1.29.25
The company provides revenue trends by end markets for the last five quarters. This is helpful information to understand demand. Cloud revenue’s climb slowed some while client revenue declined slightly. An encouraging sign is the rebound in consumer revenue. This segment saw sales increase 14% after trending down over the prior four quarters.
The NAND downturn is underway, as declared by the 640 bps sequential drop in gross margin for flash. In the quarter ended in December, flash gross margin for the quarter was 32.5%. For the prior six quarters, oldest to most recent, flash gross margins were
(10.3%), 7.9%, 27.4%, 36.5%, 38.9%, and 32.5%. Gross margin for HDD and flash have been close in recent quarters, until this period. In the December quarter, gross margins for HDD held up, coming in at 38.6% while flash was 610 bps lower. Overall company gross margin was 35.4%. WD guided for overall gross margin between 31.0% and 33.0% in the quarter ending in March 2025. Assuming HD pricing holds up, they believe flash ASPs will deteriorate further. I said last quarter that industry average NAND prices had doubled over the prior four quarters. Zooming into just WD’s pricing, their flash ASPs bottomed out sometime in the third calendar quarter of 2023. For three quarters after that, their blended flash pricing improved. Like-for-like pricing increased for four quarters in a row. At their peak, for WD, blended flash pricing increased almost 50% off the 2023 bottom. Now flash pricing is heading down. Both blended and like-for-like ASPs saw double-digit percent declines in the last calendar quarter of 2024. I thought before that it was just pricing elasticity that caused ASPs to start dropping. I think that is part of the reason, but I now believe the bigger reason is the emergence of significant supply from indigenous Chinese NAND maker YMTC. According to Micron’s most recent release, YMTC’s markets share is in the high single digits percent.
Flash shipments and pricing were mixed. Shipments rose 9% quarter-over-quarter, following a 14% sequential rise in the prior period. Over the past two years, WD’s NAND bit shipments are up a total of 20%, well below the long-term average for market growth. Pricing is the negative side of the mixed story. Blended ASPs were down 10% in the quarter and like-for-like was down 13%. NAND ASPs only rose for four quarters with this quarter marking the first period of sequential decline. A one year upturn following an historic downturn is not enough to keep the NAND business viable. Since the pricing bottom in the third calendar quarter of 2023, WD’s like-for-like pricing crested after rising approximately 45%. This quarter’s decline brings that total increase down to about 27% since the last bottom. WD’s mix of products has yielded better pricing that the market over the last year, but in this period those two line converged. I believe the reason that pricing has started to drop despite a low level of investment in NAND capacity from the major players is being caused by the entry of YMTC into the NAND market. There were too many companies in NAND prior to this Chinese upstart. Now I am afraid the market has little hope of being viable for at least the next half a decade. It will need a major shakeout to consolidate from five participants – one of which is state-funded and has motives beyond profitability – down to three.
In their analyst call, management characterized the current pricing pressure as “short-term” and said that demand in the client and consumer segments was better-than-expected.” They said the oversupply was caused by higher fab utilization rates throughout calendar 2024, coupled with customers working down inventory. Management is either not aware of the growth in supply from YMTC or they are intentionally omitting this important factor from their commentary. Either reason is a bad sign. The company is forecasting bit shipments in the current quarter (their third fiscal quarter) to be down. They see stronger demand in data center and mobile more than offset by lower shipments to PC OEM and consumer customers. They characterized the pricing headwinds as “stronger-than-anticipated.” They always are, another reminder that the executive teams at memory companies don’t know much more than the rest of us beyond the current quarter. The company characterized this as a “new era of NAND,” with lower fab utilizations and less CapEx intensity. Management is saying there has been a structural change in the NAND market, with lower bit demand growth going forward. As for calendar 2025, management believes that inventory digestion in the first half of the year will “pave the way” for a market recover in the second half, driven by demand from AI servers and the Windows refresh cycle. I think this is wishful thinking. I believe we will continue to see ASPs decline throughout 2025 as more Chinese supply comes online. The four big NAND players will lower their fab utilizations and further cut back capital expenditures on node transitions, but it won’t be enough.
In the current quarter, management expects revenue to decline sequentially in the mid-teens percentage range caused by lower blended ASPs and higher cost per bit. They forecasted an underutilization charge of $20M to $30M in the current period. The company is already cutting back NAND supply in response to the weak market, following Micron’s lead. They expect underutilization to continue at least through the June quarter. C.J. Muse asked why they are confident in the “mid-cycle pause,” as they have characterized the current NAND weakness. Their answer was that shipment volumes remain strong, and that PC and smartphone demand that was believed to be coming now has “moved to the right a couple of quarters.” This is an important comment for the non-AI DRAM market as well, because it says PC and mobile demand is weaker than what they expected three months ago, which affects DRAM as well as NAND. Said another way, they were wrong about demand from these two segments. Why do we think they are right about a similar prediction they are making now? They see pricing headwinds next quarter but expect them to be lower than what happened this quarter. This is a key comment to revisit in three months. I don’t believe pricing turns around that quickly. The CEO spent a lot of time on the answer to the weak market question. I think he protests too much.
Before this “new era of NAND” the company was targeting 15% bit cost declines year-over-year. In the new era, they think this will be around 12% annually. Underutilization charges in the fourth quarter (April through June) will be higher than those in the current (January through March) quarter. They are pulling the throttle back on their fabs and it takes time for that change to flow through the whole manufacturing system.
– Smooth Hughes (no MU position)