering qualification at multiple hyperscale customers. PC demand continues to be driven by Windows 10 end of life and AI PCs. We forecast PC unit sales to grow high single-digit percentage range in calendar 2025, above our prior expectations provided in our last earnings call of mid-single digits. As we look ahead into 2026, we expect these demand drivers to continue while memory supply constraints may affect some PC unit shipments. Micron has completed multiple OEM qualifications of our 16 gigabit, 1-gamma based, DDR5 and our G9 based PCIe Gen4 QLC SSDs.

Turning to mobile, smartphone unit volumes in calendar 2025 are on track to grow in the low single-digit percentage range. AI is driving memory content growth. The shipment mix of flagship smartphones with 12 gigabyte of DRAM increased to 59% in calendar Q3, more than twice the level from a year ago. Micron is accelerating innovation across our mobile DRAM portfolio. In fiscal Q1, we began sampling our break through 1-gamma 16 gigabit LPDDR6 product to leading OEM and ecosystem partners, marking a major milestone in next-generation memory technology. LPDDR6 will power AI at the edge delivering over 50% higher performance and improved power efficiency for flagship smartphones and AI PCs. Micron also sampled our 1-gamma LP5X 24-gigabit product and began volume shipments of the previously announced 1-gamma LP5x 16 gigabit product to multiple OEMs.

Turning to auto, industrial and embedded. In automotive, L2+ and L3 adoption is driving robust demand today and our customers' road maps indicate a significantly higher memory content in fully automated vehicles. Micron is uniquely positioned for growth with our differentiated product portfolio and automotive market share leadership. Our ASIL-rated LPDDR5X and UFS 4.1 NAND products optimized for automotive and advanced robotics that include bandwidth-enhancing features are seeing strong demand and have already secured billions of dollars in design wins.

In industrial, demand continues to strengthen, driven by the growing adoption of autonomous systems across various applications. Long-term demand trends trajectory remains robust for memory and storage in industrial applications, such as in factory automation, aerospace and defense, humanoid robotics, edge networking and video surveillance. Across both auto and industrial markets, LPDDR4X and DDR4 are also experiencing strong demand, and we are making investments to provide long-term supply from our Manassas, Virginia fab.

Now turning to our market outlook. Over the last few months, our customers' AI data center buildout plans have driven a sharp increase in demand forecast for memory and storage. We believe that the aggregate industry supply will remain substantially short of the demand for the foreseeable future. The dramatic increase in HBM demand is further challenging the supply environment due to the 3:1 trade ratio with DDR5 and this trade ratio only increases with future generations of HBM. Additional cleanroom space is necessary to address this increased demand and lead times for cleanroom buildout are lengthening across geographies.

Together, these demand and supply factors are driving tight industry conditions across DRAM and NAND and we expect tightness to persist through and beyond calendar 2026. Calendar 2025 DRAM and NAND industry bit demand growth expectations are higher than in our last earnings call outlook. We now expect calendar 2025 DRAM bit demand growth to be in the low 20% range versus high teens previously. We expect 2025 NAND bit demand growth to be in the high teens percentage range versus low to mid-teens previously. We expect calendar 2026 industry DRAM and NAND bit shipment growth to be constrained by industry supply. We expect both DRAM and NAND calendar 2026 industry bit shipments to increase around 20% from 2025 levels.

Micron is working hard to support our customers' demand during this time, and we expect to grow our DRAM and NAND bit shipments approximately 20% in calendar 2026. Despite significant efforts, we are disappointed to be unable to meet demand from our customers across all market segments. Micron plans to increase our fiscal 2026 CapEx to approximately $20 billion versus our prior estimate of $18 billion. This increase will primarily support our HBM supply capability and also our 1-gamma supply in calendar 2026. We are pulling in equipment orders and accelerating installation time lines to maximize output capability.

Micron is also investing across our global manufacturing footprint to add supply to support longer-term demand. We are seeing an enthusiastic customer response to our planned U.S. supply. We are pulling in our first Idaho fab time line, and we now expect first wafer output in mid-calendar 2027, earlier than our prior expectation of second half calendar 2027.

Earlier this year, we announced our plans for the second Idaho fab which will begin construction in 2026 and be operational by the end of 2028. We are making good progress on securing necessary permits for our New York site and appreciate the partnership with the state of New York and the Trump administration. We plan to break ground on our first New York fab in early calendar 2026 which we expect will provide supply in 2030 and beyond.

In Japan, with the support of METI, we are making technology and manufacturing investments. We are enabling future DRAM technology transition in coordination with our Boise R&D team. We are also adding cleanroom space in our Hiroshima fab to support these advanced nodes which will increase production scale and optimize fab economics.

In Singapore, our HBM advanced packaging facility is on track to contribute meaningfully to our HBM supply in calendar 2027. As HBM becomes a part of our Singapore manufacturing footprint, we expect opportunities for synergies between NAND and DRAM production. We are pleased with the progress on our assembly and test facility in India, which has initiated pilot production and will ramp in 2026. As we make progress on our strategic manufacturing initiatives, we will continue to be responsive to the market environment and disciplined with our CapEx plans.

I will now turn it over to Mark for our fiscal Q1 financial results and outlook.

Thank you, Sanjay, and good afternoon, everyone. Micron delivered strong financial results for the fiscal first quarter with revenue, gross margin and EPS, all exceeding the high end of our guidance. During the quarter, we generated record free cash flow, reduced our debt and returned to net cash.

Total fiscal Q1 revenue was $13.6 billion, up 21% sequentially and up 57% year-over-year, setting a quarterly record for the third consecutive quarter. We saw sequential revenue growth across all our business units. Fiscal Q1 DRAM revenue was a record $10.8 billion, up 69% year-over-year and represented 79% of total revenue. Sequentially, DRAM revenue increased 20%. Bit shipments were up slightly and prices increased approximately 20%, driven by tight industry DRAM supply, pricing execution and favorable mix.

Fiscal Q1 NAND revenue was a record $2.7 billion, up 22% year-over-year and represented 20% of Micron's total revenue. Sequentially, NAND revenue increased 22%. NAND bit shipments increased in the mid- to high single-digit percentage range and prices increased in the mid-teens percentage range, driven by tight NAND industry supply, pricing execution and favorable mix.

The consolidated gross margin for fiscal Q1 was 56.8%, up 11 percentage points sequentially. This improvement was driven by higher pricing with strong cost execution and favorable mix.

Now turning to quarterly financial performance by business unit. Cloud memory business unit revenue was a record $5.3 billion and represented 39% of total company revenue. CMBU revenue was up 16% sequentially driven by an increase in bit shipments and higher prices. CMBU gross margins were 66%, higher by 620 basis points sequentially supported by cost execution and higher pricing. Core data center business unit revenue was a record $2.4 billion and represented 17% of total company revenue. CDBU revenue was up 51% sequentially, driven by robust bit shipments and higher pricing. CDBU gross margins were 51%, up 990 basis points sequentially, supported by higher pricing and cost execution.

Mobile and client business unit revenue was a record $4.3 billion and represented 31% of total company revenue. MCBU revenue was up 13% sequentially, driven by higher pricing, partially offset by lower bit shipments. MCBU gross margins were 54%, up 17 percentage points sequentially, driven primarily by higher pricing. Automotive and embedded business unit revenue was a record $1.7 billion and represented 13% of total company revenue. AEBU revenue was up 20% sequentially driven by higher bit shipments and higher pricing. AEBU gross margins were 45%, up 14 percentage points sequentially, driven primarily by higher pricing.

Operating expenses in fiscal Q1 were $1.3 billion, up $120 million quarter-over-quarter and in line with our guidance range. The sequential increase was driven by higher R&D expenses in support of technology and product development on our new DRAM and NAND technology nodes.

We generated operating income of $6.4 billion in fiscal Q1, resulting in an operating margin of 47%, up 12 percentage points sequentially and 20 percentage points year-over-year.

Fiscal Q1 taxes were $977 million on an effective tax rate of 15.1%. Non-GAAP diluted earnings per share in fiscal Q1 was $4.78 with 58% sequential growth and 167% versus the year ago quarter.

Turning to cash flow and capital expenditures. In fiscal Q1, operating cash flows were $8.4 billion, and capital expenditures were $4.5 billion, resulting in free cash flow of $3.9 billion. Fiscal Q1 free cash flow was a quarterly record, exceeding our prior record in fiscal Q4 2018 by over 20%. Ending inventory for fiscal Q1 was $8.2 billion, down $150 million sequentially with days of inventory at 126. DRAM inventory days remained tight and below 120 days. On the balance sheet, we held $12 billion of cash and investments at quarter end and maintained $15.5 billion of liquidity when including our untapped credit facility.

In fiscal Q1, we repurchased $300 million of shares as permitted by the terms of the CHIPS agreement. During the quarter, we also reduced debt by $2.7 billion, paying off a $1 billion balance of term loans and redeeming $1.7 billion of senior notes. We closed the quarter with $11.8 billion of debt and a net cash balance over $250 million. Through the fiscal year, we expect to further strengthen our balance sheet as we generate additional free cash flow.

Before turning to our outlook, I would like to share an update on how we are benefiting from AI use across Micron. Today, over 80% of our professional workforce actively uses GenAI with total usage up tenfold since last year. In manufacturing, integrating AI into yield and quality management has cut root cause identification time by half in cases. Our coding teams are realizing productivity gains of 30% or more using agentic AI. In R&D, GenAI is accelerating development by reducing cycle times and design verification, product validation, issue triage and root cause analysis. Across business functions, GenAI is broadening automation opportunities and we are deploying conversational analytics to accelerate and improve decision-making. We expect Micron's use of AI across the enterprise to further strengthen our competitiveness in the coming years.

Now turning to our outlook for the fiscal second quarter. Industry demand is greater than supply for both DRAM and NAND. We expect higher price, lower cost and favorable mix to all contribute to gross margin expansion in Q2. Operating expenses for fiscal Q2 are projected to be approximately $1.38 billion. As mentioned last quarter, Micron's fiscal Q4 2026 OpEx will also reflect the effect of an additional work week in this 53-week fiscal year. We expect the fiscal Q2 and fiscal year 2026 tax rate of around 15.5%.

Micron is investing in a disciplined manner across our global manufacturing footprint to better meet demand. To address tight supply-demand conditions extending beyond 2026, we now project our capital spending in fiscal 2026 to be approximately $20 billion weighted to the second half of the fiscal year. We expect free cash flow to strengthen in fiscal Q2, and we expect to generate significantly higher free cash flow year-over-year in fiscal 2026. Any impacts that may occur due to potential new tariffs are not included in our guidance.

With all these factors in mind, our non-GAAP guidance for fiscal Q2 is as follows. We expect revenue to be a record $18.7 billion, plus or minus $400 million. Gross margin to be in the range of 68%, plus or minus 100 basis points; and operating expenses to be approximately $1.38 billion, plus or minus $20 million. Based on a share count of approximately 1.15 billion shares, we expect EPS to be a record $8.42 per share plus or minus $0.20.

I'll now turn it over to Sanjay to close.

Thank you, Mark. AI-driven demand is here, and it is accelerating and Micron is capturing these opportunities with the best competitive position in its history. This success is built on the strength of our global team, and I want to thank our team members worldwide for their hard work and dedication. We are in the most exciting time in Micron's history and the best is yet to come. We will now open for questions.

Operator, can you queue up the questions?

Yes, sir. And I show our first question comes from the line of Timothy Arcuri from UBS.

Sanjay, I wanted to ask you about customer LTAs. I know we're hearing about D5 that's being bundled with HBM and in some cases, even NAND. So can you just talk about these LTAs? I know it sounds like these are stretching out through '26 and in some cases, even into '27. I've even heard of some stuff into '28. So can you talk about the nature of these LTAs? And then I had a follow-up as well.

These are multiyear contracts that we are in discussions with several of our key customers. And these contracts, of course, involve DRAM as well as NAND. And with respect to terms, of course, these contracts that we are under discussions for are very different from prior LTAs. They have specific commitments in them and much stronger contract structure. And beyond that, I can't be giving you specifics at this point. Of course, in the future, if and when appropriate we'll be sharing further details.

And then, Mark, I wanted to ask you about CapEx. So you took it up to $20 billion net, but it still seems, I mean, you're not guiding all fiscal '26 revenue. So we don't really know what the capital intensity number is, but it seems like it's like 25% to 30%, which is a little below your 35% metric that you usually think of. So is that because you're constrained because of fab space? And can you just talk about does that sort of like roll into fiscal '27 where we would see CapEx up more near that 35% range?

And I show our next question in the queue comes from the line of C.J. Muse from Cantor Fitzgerald.

Yes, yes. They didn't answer the prior question. Operator?

Yes, sir.

Can you turn it back to management so they can answer the prior question?

Sure. Management's line is open, please proceed. I believe the speakers' lines are muted at this time. Ladies and gentlemen, please continue to stand by. Your call will resume momentarily.

Operator, can you hear us?

Yes, we can now. You are unmuted, sir. Please proceed.

Okay. Not sure what happened there. We were on mute here. So I just want to make sure that you heard Sanjay's response, correct?

We did not.

You did not hear Sanjay's response? Okay. Thanks, Tim. So Sanjay...

We heard, I heard Sanjay's response to the first question, but I didn't hear your response, Mark, to my question on CapEx and on capital intensive.

Okay, Tim. So you're right, Tim. We're not providing a full year revenue guide. We did indicate that our CapEx was going up in calendar -- in fiscal '26. Anyhow, a substantial part of that CapEx is to support DRAM and specifically HBM and the 1-gamma and 1-gamma ramp. I would say that from '25 to '26, the plan is roughly to double the brick-and-mortar construction CapEx and at this time, we would expect '27 CapEx to be up and -- but I want to emphasize that Micron is going to remain disciplined on CapEx growth to support bit demand and bit supply and that supply will be in line with demand.

As your question of capital intensity, our capital intensity, of course, is dropping as the market conditions remain very constructive. And of course, we are working to be very efficient with our capital spend.

And I show our next question comes from the line of C.J. Muse from Cantor Fitzgerald.

I guess, Mark, to follow up on the prior question around CapEx and the relative growth seems very conservative in the backdrop that we're in. And it feels like that you're just sitting here without cleanroom space. And it also doesn't sound like you're meaningfully pulling in cleanroom. So can you talk about the philosophy there? And I guess what I'm taking away from your commentary is that you're being very conservative and judicious with adding capacity here.

Well, I would say that we've been indicating issues with supply for several quarters that we were working inventories down, that node transitions were going to be the principal source of supply growth in fiscal '26 and that's exactly what's happening. And we know that cleanroom space takes time and the HBM growth, which has only picked up with AI-driven demand has further pressured supply.

So there's no near-term solution. As we said in the prepared remarks, the entire industry, we expect to be short to demand and we're no different in that case. But we are moving quickly to do our best to provide customer supply. And we have pulled in tools. We have accelerated construction in Idaho. We are doing everything we can within our existing footprint and near-term capacity expansions to deliver supply. And so we provided a bit growth number for '26 and that is supply constrained.

And I'll just add that, of course, we are continuing to make the investments in technology transitions in our existing footprint. And as we highlighted that 1-gamma node will be a majority driver of our supply growth in '26. And of course, you have seen us make investments not only in technology transitions, but also in greenfield capacity, but also enabling greater technology production capability in our existing cleanrooms in Japan, and we are making the necessary investments there as well.

So of course, we remain disciplined, but we are very much focused and trying to work hard towards increasing our supply there and pleased with our plans for Idaho 1, Idaho 2 and, of course, New York as well. And of course, in the short term, very much focused on maximizing production efficiencies, maximizing our production output from the existing footprint as well. But yes, I mean, demand fundamentals are pretty strong driven by AI, from data center to edge with the buildout of our customers and supply is significantly short. And I would say that in the medium term, we are only able to meet about 50% to 2/3 of our demand from several key customers. So we remain extremely focused on trying to increase the supply here and making the necessary investments.

Very helpful. And then I guess as a follow-up for gross margins. Obviously, the guide is quite stellar. But curious as you go through calendar '26, how should we think about cost down across both DRAM and NAND? And as you transition from 3E to 4, is there anything that we should keep in mind or we should think about it, contemplate in our models where there might be higher costs temporarily giving yields or whatnot?

Yes. So our -- C.J., our cost execution has been very good across both DRAM and NAND. Of course, we're getting some volume leverage, but spend control has been very good. Yields have been good. We do have some start-up costs coming in for the new fabs, new construction across the network. That starts to come in second half of '26 and into '27. But at these -- the size of the business at these levels, it's a relatively small impact on margin. We're not going to provide cost guidance for the rest of the year as it depends on many factors, including mix. But to our earlier question, I can say that we talked about ramping 1-gamma DRAM and G9 NAND at length for supply in '26. And those ramps are proceeding well and will be a tailwind to our cost as these nodes ramp.

And regarding your question on HBM3E and HBM4, as we have said, we'll be beginning to ramp production of HBM4 in CQ2 time frame in line with our customer demands. And of course, our HBM4 is progressing extremely well. Very pleased with our product, industry-leading product with the highest performance of over 11 gigabits per second. And so I mean that's the highest performance. And we are very pleased with its overall yield ramp, and we expect our HBM4 to be expecting -- having a faster yield ramp than our HBM3E. And of course, our mix of HBM3E and HBM4 during '26 will be very much based on our overall customer demand. And we will have both of these products with a strong profile in our '26 revenue.

Our next question in the queue comes from the line of Harlan Sur from JPMorgan.

Great job on the quarterly execution. Just over the past 3 to 4 months as we track the different ASIC AI XPU programs. There's been a significant upward revision on ASIC XPU volume shipments next year. You have Google TPU, AWS Trainium and so on, right? And all of these XPUs still going to be using HBM3E. Have you -- has the team seen this near-term positive dynamic in your order book for 3E? And given that you're fully contracted for calendar '26, like what -- and what appears to be growing upside to next year's view. How is the Micron team going to try and manage this upside dynamic in 3E alongside a strong HBM4 demand profile?

So as I mentioned, Harlan, earlier that, of course, 2026, we'll have a mix of HBM3E and HBM4. And we have shared with you in the past that we are engaged with multiple customers with the entire ecosystem here of HBM customers and very much engaged with them, and they will all contribute to our strong year-over-year growth in revenue in '26. And of course, that will be made up of both HBM3E and HBM4. So as I said, I mean, we will continue to manage the mix of 2 based on customer requirements. I can tell you that 2026 supply on HBM will be tight, non-HBM DRAM will be tight as well. So we are continuing to see, as we have highlighted in our prepared remarks, tightening supply environment. And of course, we see strong year-over-year growth in '26 for our HBM. We today upped our revenue forecast for HBM. We highlighted that by 2028, we expect it to be $100 billion TAM, and that's 2 years ahead of our prior outlook.

So of course, HBM is on a good trajectory. And what I can also tell you is that customers as their architectures are evolving, as their platforms are evolving and these are customers across the ecosystem. Of course, they are requiring more and more HBM. I mean the value of memory in terms of ability to deliver the AI capabilities and the functionality and the performance, memory is critical and more HBM is required, and that's across the various AI platforms in the industry.

I appreciate that, Sanjay. And then after 2 to 3 quarters of enterprise SSD sort of muted trends, I believe the team saw a strong acceleration in the business, right? According to some of the third-party research estimates, I think your enterprise SSD business grew like 25% sequentially in the most recent quarter, right? So you're the #3 market share leader amongst 8 or 9 competitors, right? Very strong share position. Given, I would assume, increasing demand trends here, expanding lead times, is the Micron team also entering into long-term supply agreements with your eSSD customers? And then secondarily, I mean, SSD demand, is it more tied to expansion in inferencing workloads as customers aggressively move to monetization? In other words, is storage intensity higher on inferencing versus training workloads?

So with respect to enterprise SSDs, really very proud of our engineering and business teams and of course, our sales teams in terms of our customer engagement and the strong momentum and the share gains that we have with our enterprise SSD. Of course, our enterprise SSDs are a big part, an important part, let me say, of our data center strengthening mix. Of course, DRAM is continuing to increase in mix towards data center, but data center SSDs are an important part of our overall revenue mix. And we expect to continue to focus on share gains with our strong SSD road map, customer engagement and great quality that we provide to the customers.

And as I mentioned, that our multiyear contract that we are in discussions with, with our several key customers, our SSDs, our data center SSDs are also part of that. And let me tell you that these multiyear contracts are not just about data center customers. They also are about multiple customers across our market segments here.

And in terms of your questions on is the SSD requirement growing with inferencing versus training? What I can tell you is that in the data center AI applications, I mean, as the generative AI moves to more and more video, of course, that drives greater our demand for more SSDs as well. So I mean, the rapid evolution of AI from training to inferencing and a rapid evolution of AI models and applications, they are all driving greater growth of enterprise SSDs, yes, fueled by GenAI.

And I show our next question comes from the line of Tom O'Malley from Barclays.

Sanjay, you've been helpful in the past about kind of talking about Micron's ramp in HBM and then also the ramp of the total market. You gave some new color on HBM with the $35 billion moving at a 40% CAGR. But Micron specifically, I was curious if you could give us any color on percentage of the DRAM business today that's HBM from a dollars perspective. And then on share as you move into next year, obviously, there's a large competitor that is looking to become more competitive at 3E. We haven't heard anything on 4 yet. How do you feel about your competitive positioning into next year? And do you think that you're going to make any strategic decisions differently based on the public certification of their memory in the next couple of months?

We feel very good about our competitive position. We feel very, very good about our product and our HBM4 product that we have highlighted as industry-leading performance over 11 gigabit per second, the best specifications in the industry with our performance. And of course, we feel very good about the power consumption in our products as well. In the past, we have shared with you that how our HBM3E is 30% lower power than any of the competitors in the industry, and we are maintaining that momentum of low power, which you know in data center applications is very important.

So we are maintaining our performance in power and of course, our strong capacity position with HBM4 road map as well. So we feel very good about our competitive position, about our road map and our road map going beyond HBM4 for the future years, beyond '26 as well. And we are very proud of our team's ability to execute successfully over the course of last several quarters in terms of ramping up production capabilities of HBM3E and we shared with you that in CQ3, we reached our share of HBM3E to be in line with our -- I mean, our HBM share to be in line with our DRAM share. And we have always highlighted that as we have reached that share, we will, particularly in the tight supply environment, we'll be managing the mix of our HBM as well as our non-HBM, all of it is in high demand. And HBM as well as non-HBM has strong profitability.

So looking at our strategic customer relationships as well as our overall profitability goals and growth objectives, we'll continue to manage that mix between HBM and non-HBM, but of course, HBM is growing, and we have highlighted that how we expect the TAM to be $100 billion by 2028, a couple of years ahead of our prior projection. And of course, we will grow our HBM as well, 2026 will see a strong year-over-year growth in our HBM. And this is a tight supply environment, as I mentioned, that the gap between the demand and supply for all of DRAM, including HBM is really highest that we have ever seen, and I quantified it earlier as well. So in this environment, of course, working closely with our customers, we are continuing to manage our mix of the product here. As we focus on also continuing to increase supply to better address our customers' demand requirements.

Perfect. And just as a follow-up, you've said historically kind of the $8 billion run rate. If you look at November and February, you're taking up the total TAM, but any color specifically on HBM contribution in the November quarter and what you're expecting in the guide?

We are not really providing those specifics here in terms of the breakout. I mean we highlighted that in FQ1, our HBM revenue was a record, we did highlight that. And beyond that, we are really not going to be providing the specifics in terms of revenue. And just would tell you again that year-over-year in '26, we'll be seeing strong growth in our HBM revenue. And again, our product is very well positioned. So I mean, that's a very good place to be in, in terms of managing the overall mix of the business.

And I show our next question comes from the line of Krish Sankar from Cowen.

Congrats on the phenomenal results and guidance. My first question is for Mark. I know you spoke about the sustainability for the next year. I'm kind of curious how to think about gross margins beyond the February quarter, like into May, is it going to improve or sustain at these levels? How to think about the gross margins? And I have follow-up for Sanjay.

Thanks, Krish. For -- we're not guiding margins beyond Q2. We did guide a record Q2, as you know, 68%, 11 points sequential improvement, 7 points better than the previous record. We did indicate that we would -- our business would strengthen through the year. And so we do believe margins can be up. We do believe that they will be up for DRAM and NAND. Now keep in mind that at these high gross margin levels, mathematically, we get less in gross margin percent for the same increase in price. So yes, we would expect gross margins to expand beyond fiscal Q2, but we would expect that growth to be more gradual than what we've seen in the last couple of quarters or the first quarter and the second quarter guide.

Got it. That's super helpful...

And Krish, just one last thing. We have indicated that strengthened through the year because we believe this constructive market environment will remain so through the year, these favorable market conditions. But we also -- we're executing very well on cost. And as Sanjay mentioned earlier, we're deploying the bits to the valuable part of the market and where we can serve our customers best.

Got it. Super helpful, Mark. And Sanjay, just as a follow-up to the earlier question. I understand you don't want to like put some boundary conditions, but a year ago, you totally nailed it when you said you're going to get like HBM market share close to your DRAM market share, the DDR market share. So when you talk about like a 2028 CAGR of 40%, $100 billion TAM, how do you think about Micron's HBM market share in that realm? Should we assume it's going to be the low 20%, that spectrum or is going to be lower?

So Krish, again, we are not really going to be specifying the share. As we have said, we will be managing the mix of the business between HBM as well as our non-HBM. It's like any other product in our portfolio that when you have a strong product road map across the portfolio, then of course, we manage the mix across our portfolio with all the strategic reasons and customer relationships in mind. So Krish, we are not really going to break that down. And all I would say is that in this current industry environment, which we see as durable industry fundamentals in the foreseeable future. We are in a very good position with all the tailwinds of our product portfolio and, of course, the increasing value of memory across the board, of course, HBM, but also non-HBM in data center and other markets we will just remain very focused on managing the mix of our business and of course, managing for the best in the mid-term as well as keeping in mind the longer term.

And our next question comes from the line of Chris Danely from Citi.

So I just want to dig in on these long-term customer contracts you guys are negotiating. Can you give us any more sense of when you think you'll be able to sign these and then maybe just talk about what the holdup is? Is it just the unprecedented length or size? And given that the AI companies are asking for so much of your capacity, are you able to get them to more or less contribute to the building of a new fab?

So we will not really get into the specifics around our contract discussions with our customers. But again, I will highlight a couple of important factors that customers are concerned about long-term access to adequate memory in the environment that we are heading into. And that's leading to constructive dialogues with several key customers and across our multiple markets in terms of their supply as well as other important specific commitments related to our -- these longer-term agreements.

Not getting into the specifics, but as I highlighted, our contract structures that we are discussing are not like anything before. They are far stronger contract structures with specific commitments. And of course, different from prior contracts is that those used to be like 1-year contracts, and these are multiyear in nature as well. And as we look at addressing the customer discussions, of course, we have to look at our overall supply. And I have mentioned to you that in the mid-term, medium term, we are only able to meet half to 2/3 of the demand from our several key customers. So all of that, as we are managing our customer relationships and keeping our strategic objectives in mind, all of that has to be taken into account as we manage our contract discussions. But really not getting into the specifics here.

That's still very helpful, Sanjay. And for my follow-up, just a question on HBM and the pricing there. So given that the demand is so strong, I think you said you're sold out for '26. Are you guys locked in to a set price? Or can you let that price more or less float a little bit given how strong demand is like DDR5 does, for example?

We are really pleased with our product position and our ability to work with our customers as we highlighted, in our prepared remarks that our HBM for 2026 is sold out in terms of volume and our negotiations with customers have been completed for calendar year 2026 for volume as well as pricing. And as we have always highlighted that our HBM has strong profitability. And of course, very much focused on ROI and our non-HBM business also clearly has healthy profitability as reflected in the results that we produced as well as in the guidance that we have provided here.

Great. Congrats again on the results.

Thank you.

And our next question comes from the line of Vivek Arya from Bank of America Securities.

Sanjay, I'm curious, at what point does increasing memory price impact demand for electronics? If you set aside the data center and the AI market, do you see some elasticity? Do you see any impact on demand as you look into 2026 for more consumer and kind of traditional enterprise products? How does that shape where memory pricing can go next year?

We have highlighted in our prepared remarks that in some of the consumer markets, I mean, some of the unit demand may get impacted, given semiconductor prices here, given memory prices here. And of course, some of the customers may have, for example, in smartphone and PCs, they may have some mix adjustments in their portfolio as well to address available supply to them. But these are accounted for in our forecast that we have. And so I mean, some of the possible impact on unit demand and some of the customer mix changes have been accounted for in our forecast. And we, of course, even then we see a very, very tight supply environment here in the large gap between the demand and supply.

However, I will highlight to you that AI experience across from data center to edge, including in these edge devices like smartphones and PCs and other devices, AI experience really more memory is essential. So without sufficient memory, that AI experience, the functionality, the capability does get impacted in these edge devices as well. So I mean the punchline here is that AI across the board from data center to edge is driving increase in content and increasing requirement for memory as the customers look ahead at their road maps. And that's why customers are, of course, working with us with respect to access to supply for their long-term multiyear plans here as well.

Thank you. This concludes our Q&A session and today's conference call. At this time, I would like to end today's conference call. Thank you all for participating. You may now all disconnect.

From Micron Technologies, Executive Vice President and Chief Technology and Products Officer, Scott DeBoer; Executive Vice President and Chief Financial Officer, Mark Murphy; and your moderator from RBC Capital Markets Research Analyst, Srini Pajjuri.

Good morning, everyone. Thank you all for joining the RBC TIMT Conference. Thanks, Mark, and thanks, Scott. And also, we have Satya from the IR team at Micron in the audience. We have roughly about 45 minutes or so. Mark is going to start off with a few comments, and then we'll get into Q&A. And then we should have enough time for any questions from the audience in the end. With that, let me hand it over to Mark. Mark?

Okay. Good morning, everybody, and Srini, thank you for having us here today. I'll start with the safe harbor. We'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, including our most recent 10-K. So things are good at Micron right now. Business conditions have continued to improve since our September earnings call. And we're executing well on all fronts. Technology, which you'll hear about today from Scott and products from Scott as well. And operationally, we're doing very well.

Data center demand, which includes a lot of our highest value products is especially strong, but also other markets -- all other markets are also healthy. We expect growing AI demand to drive a multiyear data center build-out globally. We've seen much more supply-demand tightness than we expected at the time of our earnings call comments. And that is allowing us to drive robust pricing trends, and we're doing that across markets. Now we project tightness to continue beyond 2026, and that's due to both supply and demand factors. Consequently, customers have approached us about entering into multiyear contracts. Now on contracts, as we noted in our fourth quarter earnings call, we said we expected to close HBM negotiations in the coming months.

Today, we are pleased to announce that our HBM supply is fully contracted for calendar '26, and that's both for HBM3E and HBM4. On supply, in our assessment, industry supply response is limited by clean room space availability in the near term to address all the demand opportunities that we have in front of us. Inventories are very lean. DRAM, as we've talked about before, is below our target levels. NAND is improving. And by the end of the year, we expect it to be near target levels. Near term, we are growing supply through the ramp of our node transitions of 1 beta and gamma on DRAM through existing clean room capacity. We're also engaging in a lot of productivity gains and optimization of that capacity. Also, construction is ongoing for new clean room capacity for future supply requirements. Now ultimately, our success is built on a foundation of strong execution that leverages our leadership technology and product position. That's the focus today. And I'm very pleased to be joined today by Scott DeBoer, Micron's CTO, where he can provide his views on Micron's technology and product position. Scott?

Great. Thanks, Mark, and thanks for having us here today. I'm really mostly going to focus on our position and where our products sit today. And Micron is in the strongest position in history in terms of our technology and products in the history of the company. We have product leadership in both DRAM and NAND, and that's really enabled by a core innovation engine that we have at Micron, a global technology team and a strong partnership with our manufacturing operations. that has led to both consistent -- 4 consistent nodes of technology leadership on the DRAM side and the NAND side, but also sequentially improving yield ramps on every one of those nodes through the partnership with our technology teams and our manufacturing teams.

So it's really something that is a strength of Micron is our ability to generate new technology in a very efficient and quick manner and then to ramp it fast into volume manufacturing. We've enabled a number of nodes, as I mentioned. Right now, we have our proven 1-beta technology in high-volume technology and ramped already. We also have our 1-gamma node, which is now at mature yield and ramped on some products and it's continuing to ramp on more products through the rest of the next 2 years. So really strong position with those 2 nodes, both in mature yields and ramping faster.

And then our next nodes, our 1-delta and our 1-epsilon nodes are both planar DRAM nodes. And those are ramping over the next several years and will be the focus both of our high-density products, our LP memory and our DDR memory as well as eventually being part of the HBM landscape. Longer term, we are focused on true 3D DRAM beyond those planar nodes. And just like we've done in the past with other technologies, our focus is on bringing that technology to market at the right time when we believe it's cost effective and brings a competitive advantage. So I think we're in a very strong position with our 3D technology on the DRAM side, and we look forward to bringing that in at the right time. We haven't picked an exact year yet because it's going to depend on how technology evolves on the planar side and when we reach both cost and performance crossovers.

On the DRAM product side, we've had a very strong year with our -- in particular with our LP and our DDR products. And both of those now on our 1-beta node and our 1-gamma node are ramping, and they're a bit different in how you think of them now. Of course, both of them have significant aspects that are commodity and the fact that we have, in that sense, leading-edge technology, power and performance that differentiate themselves in terms of just the pure performance in the market. We also, though, have significant parts of those that are more differentiated. So even those products today for us, we don't think of as products without differentiation, and that's visible in the fact that our LPDDR5 DRAM was the leading data center LP in the market and sole sourced for a big part of last year.

On the rest of the product front, we've focused significantly from a customer engagement point of view. So our customer partnerships over the past year or 2 years have been a strong focus because all these products, whether it's the LP that is with sole source that I mentioned or our HBM product that I'll talk more about in a couple of seconds, are really results of long-term customer partnerships to build in features that our customers want and foresee and take multiple years to build in. And that's really the story of our success on our DRAM products and our NAND products and for sure, on our HBM products. So following our HBM3E, which we've talked about publicly for several quarters and really was a ramp from 0 to a substantial market share position based on excellent execution, business strategy and the best product in the market.

We're now in a position with an HBM4 product that we also think will be the standard setter for the future benchmark for HBM4 performance. And a couple of things on the HBM front, and I'm sure there'll be some questions on HBM4. We have started with a product that is internally developed and designed to run all on Micron silicon. And the strategy around that was to focus on utilizing the core technology capability that we knew we needed to have a much higher performance HBM4 than the market was calling for a year or so ago. And by keeping that internal, we were able to utilize Micron's strength that metallization technology, the right kind of CMOS for memory in an HBM to optimize the product and the ability to design a much higher performance HBM4 product than the market was asking for a year ago, and it's aligned or still above what the capability is that the customers are asking for today.

So all that led to us to design a product that is going to demonstrate very high yield above 11 gigabits per second and basically was designed to operate in that kind of space. So even though the market was calling for something almost half of that, we anticipated that we needed to be in a stronger position. And that really did lead to our strategy for staying internal on the CMOS for designing with microns metallization schemes, which are much more aligned to optimization for memory and to position us with a product that is going to set the benchmark both for performance without any design revisions and also to have the leading power that we've had and enjoyed on HBM3E continuing into the HBM4 space.

And then last, on the NAND side, we've had, again, multiple generations of product -- or technology leadership on NAND. And with our Gen9 NAND right now in high-volume manufacturing and ramping across new products as the leading node in the industry and really positioned ourselves with now a focus on data center SSD in particular. And as you've seen, our growth on data center SSDs has been very substantial over the past year. And we positioned ourselves with high-density data center SSDs based on this Gen9 technology and 4-bit per cell or QLC technology leading the industry to really build our product portfolio on the NAND side at a different level than we've been before.

So over the past year, we've had significant success on PCIe Gen5, high-density data center drives. And now we're leading the industry with the first products coming out on PCIe Gen6. And this is the first time Micron's led a protocol change like that with PCIe Gen6 data center SSD drives, and those are getting great uptake and customer interest and really the first time that Micron has been there in that kind of position. So net, I think of my message to start with is the technology and products position is exceptional right now at Micron and really well positioned for the exciting opportunity in front of us, certainly focused on data center, but also across mobile, automotive and other spaces where our products are recognized as leading.

Great. Thanks for those comments, Mark and Scott. Scott, definitely, I do have some questions for you on the technology side. But I'm going to start off with what Mark just -- Mark's update just now. So I guess this is what, tenth quarter of the up cycle in DRAM, Mark. Historically, these cycles have lasted anywhere from 8 to 10 quarters. So I guess for an average memory investor, this looks like we are closer to the peak. But based on what you just said, you seem to have visibility that the industry tightness will probably continue through all of next year, which I think makes sense given how strong the GenAI has been. So talk to us about when you talk about allocations, especially customers asking for longer-term contracts, especially on the commodity DRAM side. So how are you approaching your allocation? And when you talk about longer-term contracts on the DDR side, are customers willing to commit to both volume and pricing? Or are we just talking volume commitments here?

Yes. So let me step back. The industry and Micron took decisive actions during the downturn to rightsize the capacity in the industry. And we're very disciplined in our CapEx investment, manage our inventories very carefully, and delayed node transitions and the wafer capacity at Micron and the industry came down. And so now we have a situation where the AI demand is very strong. And we'll talk today about reasons that AI drives higher performance product requirements. And then you have one of those in particular, high-bandwidth memory consumes a lot of silicon. So now that the industry -- or now that the demand has kicked in on AI and has broadened out to other parts of the market, the market has gotten very tight, and especially on DRAM, but is improving pretty dramatically on NAND.

So we are -- in this past year, we worked our inventory levels down and supply was provided by inventories that have been built up in the downturn. And this year, we are focused on node transitions to give us the supply we need. But overall, the supply that's going to come into the market, we believe, in now and beyond '26 is going to be inadequate to supply the -- all the opportunities in the market. And then we have greenfield capacity that's needed for ourselves and the industry, which you see coming on in '27 and beyond.

We are in discussions with customers. We've been approached about multiyear agreements. Supply assurance is important for them. And it's good for our visibility on investing the capital and technology that we need to support our customers. Those discussions are underway. We're not going to disclose price and volume commitments, but they are multiyear, and that is an indication of the importance of memory and the importance of us providing supply assurance. Now what I will say is we're extremely disciplined in our investments, be it what we're choosing to focus on our technology and product development. And then importantly, the rate and pace of our capital spend. And we, of course, take in demand signals, and we very regularly process those and run those through our capital planning models. And then we use external sources. We use a number of methods to get a good read on what sort of supply we need to bring on and win.

Ultimately, for Micron, as we've said, we want to achieve stable bit share and then deploy our bits to the more valuable parts of the market, which Scott's team and Sumit's and the broader Micron effort around getting the right technology and the right products and executing very well in Manish's team. I would just say that given the tightness in the market, given the duration of that tightness, which we now believe will be beyond '26, and given the customer interest in long-term or multiyear agreements, our CapEx that we had provided -- our CapEx number that we had provided, which was -- we had provided a run rate of about $18 billion for the year. There's going to be pressure on that to come up this year. So I would expect at our earnings call, which is about a month from now, we would probably -- we will bring that up and update you at that time.

Got it. And then your comment about customers asking for multiyear contracts. Are you seeing that across all end markets? Or is it specific to data center where the demand seems the strongest?

We're seeing it broadly.

Okay. Got it. And then, Mark, any time pricing gets to these levels, we hear concerns about, especially in the consumer markets, the bill of materials, the de-specing, -- are you seeing any of that, any of the behavior from customers as to, okay, I have 12 gig in my phone, either I have to raise prices or going to scale back on the memory content. Do you see that -- I mean, historically, have you seen that in terms of de-specing? Or do you think it's something we should expect in this cycle given how strong the pricing is?

I can't say that we've seen that yet. I mean it's -- I can't say that it's an elastic thing. I mean, the performance of the devices and especially as we get into greater applications with AI, I think it emphasizes the importance of DRAM. And in the case of our focused data center portfolio on NAND, the high-performance aspects of NAND. But as your question on consumer devices, we see DRAM content increasing. And we think that's a trend that will continue.

Got it. Got it. And then the other question we get this time in the cycle is about peak margins. If I look at what you reported last quarter, what you guided to 51.5% gross margins, I think back in 2018 cycle, you peaked at like 60% plus. So again, it's difficult to compare cycle to cycle. But given that HBM is a bigger portion of your mix, I mean, it was 0 essentially in 2018, is there any reason artificially for us to think that margins can't go back to that previous peaks? We would -- I would argue that they should actually be better because of the HBM mix. So how should we think about as we -- given the tightness comments that you made?

No, I think you're -- generally the drivers you're thinking about the right way. We're not going to provide guidance on margins on second quarter or beyond. We have said that these trends will be positive. So today, I talked about how tight the market is and how that's driving pricing. We are also operating very well. So our cost performance is good. Our mix of products, as you point out, has really been a wonderful story for Micron. I mean if you look at -- as we said on the earnings call, if you look at the high-capacity DIMMs, the HBM and the low-power DRAM in server, that was about $10 billion in our '25. And material part of our business, which had been back in '22, '23, it was near 0.

So -- and then you add on top of that, the high-performance SSDs, you're in low teens billion dollars of business for just those -- that group of 4. And it's premium products, it's better margin generally. And in -- as we look at -- look out, that will grow on an absolute basis, of course, substantially, and it will grow on a percent potentially of the mix. So these things will -- these mix effects, we believe, given our positioning on technology and products, we believe this mix effect will continue to help us. So we've said before that we think that the second quarter can be stronger than the first quarter on a margin. And even with the first quarter conditions improving, we do still believe that the second quarter will produce better margins in the first quarter.

That's great. Then switching gears to HBM, and we know that the demand is very strong, and you just told us you're pretty much sold out for next year. But there has been this persistent debate about your road map, Scott, you talked about you're having a very strong product portfolio. So maybe help us understand in HBM3E, you said you had a product leadership. Do you see that leadership, whether it's performance power -- I mean, transferring into HBM4 as well as you look out to the road map? And then maybe for Mark, as we look out to the HBM shipments over the next, I guess, year or so, when should we expect the crossover to happen between HBM4 and HBM3? And do you think Micron will follow the -- whenever the industry crossover happens, should we expect Micron to all kind of pretty much follow that same time line?

Okay. So we do expect our performance advantage to continue on HBM4. And from the very first concepts of how we created a better HBM3 product, we really look forward on HBM4 to ensuring that it was a sustainable advantage that we had. And so we have a combination of the process technology that we use to make our HBM, our advanced metallization sequence that we use on both the base die and on the DRAM chip that is different than others. And also then some very key IP on our design capability. And we do design our HBM differently than the rest of the industry. And of course, those things eventually normalize. But we have a road map of different design elements that we bring in on HBM4 and then we'll again bring in differently on HBM4E that we believe will continue to enhance our capability, both from a power and a performance point of view.

And on HBM4, in particular, the challenge really was looking ahead and making a decision to design the product at a much higher capability level than the industry was calling for a year or more ago when this product was in design. And we'll do the same thing on the future generations. But for HBM4 -- for 4 specifically, we do believe we'll have the industry's best performance on that product out of the chute at high yields. And we think that the power advantage, of course, has to be proven. But like HBM3E, early indications from our customers are that we're in a very solid position.

Yes. There seems to be this concern that because you're doing your base die in-house. You're not using TSMC at least for this generation or because you're still on 1-beta that somehow, for some reason, you have a disadvantage. So how do you kind of answer -- address those concerns?

Well, we're confident in the choices we made. To me, it feels a little bit like where I was sitting 4 years ago when I was describing why it made no sense to put 5 EUV levels in a DRAM chip at that time because the technology wasn't ready, and it would be an inherent problem for anybody who did, it would slow down your technology road map. And at that time, we had a lot of public news about why Micron is in trouble because we're not putting a bunch of EUV into our process technology back then. So some of that is just living through it again.

In this case, I'm even more confident that the decision on the base die was absolutely the right one. Our optimizing a logic chip for HBM for memory capability is actually quite different than optimizing a logic die for ASICs. And of course, the foundries will catch up and they will figure out the things that are most important in terms of power delivery and low-voltage operations specific to signaling for memory capability. But there's a lot that has to happen there. And we've announced we are using TSMC for HBM4E. And so we've been partnering with them over a period of time to make sure when we do introduce that, that it is optimized for building HBM and not just kind of a follow-on to the ASICs designs that have been needed for a long time.

So I think that in particular, the base die is going to turn out to be a significant advantage. And yes, there's news, but a lot of that's generated in the -- a little self-interested in a specific country. But as I mentioned, absolutely the big part of the news, which is we need to redesign and we're not -- we have some technology gap. We have the best products. Right now, we've sampled the highest speeds on HBM4 in the industry. We've given our customers HBM samples over 11 gigabits per second already. We are actually the only company that can test those right now because test capability at those speeds is also important. And part of our design is a built-in self-test regime that allows us to actually test and validate our material up at that speed.

So if anything, I'm more bullish on our position technology-wise, independent of all the articles that I read also, on our HBM4 position even than we were when we were looking at our HBM3E, and we felt like our product was going to be good, but we never ramped it before. So in reality, in the last 18 months, we've gone from 0 to the market share we've talked about in terms of HBM. Our customers know how good our product is on HBM3E. So they have more confidence in us now relative to our ability to execute. And when we put a technology out on HBM4 and we talk about it, we have an even greater level of confidence now that we'll be able to execute to that.

Yes. That's, I think, a very detailed answer. So I just want to go back to Mark. Given your comment that you're sold out for next year and that also includes HBM4, should we assume that all the qualifications are done and we're ready to ship soon? Is that what, I guess, your comments suggest or qualifications still need to be, I guess, completed?

Yes. I don't think the systems aren't ready for HBM4 yet, right? So qualification, HBM is different in how we work with our customers. It's not just that our part works, it has to work in their system. So truly, we go through a number of milestones. We've hit all those so far in HBM4, but the next milestones do depend on the system optimization that we partner with each of our customers to make sure that, that -- and we don't call it qualified until it's actually in their system and working. And since most of those systems are not yet available, nobody can actually be qualified yet.

And maybe just to build on this point, tying it back to our capital expenditures. So this deep engagement with customers and this confidence that we have in our product obviously helps us in gaining confidence in the investments we're making for DRAM capacity because our DRAM capacity now is heavily impacted or influenced by our view on HBM. So we've got a very good view on where we stand on HBM, the increased use of HBM and systems and how that's going to determine supply requirements and our -- what sort of investments we need to make.

I had mentioned earlier that based on this HBM and also a number of other products and the overall tightness in the market that our CapEx -- we're going to bring that up likely in the earnings call. But that doesn't necessarily mean that the capital intensity is increasing because as defined by as a percent of sales, that may not. So we'll work through all that. But the point is the confidence that we have to invest on the basis of our technology and product position.

Great. And then going back to my previous question about when do you anticipate, I guess, that industry crossover from HBM4 crossing over from 3E. And then when it crosses over, do you think there's still going to be some demand for 3E or is this something, I guess, just like what we saw with HBM3 today? How do you see that playing out?

Well, we start shipping in second quarter, and it will -- the systems will move and it will ramp in the second half. I don't think we've given a specific crossover point. But it is a product that unlike standard products, we're not expecting a long tail in these products. I mean the technology moves very quickly and which is important that we sustain the capability we have and the deep engagement with customers. And then we're just careful about our planning. So we would expect it to move to HBM4 and largely be that. Of course, there'll still be some HBM3E for some time, but then we're on the 4 and then the 4E.

Sure, sure. And then last quarter, you kind of -- you've been gaining share for the last several quarters, and I think you pretty much got to your target, which is your DRAM share. And as we look forward, it looks like potentially there's opportunity for you to even pick up more share. I don't know if you have the capacity or supply. So how do you think about market share as we look out to the next, I guess, several quarters? And also, do you anticipate your HBM4 market share to be somewhat similar to what you had in 3? Or do you see opportunity to even improve on that?

I would just say that HBM for us is as we've talked about in the third quarter, we believe we achieved what we said we would do and achieve a share equivalent to our DRAM share. At this point, it's a large product line that we treat like other large product lines. And we look at what our customer requirements are and the bit trade-offs amongst all our businesses, and we make those decisions. So clearly, given the trade ratio, deploying more of our capacity to HBM means fewer bits for some other markets. And these are the trades we have to make. And again, we make those all the time with our products.

Got it. Got it. And then in terms of the profitability itself, it's been a great product for you, HBM. But you've been improving that cost structure and your margin structure also, I think just based on what you reported in your segment gross margins, we could see that there's consistent improvement in gross margins as well. As we go to HBM4, do you expect the profitability to be at this level? Or do you see that maybe getting even better? Or because it's early ramps, are there any, I guess, issues that we need to be aware of in early days as we ramp this new product?

I mean Scott can talk more about yield. It's yielding well at this state. And so we're, again, confident and positive about that. I would say that HBM is providing a lot of value to the system. And each generation has gotten more complex and is providing more value. And so -- and it's very difficult to build, and we have the highest performance product. So not only would we expect HBM to be through cycle, generally more profitable than the rest of the business. We would be expecting to get value for each generation that is providing more value to the system.

Maybe, Scott, you can talk about what drives that cost increase as we go from HBM3 to 4? What are some of the components?

Yes. So I mean, Mark talked about the trade ratio, an important piece. The trade ratio is how many wafers it takes us to build HBM cube bits versus DRAM. And one of the fundamental pieces is defined for all of us by spec is the die size on the HBM. So from that point of view, the HBM4 die size will drive some cost structure increase relative to just pure math on silicon size that we'll have to overcome in other ways. Now when we look and we've talked about this ramp on HBM, we look at kind of the evolution of Micron's capability here, we did come from not even having a manufacturing entity on HBM to the position we're at with market share today. So when you go from 0 up to that significant market share, you obviously become more efficient, you learn how to manufacture better and more cost effectively.

So all through this last year, our costs have continued to get better on our HBM manufacturing because of that, because of the maturity of manufacturing and our design for manufacturing continues to get better. So when we went from HBM3E 8-high to mature yield, and then we went to 12-high. We did that -- we did the 12-high faster than we did the 8-high by a significant amount. And our expectation based on the status we're at right now in HBM4 is we'll again substantially improve the timing to mature yield for HBM4 versus what we have done previously on HBM3E.

So based on that, we will keep improving our cost structure in HBM4, and that will be an offset to some of the inherent pieces around die size and other things. And we're very much focused on consistency of manufacturing flow between HBM3E and HBM4 and making sure wherever we can utilize the learning and the process improvements and the cost structure of HBM3E and translate that over to HBM4, we're absolutely taking advantage of that. So overall, I think our expectation is we continue to improve and have a very good cost structure in HBM4.

Got it. And you expect that to continue as we go into HBM4E as well, I guess?

Yes. I mean every -- HBM4E is a little more complicated because we have the mixture of more custom products and standard products. So it's, again, a new variant. Of course, we'll optimize our manufacturing costs around yield ramp and reutilization there also. But as I mentioned, we are bringing in TSMC as the base die on that. So that changes the cost structure by itself of the HBM cube. And ultimately, it will be a different kind of optimization with custom parts versus standard.

Right. Got it. Got it. So Scott, I want to switch gears a little bit to the content side of the HBM story. I mean NVIDIA gives us pretty good visibility into their road map. So we know what to expect from a content per GPU standpoint. And AMD also just recently gave us some additional details. But at the same time, as Gen AI workloads evolve from pretraining into, I guess, inferencing, and we hear new terminology like RAG and vector databases, et cetera. So I'm just curious, there seems to be a lot of interest for low-power DRAM outside of HBM as well.

And in your cloud business, I think HBM is roughly half of what you reported and the remaining is probably low power and some other high-capacity DIMM revenue as well. So talk to us about how do you see that playing out as inferencing becomes a bigger and bigger portion of the workloads? Are we seeing, I guess, a new class of memory that we need? Or I just love to hear your thoughts on that.

Sure. So I think that will -- it will play out over the next few years. And I think the key thing to come back to is there is growing -- general and growing recognition of the importance of memory products in enabling all kinds of systems, AI systems in the data center, at the edge, memory is much more front and center of system architects thought pattern today than it ever has been in history. So it will evolve to be more optimized maybe for edge applications, for robotic situations, even for automotive and certainly for data center inferencing versus training.

The thing that we look at is there's no question it's going to be memory focused on how you solve those problems across that space. And absolutely, LP is a key part of enabling that, just like HBM. We see growth both in terms of HBM content and LP content going forward to support this. And we need to be positioned for subtle changes in that, and that's what we're focused on. And a piece of that is making sure our LP products are differentiated just like our HBM products are. And I think that's part of what we've demonstrated over the past year. And we'll continue working with customers to make sure we have absolutely the best low-power DRAM products for them, whether it's LP6 or some modifications of that in the future and certainly LP5 in the near term, making sure those products are solid. So overall, we think we can shift around to meet the different applications, and there's going to be growth across all of them.