Q3-25 Earnings Release and Analyst Call

Stocks A to Z
1

6/25/25

Press Release and Investor Presentation Commentary

Here are the points from the press release, outside the financial statements, that I found noteworthy:

  • Revenue, gross margin, and EPS were all above the high end of the guided range this quarter.
  • HBM revenue grew by “nearly” 50% sequentially.
  • For the first time ever, Micron is the #2 supplier of data center SSDs in the industry, according to a third party.
  • In June, the company completed a reorganization of their business units around the growth opportunity in AI.
  • Yields on their 1-gamma DRAM node are ahead of the record pace set by their 1-beta node ramp. The 1-gamma node has 30% higher bit density over 1-beta.
  • Two weeks ago, they made the repackaged announcement that the company will invest $200B in the US over the next twenty years; $150B in manufacturing and $50B in R&D. This is not a commitment of any kind. It is just a press release for appearances to help persuade the U.S. government to not take away Micron’s CHIPS Act money.
  • The first wafer output from ID1 (their new fab in Boise, ID) to begin in the second half of calendar 2027. That is more than two years away. Strictly speaking, this is not a schedule slip from prior statements (“meaningful” output in FY27,) but likely that is what happened.
  • The second Idaho fab, named ID2, will begin production before the first fab in New York. That is a change from prior statements. Ground preparation has not yet begun in New York for that fab. Word around the company is the New York fabs will never be built.
  • They expect the server market to see mid-single digits unit growth in CY25.
  • FQ3 saw the fourth consecutive quarter of record server DRAM revenue, driven by HBM, high capacity DIMMs and low power server DRAM products. This is a data point showing how the current strength in the DRAM market is almost all coming from AI demand.
  • Shipment crossover to HBM3E 12-high is expected in FQ4-25.
  • The company continues to expect their HBM share to reach a level similar to their overall DRAM share (my comment: that is low-20%) “sometime in the second half of calendar 2025.”
  • The company is now shipping HBM in high volume to four customers, spanning both GPU and ASIC platforms.
  • Their first HBM4 product will be built on their 1-beta DRAM node. Volume production will ramp in calendar 2026.
  • PC market units are expected to grow in the low single digit percentage range in calendar 2025. This is a down-revision from last quarter when they said PC unit growth would be mid-single digits in calendar 2025. They again accompanied their PC growth forecast by saying AI-enabled PCs and the Windows 11 upgrade cycle will catalyze higher PC DRAM demand.
  • In mobile, smartphone unit growth is expected to be low single digits in calendar 2025. This forecast is unchanged over the last three quarterly calls. AI adoption “remains a key driver” of DRAM growth in smartphones. I continue to think on-device AI workloads are overstated.
  • The automotive market must continue to be weak. Commentary on this slide skipped the current market conditions and jumped to general demand statements and comments on long term success.
  • In industrial, the company is seeing “ a resumption in our growth.”
  • As for the overall memory market, customer inventory levels have been “healthy overall across end markets.” They did call out that some tariff-related pull-ins have been seen. This is a tentative acknowledgement that uncertainty in U.S. tariff policy is distorting the demand picture in the memory market.
  • Their customers are telling them the remainder of the calendar year will see a “constructive demand environment.”
  • They expect calendar 2025 industry DRAM bit demand growth will be in the high-teens percent range. This is up from mid-to-high teens percent growth forecasted last quarter. And last quarter was up from mid-teens percent growth the quarter before. For the last two quarters, DRAM bit demand has surprised the company to the upside.
  • NAND bit demand growth will be in the low-double digit percent range. This is the same forecast they made last quarter. The falling knife has been caught.
  • They expect their bit growth will be below industry bit demand growth in non-HBM DRAM and NAND. Said another way, they are production both DRAM and NAND bits below natural demand levels and are making up the difference by drawing down their inventory, hence maintaining their market share.
  • Over the medium term, they are forecasting mid-teens CAGR in bit demand growth for both DRAM and NAND. This is the same as last quarter.
  • The company does not produce any D4 or LP4 products on their 1-beta and 1-gamma DRAM nodes.
  • The company has sent EOL notices to customers for DDR4 and LPDDR4 with final shipments occurring in two to three quarters from now. That is end of calendar 2025 to early in calendar 2026. They will continue to support longevity customers with these products off of their 1-alpha node for “several years.” For high-volume D4 products, customers are now on allocation.
  • D4 revenues are a low single digit percent of company revenues in the second half of fiscal 2025. The current rising prices of these products will have a small effect on Micron’s overall financial performance.
  • DRAM: 76% of total revenue. Revenue increased 15% Q/Q, bit shipments increased over 20% Q/Q, and ASPs decreased low-single-digit percentage Q/Q. At first glance, the strong revenue growth in the quarter suggests a strong DRAM market. It is surprising that ASPs were slightly down sequentially in such a quarter.
  • NAND: 23% of total revenue. Revenue increased 16% Q/Q, bit shipments increased mid-20% sequentially in the quarter, and ASPs were down high-single-digits in the quarter. This is following a quarter for NAND when ASPs declined 19% on flat bit shipment growth. The NAND market is oversupplied now, but not severely so. I say that because demand was still strong enough in the quarter to support bit shipment growth over 20% without a collapse in pricing.
  • HBM revenue was up “nearly 50%” sequentially. HBM revenue was over $1B last quarter, therefore, HBM revenue this quarter was around $1.5B. That is 21.2% of their total DRAM revenue, which is within a couple hundred basis points of Micron’s overall DRAM revenue share. Earlier in the prepared remarks, the company said their HBM market share will match their overall DRAM market share sometime in the second half of calendar 2025. They were so close for the quarter made up of March, April, and May, I can’t see how they didn’t reach this milestone sometime in June. I’ll take them at their word about the second half of calendar 2025, but that must mean July.
  • Three months ago, in last quarter’s results, the company said they would match their overall share to their HBM share in the calendar fourth quarter of 2025, so they pulled that milestone in by at least one full quarter. That is quite an achievement. I assume that pull-in was enabled by HBM output above plan. The other option is the company’s forecast for the size of the HBM market in 2025 has come down in the last three months. That would be shocking but I don’t think it is the case as it would be contradictory to everything else they are saying about HBM demand.
  • To review Micron’s HBM revenue: FQ4-24 was approximately $325M, $650M in FQ1-25, $1B in FQ2-25, and $1.5B in FQ3-25.
  • Storage business unit revenue was up 4% sequentially, primarily from an increase in customer-oriented revenue. This explains at least some of the decline in NAND ASPs. They sold more into the customer segment, which usually has the worst pricing.
  • This is surprising. Mobile Business Unit revenue was up 45% sequentially. That is a stunning rise but also consider last quarter bit shipments were down 30% because of seasonal weakness. The reason for the strong bit demand this quarter was inventory restocking at customers. I think the large swings in bit shipments from MBU the last two quarters are because the seasonally weakest quarter (FQ2-25 for Micron) lined up with customer inventory digestion. Then the seasonally stronger FQ3 lined up with customers restocking their inventory.
  • Also surprising is the 20% sequential increase in Embedded Business Unit revenue. The reason given was growth in the industrial and consumer embedded markets. Automotive remains an area of weakness.
  • Net CapEx (that is net of government incentives and equipment resale) was $2.7B in the quarter.
  • Free cashflow was surprisingly strong at $1.9B in the quarter.
  • The company did not buy back any stock in FQ3-25.
  • They expect near-term revenue growth to be weighted toward DRAM. That is a nice way of saying the NAND market is still weak.
  • Fiscal year 2025 capital expenditure remain unchanged at approximately $14B.
  • Their fiscal Q4 tax rate will be around 13%. For the fiscal year 2026 the company’s tax rate is expected to be in the high-teens percentage range because Singapore is adopting a global minimum tax.
  • Guidance does not include any impact from new tariffs.
  • They believe shipments will grow again in the fourth fiscal quarter. That means growth in bit shipments. With this growth, they expect to be “tight” on DRAM inventories exiting FY25 with the overall company inventory near their DIO target level.
  • Non-GAAP guidance for FQ3-25: revenue $10.4B to $11.0B, gross margin between 41.0% and 43.0%, operating expenses of $1.20B, +/-20M, diluted EPS of $2.35 to $2.65.

Financial Statements

Statements of Operations

The most impressive figure in Micron’s FQ3 results is the growth in revenue. The top line increased by 15.5%, to $9.30B. That is the highest quarterly revenue in the history of the company. The quarter also had the highest COGS in the history of the company. COGS rose 13.8%, almost the same percent as revenue. This is why gross margin only rose by 90 bps, to 37.7%. The previous quarterly revenue record for the company was two quarters ago. In that period, gross margin was 38.4%. Thus, Micron sold more product but less profitably on a per-unit basis. The general trend of cost in semiconductors is downward. For memory, it is around 10% a year, a little more for NAND and a little less for DRAM. That is all else equal. But Micron is shifting as fast as they can to HBM, which has a much higher cost per bit to go with its higher ASP per bit.

Considering these major factors, the market underlying these factors is mixed. They added $500M of HBM revenue in the quarter, but that is a mid-single-digits percent of total company sales, so not a large impact on pricing. Blended DRAM pricing was down by about 1% in the quarter and blended NAND ASPs declined by about 9%. I say “about” because the company doesn’t give numbers. Instead, they say “high single digits,” etc. Stop it. Just give us the number. It is a ridiculous game played by management to not do so. DRAM bit shipments were up about 21% and NAND bit shipments were up about 25%. These are shipment numbers that indicate a strong demand environment, yet ASPs declined. I’ll characterize the DRAM market in the third quarter was mixed while NAND remained weak. The company was able to ship a lot more bits with slightly down pricing in DRAM. Behind that -1% ASP trend in DRAM is an upward move from HBM and a downward move from the non-AI DRAM market. But not down by much. Customers were willing to buy a lot more product but not pay more for it. I think this is a combination of Micron’s desire to bring down their inventory and the customers’ desire to restock.

The rest of the statement of operations in the third quarter was up somewhat. Both major elements of operating expenses were quarterly records, thus operating margin only expanded from 22.0% in FQ2 to 23.3% in FQ3. The bottom line was $1.89B of net income, or $1.68 per share. The company again did not buy back any stock and the share count grew by something around one million shares. The story of the third fiscal quarter is one of marginal improvement in a healthy demand environment.

But that is not the end of the story in these results. The company guided for a large increase in revenue in the current quarter. For fiscal Q4 of 2025 the midpoint of the guided range is $10.7B with gross margin around 42% and EPS centered on $2.50. Backing into the rest of the statement of operations from those figures paints a picture of a strengthening memory market. At the midpoints, revenue will increase by 15.0% while COGS will only grow by 7.2%. I think most of that gross margin expansion of 430 bps will come from better pricing. Operating expenses will come down somewhat (R&D is lumpy for this company because of the timing of product qualifications), leading to operating margin of over 30% at the midpoint, up from 23.3% this quarter. That EPS guidance would be an increase of more than 50%. That annualizes to $10 per share in earnings, which makes the current stock price around $120 look cheap relative to the market.

Cash Flows

Operating cash flow this quarter was an impressive $4.6B. Depreciation and amortization were almost $2.1B, so that is 45% of the total cash from operations. But more than $2.9B went out in capital expenditures this quarter. Based on the company’s guidance for $14B of total CapEx in FY-25, that is $600M below the average for the year. Elsewhere in operating cash flow, receivables consumed $481M of cash while $280M of cash came out of inventories. Looking back over the last few years of cash flow by quarter, I don’t see a consistent pattern outside of capital expenditures that happens in response to the cycles. Receivables get pulled in during the worst times and let out when things are good, but it is not a perfect match. For example, two quarters ago, the company pulled in $1.15B of receivables. For the prior five quarters, they had let $880M per quarter in cash go into receivables. It could be the weakness in the memory market there prompted them to pull in some cash. I thought I would see more cash coming out of inventories, but it was only $280M, or 3% of revenue. In the most recent calls, company executives said their production would fall short of demand this year and the balance would come from inventories, to protect their market share. I’ll talk more about this in the balance sheet section.

As I said above, capital expenditures were below the average of the forecast for the year. Since we have the total for the year and the first three quarters done, we know capital expenditures in the fourth quarter will be $3.8B. Two years ago, when the memory market was bottoming out at the end of fiscal 2025, Micron was spending around $1.5B per quarter on CapEx. The average for this year will be more than twice that rate. But revenue has grown greatly as well, such that the rate of capital investment as a percent of revenue now is similar to what it was then. In total, FY-25 capacity investment will be 75% higher than FY-24. FY-23 and FY-24 were similar. In summary, the current absolute level of capital expenditures is quite high relative to the prior fiscal year. But as a percent of revenue, the picture is not that clear. One can look at the ratios of investment to depreciation and to revenue over time and justify whichever thesis you want to.

In financing activities, the company did take on an additional $770M of debt this quarter. They paid back almost a billion in debt but took on a total of $1.75B in new debt. Moving from interest consuming to interest bearing, Micron’s treasury bought $387M of securities in the quarters while cashing out $221M. I put these two together as they can give insight into the state of mind of the executive team. Adding this relatively small amount of debt says they don’t feel great about their financial position and about the market, but they don’t feel terrible either. If they felt great, they’d be paying down debt aggressively. On the equity side, management has not bought back shares in nine of the last ten quarters. For most of this time – the upturn has been going for almost two years – this has been a good decision. But this executive team has not shown a talent for taking smart risk on buybacks in the past. Because of the severity of the last downturn, they are now saddled with so much debt it will be difficult to buy back a meaningful quantity of shares when the next downturn comes. Micron increased their cash position by $2.6B, or a full third, to almost $10.2B.

The Balance Sheet

The improvement in the memory market over the last quarter really helped the company’s balance sheet. Cash and equivalents rose from $7.55B to $10.16B. That is around the largest cash pile the company has ever had. Finished goods inventory as a percent of revenue has been coming down for the last seven quarters, with a small bump up last quarter. This is off of a brutally high level at the end of the last downturn, when the company was carrying almost 50% of their quarterly revenue in finished goods inventory. But the 13.1% current level is more than 2x the lowest level seen in the last eight years. That trough was 6.2%, visited in the fourth fiscal quarter of 2021. That was the COVID upturn. That level was four to five months before the market peaked. Total DIO averaged 86 days in the quarter, down from 102 in the prior period and 91 in FQ1-25. Management said they would hold back inventory to support pricing and they seem to have done that in FQ2. The lowest level of inventory reached in the last upturn was just below 50 DIO. In the upturn before, the one driven by the emergence of data center demand, was below 39 days. Thus, there is some way to go before this measure is at maximum leanness.

Current assets are almost $28B while current liabilities are $10.1B. PP&E grew $2.2B sequentially as we are in the investment phase of the cycle. Total assets reached an all-time high of $78.4B. Long-term debt is also at a company record of over $15B. Management has more than doubled their debt load since the last cyclical peak and shows no sign of paying that debt down. Total equity is above $50B for the first time ever. Share count grows every quarter also, but not at the same rate as equity. Book value per share is now $45.11. With the stock trading around $120 per share, Micron’s price-to-book ratio is at 2.66x. This is near the high end of the historical range.

Conference Call

Numbers given during the conference call are non-GAAP. As I provide summary and commentary on the investor presentation slides above, this section will only include new information not included in the slides, plus bullet points on the Q&A with analysts.

Sanjay Mehrotra (President and CEO)

  • · The CEO, Sanjay Mehrotra, again spent the first seventeen minutes of the call reading the slides to the audience, with little additional commentary. Stop wasting our time on the call. Send out the slides ahead of the call, let analysts and investors read it themselves, and leave more time for Q&A. I think the management team does this because they don’t want more time answering questions.
  • · As Mehrotra just read the slides, my section above covers this content and my commentary on it. He did not deviate a single word from the prepared slides.

Mark Murphy (CFO)

  • · Blended DRAM pricing decreased low-single-digit percent as a result of more consumer products in the mix.
  • · Gross margin in the quarter was above the high end of the guided range because pricing was better than they expected, partially offset by more consumer bits in their mix.
  • · Their effective tax rate was lower than guided because of one-time items.
  • · Adjusted free cash flow in the quarter was $1.9B. Annualizing this and dividing by the current share price of $119, Micron has a cash flow yield of 5.7%.
  • · They ended the quarter with 139 days of inventory, a sequential decline of 19 days. Strong sequential bit shipment growth in both DRAM and NAND were behind this decline. In the prior quarter, inventory had risen 9 days, to 158 days.
  • · They refinance 2027 debt to mature in 2033 and 2036. Their weighted average debt maturity is in the year 2032.
  • · They expect to exit FY-25 with “tight” DRAM inventories and overall inventories near their target levels.

Mehrotra and Murphy spent more than half the call on prepared remarks. This is way too much. Sanjay loves to hear himself talk, which demonstrates a baffling lack of self-awareness as he is monotone and uninspiring to listen to. Write an investor letter and leave more time for Q&A.

Analyst Q&A

  • · Total HBM revenue was $18B in calendar 2024 and they are forecasting it will be $35B in calendar 2025. The company believes bit demand for HBM will significantly outgrow overall DRAM demand over the next year plus.
  • · Since Liberation Day, memory market conditions have been better than expected. Also, pricing conditions have improved throughout the quarter. It is a continued “good” market backdrop.
  • · In fiscal Q4, they will see a favorable shift in mix to more data center products. The CFO declined to provide any guidance on pricing beyond the current quarter.
  • · Today, the company’s run rate for HBM is around $6B annually, from an answer the CEO gave.
  • · By the end of FY-25 their NAND wafer capacity will be down about 10% from where it was at the end of FY-24. That is a structural change. There is also an underutilization layer on top of that. Their leading edge capacity in NAND is fully utilized.
  • · When asked how negotiations are going with customers for 2026 HBM purchases, Mehrotra gave one of his winding, long-winded answers. Of note in this is that he didn’t say anything about customer commitments to purchases in 2026. This may not mean anything, but it is a contrast to prior calls when these executives talked up how customers had committed to purchases throughout 2025. It may also be that it is too early for customers to commit to all of 2026.
  • · They believe customers may have a modest level of tariff-related purchase pull-ins.
  • · The company believes the second half of calendar 2025 will be better than the first half of the year, because inventory issues have been cleared.
  • · I was wrong in the prediction I made last quarter. I said that I thought NAND price declines would outpace higher bit shipments, leading to a sequential drop in revenue from FQ2 to FQ3. NAND revenue increased 16% sequentially. I was so wrong because, while ASPs declined by high-single digits, the company shipped around 25% more NAND bits in FQ3 over FQ2.
  • · Unlike last quarter, the analysts didn’t ask any questions about gross margin in this call.

Post Earnings Q&A call with Mark Murphy (CFO), Sumit Sadana (CBO) and Manish Bhatia (VP of Global Operations)

  • · Their view of the full calendar year 2025 for DRAM bit demand has improved. They are finally seeing long-awaited demand improvements in the distributor and industrial customer segments.
  • · Management expects positive free cash flow in the fourth fiscal quarter. This seems obvious given the strength of FQ3 and of their FQ4 guidance.
  • · In response to a question about deleveraging, the CFO shifted his answer to their net debt position being improved and their record cash position. He didn’t answer what the analyst was asking which is what the company plans to do to reduce their debt load and to buy back shares.
  • · NAND inventories improved in the third fiscal quarter, just not as much as DRAM did. Murphy said DRAM inventory will be below target levels at the end of FQ4.
  • · They feel “positive” about the trajectory of their business beyond the end of August. Here is where I remind readers that memory company executives don’t know much more than the rest of us beyond one quarter out.
  • · D4 pricing increases were helpful in Q3 and in the Q4 guide, but they are too small to be a driver of those improvements. The CBO noted that the news about better spot pricing in the DRAM market is mostly about DDR4 which makes up a small portion of their business.
  • · They believe they can expand gross margins in their November quarter. That is the first quarter of FY26.
  • · From an analyst (Tim Arcuri): more than half the dollar growth in DRAM revenue came from HBM. Price per bit is 4.5x to 5x, according to “most.” Yet the overall DRAM pricing came down in the quarter, which means a lot of the decline came from HBM. I think what Tim missed here is the different segments shift within each quarter. Sumit explained that like-for-like pricing increased in the quarter. But growth in consumer-oriented segments became a larger share of the mix, specifically mobile. Because mobile has a lower ASP than the average, when it became a higher portion of the mix, even though its pricing went up like-for-like, the blended ASP for all DRAM in the company declined. Pricing on the HBM front has been “steady.”
  • · On the other hand, the NAND market is not as healthy as DRAM. The DRAM dynamics are “very healthy.”
  • · The memory pricing conditions are “better than they expected” when they commented at the Wolfe conference in February.
  • · Their 300 bps improvement in gross margin is from two favorable mix shifts. One is to more DRAM from NAND. The other is to more data center from consumer.

ASP and Bit Shipments

The bottom of the last downturn was somewhere in the middle to the end of the summer of 2023. The uncertainty around the exact location of the nadir comes from how the bottom is measured. I often use memory company gross margin to mark inflection points in the cycle. Another way to do this is pricing. To do this for Micron, one can mark the bottom as the last quarter of declining ASPs. This is imperfect for a number of reasons. One is the resolution is one quarter. Somewhere in a ninety day period is the bottom. Another is the complexity of the memory market. The prices within different segments are related but they don’t move in lockstep. Demand in mobile is different from the server segment and from the PC segment. Also, it takes time to move supply from one segment to another. Typically that latency is a full production cycle, which is in the range of three to four months. All this is to say that marking the bottom and top of a cycle is inherently inexact. I usually narrow it down to one or two quarters. Three to six months is a wide window.

Having said all that, based on the ASP changes Micron reports with their financials, the bottom of the last cycle for both NAND and DRAM was their fourth fiscal quarter of 2023. That is the three months of June, July, and August. After that quarter, pricing started to rise. For DRAM, ASPs rose for six consecutive quarter, increasing at total of 90%. This is blended, not like-for-like. Over time this doesn’t matter, but within a given quarter, shifts in the mix of sold products will distort the view of pricing trends. This quarter is an example. Mobile outgrew almost everything else in total sales dollars. Because mobile has lower ASPs than the overall portfolio, this brought down the blended DRAM ASP for the company, even though pricing in each segment was either steady or climbing. With that qualifier, I can say DRAM pricing has risen for the last seven quarters, by a total of ~90%. Micron’s DRAM bit supply has risen a total of ~42% over the same seven quarters. That is an CAGR in the low-to-mid 20%, well over the mid-teens long term trend. That is to be expected coming out of a downturn, but it does show we are well on the way to oversupply.

NAND is an unhealthy business. The trends in pricing and supply over the same seven quarters bear this out. Micron’s blended NAND ASPs rose for four quarters, crested for one, and have been dropping for the last two. When NAND prices peaked, they had doubled off the bottom. In the two quarters since, they are off by 25%. From trough to crest, Micron’s NAND bit supply was almost flat. That is how bad things got at the bottom. Micron’s supply is up 25% over the last three quarters, since pricing peaked.

For DRAM, the small decline in blended ASPs this quarter doesn’t mark the top of the cycle. I say this because the decline was caused by mix shift to mobile. Also, management said pricing surprised them to the upside this quarter and they expect that to continue for at least the next couple of quarter. That said, the current quarter marks the eight one in this upturn. History tells us the top is close, within the next four to five months for DRAM. NAND is another story. NAND pricing has dropped precipitously for the last two quarters. Micron reacted quickly by derating their NAND capacity. The unusually short duration of this NAND cycle was caused by the emergence of substantial bits from YMTC in China. The one bright spot for NAND is the elasticity of demand in that market. Dropping prices today will stimulate demand for NAND bits and hasten the recovery.

Summary

Key financial metrics for the quarter – revenue, gross margin, and earnings – were all above the high end of the range. Both the DRAM and the NAND markets have improved more quickly than management had anticipated. This executive team always sandbags a little, so adding in the unanticipated rise in memory prices added to this and led to the big beat and raise. They are calling the current demand environment “constructive” and believe it will stay so for the rest of 2025. Of course, they don’t know much beyond one quarter out. They expect to be tight on DRAM inventory exiting the month of August. The engine of this upturn in memory is HBM growth. HBM revenue grew 50% sequentially in the quarter, which is a $6B annual run rate. As they are ramping, the current (mid-July) HBM run rate is probably around $7B right now. Recall the company raised their TAM estimate for HBM in calendar 2025 to $35B+ in their last call. Thus, $7B is around 20% of the total HBM market. Micron’s share of the overall DRAM market is 22% – 23%, so they are close to matching this today. HBM pricing in the most recent quarter has been “steady.” To me, that means the price-and-volume contracts Micron said they have with customers for all of calendar 2025 are holding. How HBM goes in the coming quarters, so goes Micron.

Elsewhere in demand by segment, Smartphone unit growth in CY25 will be low single digits, consistent with the last three quarters. This means the strength in MBU seen this quarter has been from inventory restocking/drying out at customers vs. a change in end demand. After many slow quarters, industrial demand has returned. The recent news on DDR4 and LPDDR4 will have little effect on Micron. These products are too small a fraction of Micron’s business. Overall CY25 DRAM bit demand will be in the high-teens. The company has been chasing this up for the last two quarters, meaning DRAM demand has been stronger than expected. Note, this includes HBM demand, which is driving a lot of the growth. So far this upturn (seven quarters in,) DRAM ASP have risen a total of 90% off the bottom. Bit supply has grown by 42% over that time, above the long-run average. This is expected given how oversupplied the market was prior. That said, the current quarter marks the eight one in this upturn. History tells us the top is close, within the next four to five months for DRAM. NAND demand in CY25 will be low-double-digits, the same forecast they made last quarter. That means the NAND market is as bad as they expected, but not worse. NAND is oversupplied, but not severely so. ASPs down 9% with 25% bit growth. These are shipment numbers that indicate a strong demand environment, yet ASPs declined. NAND upturn was four quarters long, then at the peak for one, now dropped the last two. Unhealthy business made worse by entrance of YMTC. I’ll characterize the DRAM market in the third quarter as mixed while NAND remained weak. The story of the third fiscal quarter is one of marginal improvement in a healthy demand environment.

Finally, a couple of comments from Micron’s financial trends. The current level of capital expenditures is high relative to the prior two years. But as a percent of revenue, it is not out of line. In other words, at a company level, CapEx is not a flashing red light saying oversupply is pending. A similar story is seen in inventory levels. Finished goods inventory of 13.1% is still 2x the lowest level seen previously. That was near the end of the COVID upturn, back in 2021. That cycle peaked about six months later. Micron’s free cash flow yield is 5.7% today, which is inexpensive compared to current market valuations. My view, which informs my current short position in Micron, is that the upturn will peak before the end of calendar 2025. Today, the duration of the current expansion phase is the best indicator I have of its remaining life.

  • S. Hughes (short MU)
1 Like