Operator: Ladies and gentlemen, thank you for joining us, and welcome to Micron's post earnings analyst call. After the speaker presentation, we will host a question-and-answer session. I will now hand the conference over to Satya Kumar, Corporate Vice President of Investor Relations and Treasury.

Satya Kumar: Thank you. Welcome to Micron Technology's fiscal third 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-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 levels of activity, performance, and achievements. We are under no duty to update any of the forward-looking statements or to conform these statements to actual results. We can now open up the call for Q&A.

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

Ben Reitzes: Thanks a lot. Mark, looking at the numbers here for the next quarter, free cash flow is going to be somewhere around $30+ billion. I just want to make sure investors understand what you're saying with regard to cash return. You're saying 100% will go back to shareholders. I assume the vast majority of that is in buyback. If you go from $30 billion in free cash flow and grow it, you could buy back 10% of the company next calendar year. At a $1.2 trillion market cap, that's where it is — 10% of the company. Are you prepared to do that and buy back at that level?

Mark Murphy: Sure, Ben. We're really pleased with the financial trajectory of the business. The combination of memory being so important to so many markets — AI data center, the edge — is enabling this technology revolution we have underway. Between our technology products and manufacturing performance, we are delivering record cash flow numbers. The last two quarters, we've generated as much as the company's entire history. We expect cash flow growth will increase in the fourth quarter. We've paid down quite a bit of debt over the past year, and cash will build. We will maintain levels of cash that we feel comfortable allow us to invest through all seasons in the business. We have these Strategic Customer Agreements, which we announced today — a meaningful number of them, and we expect more. We will hold what we believe is appropriate excess cash. We've always said we intend to grow the dividend over time — you saw us do a 30% increase recently — but the principal capital return will be share repurchase. I said today in the prepared remarks that we intend to increase our capital return from December 9th, which is the second anniversary of our CHIPS agreement signature. The rate and pace from there will be determined based on a number of factors. Absolutely committed to capital return.

Ben Reitzes: Thanks. One more — we're really pleased to see 40% and eventually half of your business on SCAs. I cover Apple, and they've never publicly said they're willing to pay full price for a component and pass it through to customers. To me, it was an advertisement that they are open for business for full-price DRAM. Are you tempted to do a lot more in DRAM? Is there an appetite for maybe a higher mix than expected of DRAM?

Sumit Sadana: When you say higher mix of DRAM, you mean versus what?

Ben Reitzes: Versus HBM, versus NAND — commodity DRAM to consumer would certainly not be on SCAs as much as a hyperscaler would, I would think.

Sumit Sadana: Just a couple of thoughts. Our mix of DRAM versus NAND tends to oscillate between 75% to 80% DRAM and 20% NAND, and we are comfortable with that mix. As it relates to HBM, we have made a strategic decision that our goal is to have our HBM share consistent over time with our overall DRAM share. We support our customers on HBM and also support them on the non-HBM portion of the DRAM business across all market segments. We definitely believe in the strength of diversity. The AEBU and MCBU businesses — both non-data center — are almost 40% of our company revenue. We like that diversity and continue to service customers across all segments, including non-HBM DRAM, HBM, and NAND.

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

Harlan Sur: Good afternoon. Back in September of last year, the Micron team said they were booked out through calendar 2026 on HBM3E. Fast-forward to now, with stronger XPU ASIC demand, is the team already booked up on volume and pricing for HBM3E and HBM4 for calendar 2027?

Sumit Sadana: The demand we have for our HBM products — HBM3E, HBM4, and even ahead of HBM4E quals — the asks from customers for volume not just in 2027 but through these SCA agreements going into 2028 and beyond are far in excess of our ability to supply. The demand continues to be well above our supply. Even when we do these multiyear SCAs, they contain volumes that are less than customers would actually like to sign up for. In a lot of these negotiations, we spend a lot of time helping customers understand that this is all we can do in this timeframe. Our demand for HBM is well above our ability to supply not just in 2027 but even 2028, across all HBM flavors. The same is also true for non-HBM DRAM.

Mark Murphy: Maybe just something to add while we're on HBM. Today we indicated that we expected market tightness to continue beyond 2027. Part of that reason is we did see the HBM TAM increase. We had said previously it would cross $100 billion in 2028 — we now see the HBM TAM easily crossing $100 billion in 2027.

Harlan Sur: That's great. The other thing I wanted to ask about is your midterm to long-term view on industry DRAM and NAND bit demand growth. Inferencing workloads have crossed over training workloads, and on the server CPU side, customers are now forecasting 30% to 40% per year CAGRs given agentics. Could you give us an update on your midterm views on DRAM and NAND bit demand CAGRs over the next few years?

Sumit Sadana: That's a good question. We have provided some updates on how we see 2026 bit demand forecasts change — DRAM forecasts up a little bit, NAND relatively similar. The reason we are not providing a lot of forward-looking views on CAGRs is because for the foreseeable future, shipment growth for bits is not really determined by demand anymore. It's determined by supply. The demand is so much above the industry's ability to supply that supply growth will determine how shipping growth occurs, far less so the demand growth. Because of that, we are trying to point out what kind of growth trajectory we believe is unfolding. Our expectation is that supply growth will continue to remain short of what is needed to meet demand. We don't really see when supply is going to be able to meet demand — that is not something we are able to project at this time.

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

Tom O'Malley: I wanted to go back to the long-term agreements. Is there any way to walk us through what happens if a customer were to cancel the agreement? What financial hooks do you have in it?

Sumit Sadana: These Strategic Customer Agreements cannot be canceled. There is no provision in this agreement that allows a customer to walk away. These are designed to be take-or-pay agreements. Outside of automotive, generally these are five-year agreements with annual volume commitments for each year. 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 — a ceiling and a floor. The price gets negotiated every quarter based on market conditions but cannot exceed the ceiling or go below the floor. There are also premiums in the agreement for more sophisticated products, like LPDDR6 versus LPDDR5 or a new version of HBM. The one provision that doesn't exist is any customer's ability to walk away. Beyond the take-or-pay obligations, there is also an upfront cash deposit where customers have provided cash deposits and related financial commitments like letters of credit. For the 16 agreements we have already signed, these aggregate to $22+ billion in total cash and related financial commitments, of which the cash alone is almost $18 billion. As we get to our target of roughly half of company revenue covered by SCAs, you can imagine that cash will significantly increase from the current $22 billion number.

Tom O'Malley: Just to clarify on the cash deposits — it's my understanding they end up getting returned to customers at the end of the agreement. Any strategic rationale for receiving them upfront and being able to use that cash? Why does it get returned rather than just being recognized as revenue?

Sumit Sadana: The cash is a contingency and a show of good faith and confidence in this new business model from our customers. In the unlikely event that a customer is unable to purchase the volume at the agreed price, we do have the right to decrement the cash balance as one available remedy — but not the only remedy. None of this ultimately relieves customers of the liability of having to purchase the volumes over the term at the agreed-upon prices. The cash doesn't get returned all in one shot at the end of the term — it gets returned over a period of time, weighted towards the second half of the agreement term.

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

Melissa Weathers: I wanted to ask on the non-HBM side of DRAM within the data center. You've talked about SOCAMM and using low-power DRAM for data center applications, especially as the mix of server CPUs increases with agentic AI. Can you give an update on how you're seeing the growth in demand for SOCAMM attach and what trends you're seeing in adoption?

Sumit Sadana: Agentic AI drives a lot of growth in CPU demand and CPU-based servers, and that is certainly a trend we are seeing. These CPU-based servers are coming from multiple suppliers — x86-based CPUs, CPUs from NVIDIA, Qualcomm announced a CPU. There's lots of different possibilities for CPU types in the data center. We have CPUs that use DDR5 as well as plans from customers to increase the use of LPDRAM in the data center. When LPDRAM gets used, it will be in the SOCAMM form factor. Micron has been a pioneer there — we were first in the industry to drive usage of LPDRAM in the data center and were sole sourced on it for the longest time. We were also the first to bring out the SOCAMM form factor. We continue to expect this to be an area of differentiation, and we have really strong engagement with customers who intend to use LPDRAM as a way of reducing power consumption, increasing performance, and reducing memory footprint. There are RAS-related complications — reliability, availability, and serviceability — that have to be worked through for LPDRAM since it wasn't designed for data centers. That's where we're bringing in differentiation and helping our customers deal with that. We expect LPDRAM to grow over time as a percent of DRAM consumption in the data center, and we expect to be leaders on that front.

Melissa Weathers: As we think about Idaho One, Tongluo, Idaho Two starting to have wafer outs next year and the year after, how should we think about the impact of startup costs on cost per bit as those greenfield fabs come online?

Manish Bhatia: Given the industry-wide trend towards higher performance solutions such as HBM — and even within HBM, higher die ratios expected in the future which require more silicon per bit versus traditional DRAM — as well as greenfield build-out which doesn't get the same leverage on existing capacity as traditional technology transitions, both of these trends are going to increase DRAM bit costs in the near term. That goes for Idaho One, Tongluo, and Idaho Two. This will be an industry-wide phenomenon.

Mark Murphy: We begin to see startup costs more meaningfully in the fourth quarter here and then into the first half of next year. You'll see 2027 at elevated levels — think about $100 million to $200 million per quarter effect versus previous run rates. What that will be over time is a function of the various ramp profiles of the fabs. I had said previously this might be a 0.5 to 1+ point of margin effect, but today, with the size of the business, this effect is much reduced from before. The benefit of the incremental bits will outweigh the incremental associated startup cost.

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

Vijay Rakesh: On the 16 SCAs you announced — do they include any HBM, and do they include some of the major CSPs within that?

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

Vijay Rakesh: On the $20 billion deposit — is the intention that you hold it for the five-year contract period, or as customers buy the product, is it prorated and runs through that way?

Sumit Sadana: It's not a prorated type of thing. It is a customer cash deposit — not prepaid revenue. It gets returned to customers on a predefined agreed-upon schedule, and it is back-end loaded, meaning the bulk of the return of the customer deposit occurs in the second half of the agreement term. The cash would be returned assuming it hasn't been decremented for reasons determined by the terms of the agreement.

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

Jim Schneider: Stepping back for a moment — as you've had discussions with customers about their forecast demand needs through the end of calendar 2028, where do you think you will end up in terms of the percentage of forecasted demand you are able to supply? Is it something like 70%, 90%? How do you expect that gap to close over time?

Sumit Sadana: There isn't a homogeneous percentage number we can provide because our strategy is different for each segment of the market. If the automotive industry can only get 50% to 70% of the units, that would be catastrophic for that part of the market — we can't have a one-size-fits-all approach. There are complex factors around what fulfillment rate is appropriate for what customer, in what segment, in what geography. With that said, the general sentiment among customers is that we are very short of their demand. For some customers, we are extremely short — our supply numbers are a fraction of what they want. For smaller but super important sectors like automotive, defense, aerospace, and critical industrial and medical markets, we try to minimize the impact. Broadly speaking, the overall aggregate supply is substantially below aggregate demand for both DRAM and NAND. DRAM is extremely constrained, and HBM is very constrained.

Mark Murphy: If I could just add — as you heard, we're doing everything we can to bring up supply. Between adding construction and tool installs in Q4, we're increasing our fiscal 2026 CapEx to around $27 billion. We're also going to increase CapEx substantially next year, with more than half of that increase being construction. Based on our comments, you may have come up with numbers in the low to mid-40% range for FY2027 CapEx as a percent of revenue — we will be spending above that level. We'll do about $10 billion this quarter and step up from there into 2027.

Jim Schneider: What are the chances you're going to do materially above 50% — is something like 55% or 60% even in the cards?

Mark Murphy: We're not going to give a specific CapEx number. If the number were that order of magnitude, I think we'd owe it to you to be more specific. We're going to run about $10 billion this quarter and step up from there. We will be higher than the mid-40% range. We're going to remain extremely disciplined. The ops team has been amazing at figuring out how to sweat these assets as much as possible while accelerating all these greenfield capacity adds.

Manish Bhatia: The majority of the fiscal 2027 CapEx is for construction, which gives you some indication that those construction dollars are not going to be producing bits in that time horizon. That's why we also said that greenfield capacity really starts to contribute to bits in calendar 2028. Even with that supply improvement, we don't see an intercept for supply with demand.

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

Aaron Rakers: On the SCAs, I'm curious about the NAND flash market specifically. As you're engaging with customers on these SCAs, is there a strategic advantage in having both NAND and DRAM in your portfolio competitively in these engagements? Is NAND pervasive across these SCAs, particularly as it relates to enterprise SSDs?

Sumit Sadana: Our enterprise SSD momentum is exceptionally strong — we had a $5 billion quarter in FQ3 for enterprise SSDs inside of the $25 billion overall data center revenue for the quarter. We feel really good about having a strong portfolio on both the NAND and DRAM side. We have hit record share after record share in data center SSDs due to the strength of that portfolio. In terms of the levels of constraint, both DRAM and NAND are very constrained. When we talk to customers across this long horizon through the end of calendar 2030 — which is the term of a lot of these large SCAs — they are definitely interested in getting their hands on NAND, but DRAM is far more constrained and more difficult to supply in the quantities customers need. The sense of concern and urgency in the minds of our customers around DRAM is very, very high.

Aaron Rakers: My final question on the competitive landscape — how have you evolved your thoughts around China and the competition from CXMT or YMTC? Have you seen any changes on that front?

Sumit Sadana: Those two companies have grown over the years in capabilities and share. Most of their output — the overwhelming majority — tends 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 on driving the highest performing, most complex products in the portfolio. On NAND, we are focused on data center SSDs in a very single-minded way. We are the QLC leader in the world. We are leaders in Gen6 — the first company to come out with Gen6 drives and advance them in volume. We lead in the highest capacity 245 TB drives. On DRAM, our strength spans everything from HBM to high capacity DIMMs, to LPDRAM leadership in the data center, to mobile and client LP leadership. Our focus is to find these complex, difficult-to-get-right products, get into deep customer engagements across multiple years on the roadmap, gain their confidence through a track record of meeting and beating time to market with best-in-class specs, and maintain a track record of innovation. We have one of the best IP portfolios in the world — almost 65,000 patents — and we are very aggressive in defending it. Overall, we feel very good about where we are. The structural foundational changes in our business model — the combination of demand, structural supply challenges, the newfound strategic importance of memory in the AI era, and now these SCAs — are completely transformative for our business.

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