Q4 and FY24 Earnings Release and Analyst Call

09/25/24

Press Release and Investor Presentation Commentary

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

  • Management said that FY-25 will be a “substantial revenue record,” up from just a “revenue record” in prior commentary.
  • FQ4-24 revenue was at the high end of the guided range while gross margin and EPS were above the high end of the range.
  • The company continues to lead the industry in both DRAM (1-beta) and NAND (G8 and G9) process technology.
  • DRAM front-end cost reductions for the fiscal year 2024 were at the high end of the outlook provided at the beginning of the year (high-single digits %.) NAND FE cost reductions were consistent with their forecast (low-teens %.)
  • FY-25 DRAM front end cost reductions, excluding HBM, will be in the mid-to-high single digits percentage range.
  • FY-25 NAND cost reductions (they didn’t specify FE) are expected to be in the low-to-mid teens percentage range.
  • Their new fab in Idaho is under construction and they are in the permitting process in New York. The assembly and test facility in India is under construction, as is the AT expansion in Xi’an China.
  • The company acquired an LCD factory in Taiwan which they will convert into a test facility.
  • Overall server unit shipments are expected to grow mid-to-high single digits in calendar 2024. This is a mix of AI and traditional servers. They didn’t quantify AI server unit growth but did say traditional servers will grow low-single digits.
  • FQ4 HBM gross margins were accretive to overall company DRAM gross margin.
  • HBM TAM will grow from $4B in calendar 2023 to $25B in calendar 2025. As a percent of overall industry DRAM bits, HBM is expected to grow from 1.5% in calendar 2023 to 6% in calendar 2025.
  • 12-high HBM3E output will ramp in early calendar 2025 and mix of this product will increase throughout the calendar year.
  • PC unit volumes will grow low single-digits in calendar 2024. The company expects unit growth in PCs to accelerate into the second half of calendar 2025 as the PC replacement cycle increases pace and AI PCs are introduced. This is wishful thinking.
  • Mobile customer inventory dynamics are similar to that of PC customers. Smartphone unit volumes will grow low-to-mid single digits percentage in calendar 2024. They expect unit growth to continue in 2025, which is a soft way of saying the smartphone market will continue to be weak in calendar 2025.
  • The auto industry has an overhang of memory inventory. In another example of wishful thinking, the company expects resumption in their automotive growth in the second half of fiscal 2025, which is next spring and summer.
  • The demand outlook for the DRAM industry in calendar 2024 has improved, driven by strength in data center servers. Demand in other markets is unchanged. The company has upgraded their expectation for calendar 2024 industry DRAM bit demand growth to be in the high-teens percent range. This is up from mid-teens percent last quarter.
  • Their view of NAND industry bit demand growth in calendar 2024 is unchanged, in the mid-teens percent range. This is unchanged from last quarter.
  • In calendar 2025, both DRAM and NAND industry bit demand growth are both expected to be around the mid-teens percent range.
  • On the supply side, they expect industry wafer capacity in both DRAM and NAND in calendar 2024 to be below 2022 levels, with NAND “meaningfully so.”
  • The company expects a healthy industry supply-demand environment for both DRAM and NAND in calendar 2025. For DRAM, this statement is based on HBM demand drawing bits away from other segments. Continued low investment in node migration is the root of health for NAND in 2025.
  • NAND technology transitions generally provide more annual bit growth than the high-teens demand CAGR. Consequently, they anticipate longer periods between NAND industry tech transitions ad a moderation of capital investment in NAND.
  • Fiscal 2024 CapEx is expected to be “meaningfully higher” than the $8.1B of CapEx spent in FY-24. They expect FY-25 CapEx to be around mid-30s percent of revenue. The “overwhelming majority” of the year-over-year CapEx increase from FY-24 to FY-25 will be in greenfield fab space and HBM capacity.
  • The fabs in Idaho and New York will not contribute DRAM bit supply until at least FY-27.
  • The company closed the CEO section by saying they will focus on profitability when considering CapEx investment and will walk away from less profitable business, building inventory to do so. They ended by saying they are committed to maintaining their market share in DRAM and NAND.
  • DRAM: 69% of total revenue. Bit shipments flattish quarter-over-quarter and ASPs increased mid-teens %. NAND: 31% of total revenue. Bit shipments increased by a high-single digits percent quarter-over-quarter and ASPs increased high-single digits %. DRAM pricing increased 20% in the prior quarter and high teens % the quarter before, so the rate of DRAM ASPs rising slowed last quarter but was still surprisingly strong. The pace of NAND percent increases has steadily come down over the past three quarters. Those ASPs rose 30% in FQ2, 20% in FQ3, and then high single digits % in FQ4.
  • The Storage Business Unit (SBU) saw revenue rise 24% sequentially, supporting the statements from management that they are gaining share in enterprise SSDs. HBM is sold through the Compute and Networking Business Unit (CNBU) and their revenue rose 17% sequentially to $3B. Even in a smaller pool of revenue, Micron’s HBM is still not a large percent of revenue. The Embedded Business Unit (automotive, most notably) was weak, clocking a 9% sequential decline in revenue. This market is at least two quarters away from recovery, based on other statements from the company.
  • GAAP guidance for FQ1-25: revenue $8.5B to $8.9B, gross margin between 38.5% and 40.5%, operating expenses between $1,070M and $1,100M, diluted EPS of $1.66 to $1.82.

Financial Statements

Statements of Operations

Revenue rose by 13.8% sequentially, to $7.75B. This was near the high end of the guided range, the upper bound of which was $7.8B. Almost all of this higher sales volume came from increased pricing. Management stated that DRAM ASPs rose mid-teens percent sequentially and NAND ASPs rose high single digits percent. The split is 70-30 between the two memory types, so this averages to a low-to-mid teens percent increase in blended ASP. COGS grew slightly in the fourth quarter compared to the third, rising 0.7%. Gross margin expanded by 840 basis points, from 36.9% to 35.3%. The company guided for GM to rise around 300 bps in the current quarter, which would continue a slowing trend of gross margin increases. The shape of the gross margin curve over the next three or four quarters will have a strong effect on the stock price from here. The company strategy of holding back inventory to constrain supply and support higher pricing is working so far. Gross margin dollars rose by a third quarter-over-quarter, to $2.74B. Higher R&D, driven by the reinstatement of some bonus programs and the timing of new product qualifications, caused OpEx to rise by almost 10% sequentially. There is a lot of operating leverage in this business. Operating income doubled, to $1.52B. Lower debt and lower interest rates brought both interest income and interest expense down in the quarter. The company again provisioned a large percent of income (41.3%) for income taxes in the quarter. With a normal tax rate (15%,) net income for the quarter would have been around $1.28B. Instead, it was $887M. GAAP EPS was $0.79. The company bought back 3.2 million shares of stock in the quarter, yet the share count only declined by about 1 million, keeping the total effectively flat from 90 days ago. Buybacks at the level they were done this quarter are mostly symbolic given the rate of shareholder dilution from stock-based compensation to employees.

Cash Flows

For the full fiscal year, the company generated $8.51B in operating cash flow, yet the total cash position on their balance sheet declined by $1.60B. The difference starts, as it almost always does with this company, on the line for expenditures for PP&E. Total capital expenditures were $8.39B in the year compared to $7.78B of depreciation and amortization. The fourth fiscal quarter was the crossover point in this cycle when CapEx went higher than D&A. Management says most of this is for construction of new facilities and for HBM. I think this is mostly true, but that HBM category may be misleading. Investors are excited about HBM because it is “AI memory” so they are likely to see investment in more capacity here as a positive. While one category of this spending is on assembly and test equipment, it also includes some amount of WFE. How much management categorizes as HBM is up to them. Since leading edge front-end wafer capacity is needed for HBM products, it is true that investment in node transitions to 1-beta DRAM is HBM CapEx. The takeaway investors should draw from this is when management says they are investing in HBM, recognize that some amount of this is in capacity to make more bits via tech node transitions. Receivables rose by $3.58B over the year while AP was a source of $1.92B in cash. Second in noteworthiness behind CapEx is inventory, which consumed $488M of cash in the fiscal year. $363M of this came in the fourth quarter alone, which confirms the company is executing a strategy of holding back inventory to support pricing. Inventory at memory companies is usually declining at this point in the memory cycle. I like this strategy and shows a willingness to take risk that this management team has been woefully short of in the past. In the fourth quarter, the company bought back $300M shares of stock and paid off around $80M of their debt. The company really needs to make a lot of hay in the coming few quarters. The memory market turned around nine months ago and yet the company is still generating little free cash flow. For the full year, which includes about four months of the end of the downturn, Micron had free cash flow of only $121M.

The Balance Sheet

In the last three months of fiscal year 2024, Micron’s book value increased by around $900M, to $45.1B. Receivables rose by $1.5B and inventory increased around $360M. I will have to look at the 10-Q when it is released to see the distribution of this change in inventory between the three categories. Cash and equivalents declined a couple hundred million dollars and stands at $8.1B. Total debt is now $13.4B, most of which is long-term debt at fixed rates. As I said last quarter, I hope the need to pay down this large amount of debt saves this management team from the poor stock buyback timing they have displayed during past upturns.

Conference Call

Numbers given during the conference call are non-GAAP.

Sanjay Mehrotra (President and CEO)

  • · Data center demand is exceeding their leading-edge supply and driving overall healthy supply-demand dynamics.
  • · They are now calling their 232-layer NAND node the “G8” node. The next one will be called the G9 node. Just a nomenclature change.
  • · Industry server units will grow mid-to-high single digits in calendar 2024. Traditional servers will see low-single digits growth.
  • · FY-25 revenue from HBM will be “multiple billions of dollars” in revenue.
  • · Even with higher DRAM gross margins, FQ4 HBM gross margins were accretive.
  • · Their HBM is sold out for calendar 2024 and 2025, with pricing already set.
  • · The company believes they have gained market share in data center SSDs. They have achieved a quarterly record revenue in data center SSDs of over a billion dollars.
  • · PC customers have built inventory anticipating higher pricing and lower supply. They expect PC customer inventories to reach healthy levels in the spring of 2025.

Mark Murphy (CFO)

  • · Higher pricing and improved product mix were behind the rise in gross margin.
  • · R&D program expense increases drove the sequential rise in operating expenses.
  • · Taxes were higher than guidance because of a jurisdictional shift in earnings.
  • · They repurchased shares in the quarter at an average price of $93 per share.
  • · The company expects a healthy supply-demand environment in FY-25. They plan to draw down inventory over FY-25 to meet demand needs while controlling investment.
  • · Gross margin increasing to 39.5% in FQ1-25 is expected to come from better pricing and improved product mix.
  • · For full year FY-25 they see growth in OpEx to be mid-teens % over FY-24.
  • · Their non-GAAP tax rate in FQ1-25 will be mid-teens.
  • · FQ1-25 CapEx will be $3.5B. Total CapEx in the year will be mid-30% of total revenue.

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

Analyst Q&A

  • · DRAM bits are expected to be “somewhat higher” sequentially from Q4 to Q1. NAND bits will be flattish over this same period. They have increased their bit shipments assumption for DRAM higher a couple of times over the last month.
  • · The executives declined to give more detail than “several hundred million” in HBM revenue in the fourth fiscal quarter.
  • · DRAM cost in the first quarter of FY-25 will go up slightly with the ramp of HBM.
  • · The executives strongly believe demand will continue to exceed supply throughout fiscal 2025. They plan to run down demand over the next twelve months to close the gap between product levels and market demand.
  • · To a question about the possibility of HBM becoming oversupplied in 2025, Mehrotra responded that leading edge nodes are tight on supply. This is where HBM bits come from. This isn’t really a fair question directly. It is more the analyst fishing for any information the executives will disclose. He reiterated that Micron expects a revenue record in FY-25. Mehrotra is good at talking a lot and not saying much. It is good to remember that memory industry executives don’t know much more than the rest of us do beyond one quarter out.
  • · They expect the HBM sales to continue to be accretive to overall margin in FY-25.
  • · Data center demand strength is behind their increase in estimated DRAM bit demand in calendar 2025. They are not reflecting a strengthening in PC or mobile, beyond what they had loaded a month or a quarter ago.

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

  • · The decline of inventory they expect across FY-25 they expect to be second half weighted. This increases the risk of the strategy because they are holding inventory longer and thus exposing themselves to more uncertainty.
  • · Demand continues to be strong from both AI servers and traditional data centers.
  • · WFE CapEx will be “up a bit” from FY-24 to FY-25. Most of the spending growth will come from HBM capacity and facility construction.
  • · Chinese memory manufacturing that has come online are making products at the low end of the market. Micron aims to keep their market share stable over time but grow their fraction of the more profitable products. This is an indirect way of saying the Chinese memory entrants are making gains in parts of the market that Micron is least interested in and that are shrinking over time.
  • · Their HBM contracts have prices determined for all of calendar 2024 and 2025. I am still not convinced this means the volume and pricing (and thus revenue) for all of Micron’s HBM through the end of calendar 2025 is locked, in the form of a take-or-pay with their customers.
  • · Overall bit volumes will be second half weighted. The second fiscal quarter (December through February) is the seasonally weakest quarter of the year.
  • · The managers on the call hinted heavily that their second fiscal quarter is the seasonally weakest quarter of the year. They also reminded the audience that some of their customers are bringing down their inventory over that same period, which says indirectly that this will contribute to weaker demand in late calendar 2024 and early calendar 2025.

Summary

As far as guidance, this quarter was the opposite of the FQ3 results. Three months ago, Micron disappointed investors with weak guidance and the stock dropped as a result. Since then, shares are off from the low $140s right before the FQ3 release, to the mid-$90s just before the FQ4 announcement. They made a couple of visits around $86 on the way there. This result was a beat on margin and earnings and at the high end for revenue. The surprise was in guidance for next quarter, at $8.8B of revenue. The Street was expecting something much lower and the stock was up to nearly $115 at one point before settling down to close just below $110. There has been a lot of pessimism around weak demand in PC and mobile markets. These results did nothing to change that view. You can also add a weak automotive market to the group of bad segments. Despite this weakness, DRAM ASPs were still up mid-teens % on flattish sequential shipments. This is Micron’s strategy of holding back inventory to get better pricing. They believe this will work because their ramp of HBM is taking wafers and bits away from other segments. Even though demand in almost all segments that aren’t data center remains weak, the reduced supply from holding inventory and lack of bit growth in manufacturing is supporting non-HBM DRAM pricing to a surprising degree. NAND finally showed some weakness in the fourth quarter, with a high-single digits percent ASP rise on a high-single digits sequential increase in NAND bit shipments. One quarter in, the strategy of holding inventory seems to be working. Non-GAAP gross margin was 200 bps over the high end of the guided range. If they overperform by an equivalent amount in FQ1-25, non-GAAP gross margin will be 42.5% for that period. One comment worth noting is Micron’s management team believes the pace of node migrations across the industry will slow in the coming years. The bit growth from tech node migrations alone is more than demand growth. That is the reason they gave. NAND continues to be a bad business. While there is optimism for the future with the strong guidance for FQ1-25, the executives mentioned twice that their second fiscal quarter (December through February) is the seasonally weakest of the year. Sounds like they are prepping investors for a softer guide in three months. AI data center demand continues to carry water for all of memory, yet the improved pricing Micron is seeing is almost all from non-HBM memory ASP increases. As HBM ramps and becomes a larger percent of total sales, we should see gross margins expand further. If we can get some strength from other segments, all the better.

-S. Hughes (cyclical long MU)