**Q3 FY24 Earnings Release and Analyst Call**

Stocks A to Z
1

06/26/24

Press Release and Investor Presentation Commentary

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

  • All three bullet points on end markets were about AI. The company is playing heavily to the hype around AI technology and they are trying hard to position themselves as integral to it.
  • Over 80% (was 75% last quarter) of DRAM bits are on 1-alpha and 1-beta nodes and over 90% (was the same >90% comment last quarter) of NAND bits are on 176-layer or 232-layer nodes.
  • Data center server unit shipments are expected to grow mid-to-high single digits in CY24. Inventories at “most data center customers” have normalized. This is in-line with their forecast last quarter that DC inventories would reach normal levels by the end of the first half of calendar 2024.
  • In FQ3, the company generated over $100M in HBM3E revenue, at margins accretive to DRAM. They expect “several hundred million dollars of revenue from HBM in FY24, and multiple $Bs in revenue from HBM in FY25. They expect to match their overall DRAM market share in HBM sometime in CY25. I read this to mean they will match market share sometime in the late third quarter or fourth quarter of CY25.
  • PC unit volumes will grow low single-digit range in CY24. Most of their commentary on the PC segment focused on how PCs will now be “AI PCs”. The number of places “AI” was inserted into their presentation and answers to analyst questions has a whiff of desperation.
  • In the industrial segment, after qualifying the statement by saying this is a smaller part of their business, they said they are seeing some “near-term demand uncertainty from our distribution partners and end customers.”
  • Over the medium term, they expect bit demand growth CAGRs of mid-teens in DRAM and high-teens in NAND. This is a down-revision from what they said about NAND last quarter, when they forecasted low 20% for NAND for the medium term demand growth CAGR.
  • For CY24, the company is forecasting industry bit demand growth to be in the mid-teens percentage range for both DRAM and NAND. They expect CY25 industry supply growth to be below demand growth for both DRAM and NAND.
  • The closing bullet in the “Outlook” section was a signal to other suppliers that they won’t try to gain market share. They see the end of FY24 (end of August) to see Micron with low double digits percent less wafer capacity in both DRAM and NAND than what they had at their peak levels in fiscal 2022.
  • The company will continue to work down inventory to support revenue growth in fiscal 2025.
  • CapEx for FY25 will be mid-30% of FY25 revenue. This is a signal that if the market is stronger, they will invest more.
  • Construction CapEx in Idaho and New York greenfield fabs will be “half or more of the expected increase in total capex.” Cleanroom construction and HBM capacity investment will make up the “overwhelming majority of the year-over-year capex increase.” This is saying that they won’t add a lot of wafer capacity until the market gets a lot better.
  • DRAM: 69% of total revenue. Bit shipments decreased mid-single digits percent quarter-over-quarter and ASPs increased approximately 20%. NAND: 30% of total revenue. Bit shipments increased by a high-single digits percent quarter-over-quarter and ASPs increased approximately 20%. DRAM pricing increased high teens % sequentially last quarter, so DRAM price increases have accelerated. The opposite is true for NAND, where FQ2 ASPs rose more than 30% quarter-over-quarter and FQ3 prices rose 20% sequentially.
  • NAND continued from last quarter to be an overlooked strength in this recovery. The Storage Business Unit (SBU) saw quarterly revenue increase 50%, to $905M. On the other side, the mobile market is still weak, with revenue down 1% quarter-over-quarter in the Mobile Business Unit (MBU). In a market that saw ASPs rise 20% for both memory types over the same period, this indicates really weak mobile demand. The Compute and Networking Business Unit (CNBU), where HBM resides, was only up 18%, showing the continued weakness in PC and non-AI server demand.
  • Non-GAAP results were at or above (below) the high end of the guided range for revenue, gross margin, EPS, and (OpEx).
  • Guidance (non-GAAP) for FQ4-24: revenue $7.4B to $7.8B, gross margin between 33.5% and 35.5%, operating expenses between $991M and $1,021M, diluted EPS of $1.00 to $1.16.
  • GAAP guidance for FQ4-24: revenue $7.4B to $7.8B, gross margin between 32.5% and 34.5%, operating expenses between $1,040M and $1,340M, diluted EPS of $0.53 to $0.69.
  • Capital expenditures (net of grants, etc.) in Q4 were $2.06B.
  • In the third quarter, the difference between GAAP and non-GAAP net income was large. GAAP income was $332M and non-GAAP income was $702M. $367M of this $370M was stock-based compensation, either the direct expense or the tax effects. This unusually large difference is because Micron’s stock price has increases so much recently.

Financial Statements

Statements of Operations

Revenue in the third quarter was $6.81B, a sequential increase of 16.9%. I made a strongly-held prediction that the company would pull out all the stops to get to $7.0B in revenue (shipping extra inventory, etc.) I was wrong. Instead, they seem to have held back to support higher pricing, as evidenced by the 20% rise in ASPs. COGS declined in the third quarter from the second quarter, leading to an expansion of gross margin to 26.9%. Pricing was the primary driver of this higher gross margin. DRAM, which makes up almost 70% of revenue, saw ASPs rise sequentially by 20%. There was a mid-single-digits decline in DRAM bit shipments. NAND also saw a 20% increase in average selling price sequentially, along with bit shipments rising in the high-single digits range. Given the constrained supply conditions of the market, I wonder if the company is holding back inventory to support pricing. The commentary in the conference call about not reaching their inventory target for another year, in a constrained market, suggests this is their strategy. Operating expenses rose modestly, so the large increase in gross margin led to an expansion of operating margin, from negative 3.3% in Q2 to 10.6% in Q3. Back to gross margin, the increase of 1490 basis points quarter-over-quarter (using the Q2 COGS with inventory write-downs taken out) is massive. This is on top of a gross margin increase of around 2500 basis points from Q1 to Q2, and ~1100 bps from FQ4-23 to FQ1-24. In three quarters, gross margin has gone from negative 24.9% to 26.9%, mostly from improved pricing. Guidance for Q4 has a further 700 bps expansion of gross margin, which I predict will prove to be low. Interest expense and interest income rose by the same amount. Income before income taxes is a cleaner look at quarter-to-quarter performance. This rose from ($412M) in Q2 (adjusted for a one-time patent payment) to $715M in Q3. That is a gigantic swing in profit, more than one billion change in a single quarter. However, because of higher tax rates, largely driven by tax treatment of stock-based compensation, Micron’s GAAP EPS was only $0.30 in the quarter, still up from a loss of $0.37 per share in the prior quarter (adjusted for one-time items.) These two quarters are the unusual case where I think non-GAAP figures are more useful than GAAP. Non-GAAP EPS was $0.42 in Q2 and $0.62 in Q3. Share count rose by 12 million.

Cash Flows

The company generated $5.10B of operating cash flow in the first nine months of the year. D&A is still running well above PP&E, by about 10%, $5.79B vs. $5.27B. This is how it should be since the market is still not healthy and doesn’t need any new supply. Most surprising in the cash flow statement is, for the fiscal year so far, inventory has consumed $125M of cash. Typically, at this point of the cycle, memory companies are selling down their inventory into the strong demand. I think what Micron is doing is releasing their inventory in a measured way to support pricing. One reason I say this is the executives commented that they won’t hit their DOI target until well into 2025. Receivables have grown by $2.56B so far this year. Even with low capital expenditures, all of Micron’s operating cash flow so far this year, and then some, has been consumed by PP&E. The memory market really needs to be healthy for a couple of full years to make it all worth the trouble. They have paid down $815M of debt so far this year. Between capital expenditures, not turning any inventory into cash, growing receivables, paying down debt, and paying their dividend, the company’s cash position is down by around one billion dollars in the last three quarters. This is not a cause for concern as the market is seeing a strong recovery. While I was wrong with my $7B in revenue prediction for this quarter, The forecast I made six months ago that FQ3 would be free cash flow positive turned out to be correct.

The Balance Sheet

Micron has about $8.3B in cash and equivalents against $13.3B in debt. Current assets and current liabilities are both up by two billion dollars, so a wash in change to total value so far this fiscal year. In fact, the whole balance sheet has gone sideways in the first three quarters of FY24. This is not surprising given that these three quarters straddle the nadir of the market. Receivables have gone up faster than inventories. Within those inventories that haven’t changed much, finished goods have actually come down from $1.616B at the end of August 2023 to $1.280B at the end of May 2024. Work in process has increased by more than $400M and raw materials are up some. Hence, the net increase in inventory is from higher WIP offsetting declines in finished goods. The memory market has turned but Micron’s balance sheet doesn’t have much to show for it yet. Book value is flat from three quarters ago and they still have net debt of $5B. All the debt they added in the last two years may protect this executive team from their poor capital allocation skills. They will hopefully take the coming torrent of free cash flow and pay down their debt instead of repurchasing their equity at up-turn prices.

Conference Call

Numbers given during the conference call are non-GAAP.

Sanjay Mehrotra (President and CEO)

  • · They continue to expect prices to increase through the rest of calendar 2024.
  • · DRAM front-end cost reductions in FY24, excluding HBM, will be high single digits and FE reductions for NAND will be in the low teens.
  • · Near term bit demand growth from the PC and mobile segments will be flat. But because of strong AI-driven demand from the data center segment, the company still expects price increases to continue throughout calendar 2024. I wonder if their guidance being in-line with expectations instead of above them is because mobile and PC demand is still not growing. If it is, and these segments were to surprise to the upside, it would drive larger memory price increases.
  • · They signed the CHIPS act grant agreement with the U.S. government to build fabs in Idaho and New York. Construction of the Idaho fab is underway.
  • · Data center inventories at most customers have normalized. PC and smartphone customers have built additional inventories in anticipation of tight supply.
  • · They generated more than $100M of revenue from HBM3E in FQ3 at margins accretive to DRAM. They expect to generate “several hundred million dollars” of revenue from HBM in FY24. “Several” is defined as more than two and fewer than many, so this means they will have at least $200M of HBM revenue in FQ4, assuming that HBM3E is their only meaningful source of HBM revenue.
  • · HBM is sold out for 2024 and 2025 with pricing already “contracted” for the “overwhelming majority” of the 2025 sales. They reiterated their market share in HBM will match their overall DRAM market share sometime in 2025.
  • · PC unit growth in calendar 2024 is on track to rise low-single digits range. Micron believes the PC replacement cycle will begin near the end of 2024. I think this prediction is neither accurate nor is that event meaningful. When the replacement cycle does start, it is diffused over time.
  • · Data center SSD demand is strong, Micron is gaining market share, and customers in this space have worked through their 2023 inventory.
  • · Smartphone unit volume growth in calendar 2024 remains on track to grow low-to-mid single digits percentage range.
  • · Calendar 2024 bit demand growth will be mid-teens percent for both DRAM and NAND. They expect supply to be below demand for both DRAM and NAND. Here Mehrotra gave a long explanation of how tight supply needs to continue to support more investment in capacity and R&D. I believe this is a signal to the rest of the industry that the current ASP levels are not sufficient, and Micron will not add capacity until prices rise more.
  • · Micron’s bit supply growth for FY24 remains below demand for both DRAM and NAND. This isn’t much of a prediction given there are only two months left in FY24. They said again here that they will “continue to exercise supply and CapEx discipline and focus on improving profitability while maintaining our bit market share.” This is telling Samsung and Hynix to be careful investing in capacity.
  • · Mehrotra said something I have not heard from a memory company before at this point in the cycle. He said Micron will “use our existing inventory to drive a portion of the bit growth supporting our revenue in fiscal 2025 to enable more optimized use of our CapEx investment.” He is suggesting that they will bias towards squeezing down their inventory rather than adding wafer capacity, to meet revenue in FY25. This is a way of saying Micron will have good revenue growth but will not do so by adding a lot of new wafer capacity. He is hoping investors will give Micron’s equity price the benefit of strong sales and earnings growth without the concerns about oversupplying either memory market.
  • · They will raise CapEx materially next year, to around mid-30% range of revenue in FY25. The CEO qualified this increase in CapEx by saying that at least 50% of the increase in CapEx from FY24 to FY25 will be cleanroom construction in ID and NY. Further, between HBM and cleanroom investment, the “overwhelming majority” of growth in year-over-year CapEx will be from these two sources. The first comment is meaningful to me but the second is less so. HBM investment could be wafer investment, which is capacity investment. But HBM has a higher percentage requirement of assembly and test investment needed than DDR5, NAND, etc., so it is still worth noting.

Mark Murphy (CFO)

  • · The 800 bps increase in gross margin was mostly from pricing, with some help from product mix and cost reductions. There was no impact from prior inventory write-downs in FQ3.
  • · Excluding the effects of previously written-down inventory, the sequential increase in gross margin was 1500 basis points, from 12% to almost 27%. This is a massive increase in gross margin
  • · Operating expenses were at the low end of the guided range. EPS was above the high end of the guided range.
  • · They paid off $650M of debt that was to mature in 2025. The company returned to profitability
  • · For the fourth fiscal quarter, DRAM bit shipments to be flattish and NAND bit shipments to be up slightly.
  • · HBM growing as a percent of Micron’s total portfolio will be accretive to gross margin.
  • · Consistent with what the CFO said last quarter, their non-GAAP tax rate in FY25 will be in the mid-teens percent range. They expect fiscal Q4 tax expenses to be approximately $320M. I do not recall Micron giving a specific number for tax expense guidance in the past. I think they did this because of the size of their tax expenses, caused by stock-based compensation.
  • · DIO will approach their target by the end of FY25.
  • · CapEx in FY24 will be $8.0B and free cash flow will again be positive in the fourth fiscal quarter.
  • · Market conditions are improving and supply tightness on the leading edge continues. It is telling that they are excluding the lagging edge nodes from this comment.

Mehrotra and Murphy spent 34 minutes in this call on prepared remarks (was 35 minutes last quarter.) This is still way too much. Write an investor letter and leave more time for Q&A.

Analyst Q&A

  • · HBM is accretive to DRAM margin, not just overall margin, for the company.
  • · Gross margin in the first fiscal quarter of FY25 (ending in November) will be up “a few hundred basis points.”
  • · “Several hundred million dollars” of HBM in FY24 and “multiple billions” of HBM revenue in FY25. They are staying with this party line, same as last quarter.
  • · In response to an excellent analyst question on how fixed the pricing is in the HBM sales contracts, Sanjay gave a non-answer that never visited the neighborhood of useful information.

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

  • · The new fab in Idaho will not provide meaningful supply until 2027. DRAM supply growth from Micron will come from bit growth from technology transitions at their DRAM fabs in Taiwan and Japan.
  • · They see inventories continuing to be worked down in FY25 and they will be lean on inventory in 2026. This is bold to predict what will happen in the market more than a year out. I think they are trying to hold back inventory going into this upturn to try and get pricing to stay higher for longer.
  • · Sadana said they are almost done setting pricing for FY25. He didn’t say anything useful for what the gross margin, etc. will be. He did say as they improve their yields the HBM gross margin will continue to get better.
  • · The analysts on the call kept asking about the pricing agreements for HBM. Micron is promoting these in a way that suggests they have pricing security, which is something they have not had before. However, they have stopped short of saying that outright, which makes me believe their customers aren’t committed to pricing.
  • · OpEx will be up mid-single-digits from Q4 to Q1.
  • · They have seen a strengthening in data center demand recently, principally from AI. Demand is also increasing in traditional servers. Data center inventory has reached normal levels.
  • · They define high cap DIMMs as anything that is 96GB or higher.
  • · Mobile unit growth in the first calendar quarter of 2024 was a little better than they expected, but they are staying with their prior full year forecast of low-to-mid single digits percent growth in this measure.
  • · They can maintain their market share in DRAM and NAND with planned investments in Taiwan, Singapore, and Japan. The greenfield capacity coming from ID and NY is not needed until the latter half of the 2020s.
  • · Part of how Micron will maintain their bit share next year (I think fiscal year) is by selling down inventory. I think one reason they are doing this inventory strategy is they can better modulate market supply in response to demand than if they were to support revenue growth purely with added capacity. If so, this is a smart strategy. The risk is that Samsung and Hynix take market share by adding wafer capacity and Micron can’t respond fast enough. The risk of this is somewhat mitigated by always having empty cleanroom space ready for WFE, when the need arises.
  • · One answer that the CBO provided indicated that their long-term supply agreements in FY24 and FY25 contain prices that are fixed ahead of time. I have been dubious on how fixed these prices are, but the more they say it (and risk lawsuits if these statements turn out to be misleading/untrue) the more I tend to believe they have locked in prices for HBM. Sadana said the pricing they are setting is based on their assumption that the overall DRAM market will be constrained and see rising prices for the foreseeable future.
  • · DRAM bit cost declines in Q3 were flattish sequentially. NAND was down several percent sequentially. For FY25, DRAM costs will be “down only modestly” because of the higher cost from more HBM in their mix.
  • · The mid-30% of revenue range given for FY25 CapEx is net of grants from the CHIPS program, Singapore, etc. Thus, the gross CapEx could be higher.
  • · FY25 WFE capex will be up compared to FY24. The executives declined to provide more specific guidance than this.

Summary

After the release of FQ2-24 results three months ago, I felt overwhelmingly positive about the rest of 2024 for Micron. The reason for this is the market seemed to me to be clearly heading for severe undersupply. Here we are three months later. The memory market has strengthened. ASPs for both DRAM and NAND increased by 20% from Micron’s FQ2 to FQ3. This is following a 20% rise in DRAM and a 30% rise in NAND in the prior quarter. Gross margin is up 5180 basis points over the prior three quarters, from a bottom of negative 24.9% in the fiscal quarter from September through November of 2023. Micron’s stock price has gone from the mid $90s before the FQ2 beat-and-raise to a high of $157.54 in mid-June. MU has been buoyed by the excitement around AI, combined with the strengthening memory market. My positivity on Micron from here has been tempered by these results, because of their guidance for FQ4. I thought they would be over $7B in Q3. Instead, the company seems to have held back inventory in favor of higher pricing. If this is the strategy, then I am surprised they don’t see stronger revenue in the current quarter, which ends on Judgment Day, August 29th. The midpoint of guidance provided works out to a ~15% increase in blended ASPs from Q3 to Q4. While that is healthy, given the growing undersupply in the memory markets that I see across calendar 2024, plus Micron’s hold-inventory-for-better-pricing strategy, ASPs should rise more. Maybe they are sandbagging. Maybe they really do believe prices won’t grow more than 15%. Executives typically under call the strength of the memory recovery. If I put revenue of $8B in my model for Q4, which would be $200M over the high end of guidance, I get to a blended ASP increase of 20%. But the lack of a strong beat in Q3 makes me hesitate to predict a stronger Q4. Yes, if the market tightness is increasing across 2024 as I believe it will, then ASPs rising 20% for a third quarter in a row is not unreasonable. Yet, my confidence is shaken by the lack of a stronger beat and the weaker-than-expected forecast. Also, the conference calls had a lot of AI marketing commentary. Too much. The executives are clearly leaning fully into the AI connection to Micron. The tone had a whiff of desperation. While HBM is a revolutionary product, a year or so from now there will be plentiful supply of similar offering from Samsung, Hynix and Micron. HBM is not going to make Micron a single source of one type of AI hardware the way Nvidia is for GPUs. AI workloads are creating what will be another large segment of the memory market a few years from now, probably comparable in size to PC, mobile, and non-AI data center today. But in that coming world, Micron’s HBM product will be just another SKU, and HBM bits will be cyclical in their gross margins just like the rest of the DRAM bits. Micron’s inventory strategy is worth noting. They seem to be tacking more to counting on inventory to meet demand growth in the next twelve-to-eighteen months vs. adding more manufacturing supply. I think this is a good strategy. It carries inventory risk, but that is low at this point in the cycle. Finally, the analyst community seems to be taking Micron at their word that pricing for these HBM contracts in 2025 and 2026 is fixed, as are the volumes. My experience with LTAs in memory is that, if the market goes into oversupply, the customers can just get a lower price from Micron. In other words, if the market is stronger than expected, Micron’s customers will hold them to the lower contracted prices. If the market is weaker than expected, Micron will have to lower the price down from what is in the contract. Micron’s CBO said these HBM contracts are different than memory LTAs in the past, but I am still doubtful that Micron has locked in take-or-pay contracts for all their HBM through 2025 as they are suggesting. And even if they have, let’s remember that, even through FY25, Micron is still mostly a non-HBM memory company. Even at the “multiple billions” of HBM revenue they are saying will be had by the company in FY25, that is still $2B to $3B in a company with $25B in revenue in FY24 and perhaps $35B to $40B in total revenue in FY25. At the high end, that is about 10% of Micron’s total FY25 revenue coming from HBM. Even with take-or-pay contracts on HBM at high gross margin pricing, Micron’s FY25 financial performance will mostly be determined by how the NAND and non-HBM DRAM markets perform.

-S. Hughes (cyclical long MU)