Press Release and Investor Presentation Commentary
Here are the points from the press release, outside the financial statements, that I found notable:
- The sub-headline is “Increasing demand and disciplined supply improving industry outlook.” ‘Increasing demand’ is the more important part of that statement.
- Operating margin swung from 31.5% in FY22 to negative 37.0% in FY23. This is why the cyclical trade exists in Micron’s stock.
- The company’s 1-beta DRAM node and their 232L NAND node have reached mature yields. This means they have had the best yielding material at maturity, not their entire production line. I say this because the 1-gamma node isn’t going into production for at least another fifteen months. There wouldn’t be a delay of that long between a node reaching maturity and the subsequent node entering production.
- Their 1-gamma DRAM node will be in production by calendar 2025.
- “Most customer inventories” in PC and smartphone markets are at normal levels, as well as in the automotive markets.
- Data center customer inventory is also improving and will “likely normalize in early calendar 2024.”
- Total server unit shipments are expected to decline in 2023 (strong AI demand, lackluster traditional server demand.) That has not happened since 2016. The company expects total server unit growth to return in calendar 2024.
- The company will begin their production ramp of HBM3E in early calendar 2024 and to achieve meaningful revenue in fiscal 2024. That timing indicates the first meaningful revenue from HBM3E products will be late in FQ3-24.
- Micron DDR5 volume will cross over DDR4 (volume means measured in bits, vs. die, etc.) in early calendar 2024, ahead of the rest of the industry. The reason for going ahead of the industry on the conversion to DDR5 is likely to absorb some of the idled capacity. The DDR5 die takes up more wafers per bit than DDR4 does, because it has a lower array efficiency.
- PC unit shipments will decline by low double-digit percentage in 2023 and then increase by low-to-mid single digit percent in 2024.
- Mobile phone unit volume will be down mid-single digit percent in 2023 and is expected to grow mid-single digit percent in 2024.
- The industrial market showed signs of recovery in fiscal 2024.
- Calendar 2023 industry bit demand growth have increased some, from low-to-mid single digits percent to mid-single digits percent in DRAM. In NAND, their forecast for industry bit demand growth in 2023 rose from high-single digits percent to high teens % demand growth. That is about a 1000 bps growth rate increase in one quarter for NAND. I think the reason for this is the price elasticity in the NAND market. The company’s long-term CAGR expectations are mid-teens percentage for DRAM and low 20s percentage for NAND.
- For the industry in calendar 2024, they expect DRAM bit demand growth to be above mid-teens percent and in NAND they expect it to be around 20%.
- The company believes supply growth in 2024 for both DRAM and NAND will be below bit demand growth, more so in DRAM. They are forecasting an undersupplied market for next calendar year.
- The company has redeployed a portion of their underutilized equipment to ramp leading nodes faster, resulting in a structural reduction in their overall wafer capacity. They will not return to the 2022 level of overall wafer outs in “the foreseeable future.” Their underutilization of legacy nodes will continue well into calendar 2024.
- DRAM: bit shipments increased in the mid-teens percentage range Q/Q. ASPs declined in the high-single digit percent range Q/Q. Revenue rose 3% Q/Q.
- NAND: bit shipments increased over 40% Q/Q. ASPs declined in the mid-teens percentage range Q/Q. Revenue rose 19% Q/Q.
- MBU revenue rose almost 50% sequentially. Most of this looks to have come from selling NAND bits.
- GAAP guidance for FQ1-24 is revenue of $4.40B +/-$200M, gross margin of (4.0%) +/- 2.0%, operating expenses of $900M +/-$15M, and diluted EPS of ($1.07) +/- $0.07.
Statements of Operations
Revenue was up 6.8% sequentially, to $4.01B. Gross margin improved from negative 17.8% to negative 10.8%, an improvement of 700 bps. Still, operating margin was deep in the red, at negative 36.7%. Because of the amount of operating margin in Micron’s business, this was an improvement of 1020 bps. Cuts to spending were fully felt in operating expenses as well, with R&D spending down to $719M from $758M. The company made a profit of $5M in the quarter on the difference between their interest income and interest expenses. At the bottom line, Micron lost $1.43B, or $1.31 per share. The company guided for a 10% increase in revenue in Q1 with much of that falling all the way through to the bottom line. They are forecasting a loss of $1.173B in Q1, $257M better than Q4. Operating leverage is expected to expand with total opex declining from $738M to $700M. The rate of the recovery in Micron’s earnings will be determined by how fast pricing improves as they have squeezed all the cost reductions out of their business.
Free cash flow for the full fiscal year 2023 was negative $6.1B. The prior fiscal year, which contained four consecutive strong quarters, generated $3.1B of free cash flow. The downturn was so bad that Micron’s capital expenditures ($7.68B) were below depreciation and amortization ($7.76B). CapEx was $12.07B in the prior year. It was the still-large amount of CapEx spending, combined with a $1.8B inventory write-down, that made operating cash flow positive by $1.6B for the full fiscal year. More than $3.5B of cash went into inventories this year, of which the above-mentioned $1.8B was then written down. To say it another way, half of the value of Micron’s inventory growth in the fiscal year was written off. How did the company keep their business running in such a brutal market? They added $6B worth of net new debt. The company also collected $710M worth of cash from government incentives, mostly from Singapore. $504M of cash went out to shareholders in dividend payments and a further $425M was used for share buy-backs before the program was suspended. All in, Micron’s cash balance increased by about $300M, at the cost of adding $6B in debt.
The Balance Sheet
During the last twelve months, Micron squeezed almost $2.7B out of their customers, taking receivables down to $2.4B. But at the same time, inventories grew by more than $2.7B, to $8.4B. This is net of the $1.8B+ worth of inventory value that was written down. Long-term investments were cut in half, to $844M. Total assets came down by $2B during the year, to $64.3B. As I mentioned in the cash flow section, the company added almost $6.3B in total debt in order to maintain their cash balance, bringing their long-term debt balance up to $13.1B. Micron’s long-term debt balance was last in this regime in 2017, when it reached $10B. There was a strong upturn at that time which enabled Micron to pay down $6B of that balance. We shareholders are hoping for an upturn as strong as the one seen six years ago to happen in 2024 and 2025. The rest of the changes to the balance sheet mostly offset, leading to a decrease in book value from $49.9B at the start of the year to $44.1B by the end.
Numbers given during the conference call are non-GAAP.
Sanjay Mehrotra (President and CEO)
- · Revenue and gross margin were above the midpoint of guidance and EPS was above the high end of the range.
- · The company believes memory pricing has bottomed. They think prices, inventory, and profitability will improve throughout their fiscal year 2024, which began at the start of September.
- · They have reached mature yield on their 1-beta DRAM and 232L NAND. 1-gamma DRAM development using EUV is on track for production in calendar 2025.
- · PC and mobile market inventories are normal, as are inventories at automotive customers. Data center inventories are improving and are expected to reach healthy levels in early calendar 2024.
- · Data center revenue is bottoming in Q1 and the company believes this will improve going forward.
- · Micron DDR5 is expected to cross over with DDR4 DRAM in early calendar 2024, ahead of the industry.
- · PC unites will decline low double-digit percentage in 2023 and grow low to mid single digit units in 2024.
- · 2023 smartphone unit volumes will be down in 2023 and will grow by mid-single digit percent in 2024.
- · Fiscal 2023 was another record for Micron’s automotive memory business.
- · Industrial memory market showed signs of recovery in Q4 and volume is starting to recover in this segment.
- · CY23 DRAM bit demand will grow mid-single digit % range. In NAND, they raised 2023 from high-single digits to high teens % demand growth.
- · They expect “robust” Y/Y growth in CY24 for both DRAM and NAND. They expect DRAM bit demand growth to be above the long-term trends and NAND to be near the long-term trend.
- · They expect both DRAM and NAND bit supply growth to be below demand growth in CY24.
- · FY24 capex is projected to be slightly above FY23 levels but WFE will be lower in FY24 than it was in FY23.
- · Because of increased process step counts and faster note transitions, Micron anticipates their total wafer capacity will remain below 2022 levels for the “foreseeable future.” Lead times to add capacity here will have a long lead time. This is managements way of saying that their wafer capacity won’t come back quickly as demand recovers, because they transitioned fabs to more advanced nodes. This is a concern hanging over the industry as many believe that the memory makers will be able to dial output back up quickly in underutilized fabs.
Mark Murphy (CFO)
- · DRAM revenue in the fourth quarter was 69% of the total. It was 71% of total revenue for the year.
- · Their mobile BU saw revenues increase 48% sequentially. Wow.
- · Gross margin was hurt by pricing and underutilization, with previously written-down inventory flow through helping some.
- · They expect to have a few weeks of inventory above their target level by the middle of fiscal year 2024. This is bad news as that is the end of February of 2024. In other words, they expect to still have extra inventory five months from now. I thought by the end of calendar 2023 they would have reached healthy internal inventory levels.
- · Beyond Q1-24, they expect gross margin and pricing to continue to improve. They believe gross margin will be positive in the second half of fiscal year 2024.
- · Low single digit % up operating expenses in FY-24 compared to FY-23.
- · Full year fiscal 2024 will be under $200M. This is because, while the year as a whole will be loss-making, local jurisdictions will see profitability, resulting in taxes owed.
- · Capital expenditures on WFE in FY-24 will be lower than WFE spending in FY-23. Overall capital expenditures will be spread more evenly in FY-24 than in FY-23 and they will be driven by construction spending for long-term growth.
Mehrotra and Murphy again spent way too long on prepared remarks. They would better serve investors and analysts by publishing a shareholder letter rather than spending the first 40 minutes of a 60 minute call reading slides to the audience. Sanjay likes to hear himself talk. This is painful as his droning monotone is hard to listen to.
- · The first question reminded me of an observation I made about Sanjay Mehrotra in previous calls. He doesn’t thank analysts for their questions the way that most executives do. This especially stands out to me because if the experiences I have had with Sanjay from within the company. He is an arrogant jerk who is, at best, inconsiderate of others.
- · The full effect of the China restrictions on Micron’s sales are included in their Q1-24 guidance.
- · Their view of gross margin improvement has gotten better in the last three months. This means memory prices are rising faster than they expected ninety days ago.
- · ~100 bps of the sequential gross margin improvement they are forecasting for the first quarter of FY-24 is from the effect of written-down inventory.
- · Pricing has bottomed. They are forecasting a “slight improvement” in ASPs in the first quarter. For the first half of their fiscal year, they see price recovery being gradual and then being sharper in the second half.
- · “Several hundred million dollars” of revenue from HBM3e are expected in FY-24.
- · Murphy would not say that FQ-2 gross margin will be positive. He also wouldn’t say that it won’t be positive, so he left Q2 gross margin open. He did say that gross margin in fiscal Q3 will be positive and that the gross margin trend throughout FY-24 will be upward.
- · Their overall product mix will shift to more DRAM in the first half of the fiscal year. This will help margins.
- · NAND pricing will be up sequentially in the first fiscal quarter.
- · The move to more DDR5 and more HBM3e bits as a percent of total absorbs more fab equipment capacity. This lowers underutilization without increasing wafer starts. In other words, there aren’t going to be as many wafer outs across the DRAM industry when output is increased to full capacity as there was when the capacity was taken offline. The executives said again that their criteria for adding wafer capacity for DRAM and NAND is quite high, in terms of memory pricing and profitability. This is a way of saying they will be cautious to add more capacity. They want the upturn to last as long as possible.
Post Earnings Q&A call with Mark Murphy (CFO), Sumit Sadana (CBO) and Manish Bhatia (VP of Global Operations)
- · HBM is constrained today and is continued to be in short supply.
- · Of the $1.8B total inventory write-down from FY-23, $560M benefitted in Q4-23. They will see $600M worth flow through in Q1-24 and the remaining $400M in Q2.
- · Their manufacturing network will see the underutilization percentage reduce over time as their fabs are migrated to more advanced process nodes, which lowers total wafer capacity. The company will emerge with lower total wafer output capacity than they started with in 2022. This structural reduction in wafer capacity is happening across the memory industry. Thus, bringing wafer capacity back up to the prior level will require new capital investment in WFE.
- · By the second half of FY-24, they believe they will be down to a few weeks of inventory above their target level.
- · Total operating expenses for FY-24 will be $3.7B or a little below.
- · They skipped the HBM3 technology, going from HBM2e to HBM3e. Their production and revenue will be “pretty big” in the end of calendar 2024. They are targeting to reach similar market share in HBM that then have in overall DRAM sometime in calendar 2025.
- · Their view of return to positive gross margin has improved in the last ninety days. They said in their Q3 call that gross margin would be positive in Q4-24. They now forecast gross margin to become positive in Q3-24. They won’t say either way on Q2. In my experience from past downturns, companies are too optimistic at the start of a downturn and too pessimistic at the end, so I expect pricing to recover faster than Micron is modeling, leading to positive gross margin in Q2.
- · The China restrictions don’t affect smartphones in China, just data centers and networking.
- · Higher wafer need per bit for HBM comes from lower array efficiency, lower overall yield, and the need to use some production capacity for a logic die that is needed to make the HBM product.
- · Management believes price recovery in the second half of FY-24 will be much stronger than the increase in the first half.
My summary in the June quarter update was long. I’m going to keep this one shorter because much of what I said back then still applies. When the upturn comes, it will probably come faster and stronger that most expect. Internally, Micron’s view of the coming cycle has improved in the last three months. They now said gross margin will be positive in the third fiscal quarter of this year. Last update they said it would be positive in the fourth quarter. Management also left it open that it could be positive in the second fiscal quarter. Memory executives are typically too optimistic in the end of the upcycle and too pessimistic at the end of the downturn. One noteworthy point the company made is that they have been migrating their fabs to more advanced nodes at a faster pace than they normally would, because of the under-utilization. This is more capital efficient. More importantly, it takes wafer output out of the industry. Micron said they won’t return to their calendar 2022 level of wafer out in “the foreseeable future.” It will take them adding capex to do so, which they made clear they have a high internal bar to clear to do this. The cycle has passed all the milestones of the inflection – pricing turning up was the last one – but all the companies are still deep in the red. Micron’s guidance for Q1 is still for a quarterly loss of more than $1.3B. Memory pricing needs to increase by something in the range of 40-50% to reach break-even. A doubling of prices from here still only gets Micron to about $1.00 per share per quarter in earnings. Micron and their competitors need to hold tight on adding more capacity well into the regime where customers are screaming for product. The pain needs to shift to the buyers from the sellers for this industry to be healthy and viable. Micron has earned an average of $3 per share in each of the last five years. That is not a good enough return for a industry with high capital intensity and so much execution risk. I’m going to make the bold prediction that Micron’s Q2-24 guidance, which will be out in December, will be for gross margin above 5%. That is still an EPS loss but it is better than what most are predicting for the speed of this recovery.