8.14.25
So far the prediction by SanDisk executives that the recent weakness in the NAND market is a “mid-cycle pause” is coming to pass. Revenue of $1.9B this quarters was $100M above the mid-point of guidance. Gross margin expanded by 370 bps, to 26.4%. The last eight quarters of gross margin have been (10.3%), 7.9%, 27.4%, 36.5%, 38.9%, 32.5%, 22.7%, and 26.4%. There was a peak three quarters ago, two quarters of decline, and now one quarter of recovery. One data point does not make a trend, but they are guiding for continued growth in their fiscal first quarter of 2026. Specifically, revenue will be $2.1B to $2.2B. Gross margin between $28.3% and 29.2%. Those are GAAP numbers. The company did not provide GAAP EPS guidance, but their non-GAAP guide is a range of $0.70 to $0.90. If this guidance happens, we can say the NAND market is recovering. The company sees favorable supply-demand dynamics in the NAND market right now.
Revenue increased across all three of SanDisk’s business unites. Cloud, the smallest, was up 8%. Client (the largest) rose by 19%. Consumer sales increased 2%. The strength of this quarter was driven by the client business. That segment’s revenue is larger than the other two combined, by a significant margin. Management said there is “clear momentum” with hyperscalers to meet AI-driven demand. Hyperscaler CapEx is growing 47% year-over-year to $368B. They didn’t specify but this seems to be referring to calendar 2025 over 2024. SanDisk is the first company to credit demand from AI-enabled PCs as one of the reasons for market strength. They also cited the refresh cycle for smartphones as well as average capacity increases and Windows 10 EOL. The memory companies have been saying all these sources of demand are coming for well over a year. I wonder if it is happening now. Within their consumer segment, seasonal demand was the first-bullet reason for strength.
As I said last quarter, boo to the SanDisk investor relations for no longer breaking out blended and like-for-like pricing. Total bit shipments were up mid-single digits. ASPs also rose mid-single digits. Recall last quarter management gave pricing guidance, something that is almost never seen from memory companies. Back then, they said prices would rise “mid-to-high single digits,” which in interpret to mean ~7.5%. The mid-single-digit (5%) rise they actually saw was lower than their guide, but not by a lot. On the other side, bit shipments were higher than they forecasted a quarter ago when they said bit shipments would be flat. The rise in ASPs this quarter of ~5% breaks a streak of three consecutive quarters of price declines. Among the four NAND manufacturers I follow (I don’t analyze YMTC because they don’t report reliable results, and they are the smallest player in the market,) SanDisk is the only company that saw NAND prices rise this quarter. Samsung was the closest to positive pricing, at minus 2.5%. Both Micron and Hynix saw their prices drop by almost 10%. Given that Client is the biggest segment for SanDisk, the rise in prices may be a result of outsized gains there. Of course, client is the biggest segment of the overall market. Point is, SanDisk’s ASP rise is an outlier and likely has more to do with their product mix this quarter and seasonal strength than about a recovery in the NAND market. But their strong guidance is contrary to this. On bit growth, going back to the end of the last clear downturn (CQ1-23), SanDisk has lost the most market share to the other three non-Chinese NAND companies. Over that same period of nine quarters, their ASPs are also about flat, which is the smallest change among those same four NAND companies. So SanDisk has both lost market share and had the worst pricing performance. That might explain why they no longer provide like-for-like pricing. When you are underperforming in your business, take information off the scoreboard.
This section is commentary from management’s conference call with analyst.
· During the quarter, inventory was reduced from 150 days to 135 days.
· Overall demand exceeded supply in the quarter. The company again made the bold prediction that this situation will persist through the end of calendar 2026. Given that memory executives don’t know much beyond one quarter out, this is borderline insane.
· Prepared commentary intimated that management recognized that they are below their average market share in the cloud/data center segment and that they intend to grow in those markets at a rate above their competitors.
· The company has created a product called High Bandwidth Flash, or HBF. This is the only non-DRAM memory company outside China looking over at all the fun the Big Three are having with HBM and inventing their own version of this. NAND doesn’t have the same effect on system performance from a bandwidth bottleneck that DRAM does. Good luck to them anyway.
· Revenue was up primarily because of better-than-expected bit growth. Last quarter, their guidance assumed flat bit shipments and mid-to-high single digit price increases. What actually happened was they shipped more bits than they expected to but prices were a little lower than they forecasted.
· They took $51M in fab underutilization charges in the quarter.
· The company did a 200 person reduction in force during the quarter.
· Several of their products are on allocation and they expect mid-single-digit undersupply throughout FY-26.
· The increase in revenue in their guidance is half from more bit shipments and half from better pricing. That is similar to what happened this quarter. Higher gross margin is primarily from better pricing.
· There is an extra week in the current quarter. Conveniently, the CFO didn’t mention this until he got to the expenses in the guidance.
· The company claims to be moving from a market of pricing elasticity to one driven by varying supply to match demand and by differentiated pricing. They are saying they hope to manage supply better as an industry and to create differentiated products to de-commoditize the NAND industry. Executives have been trying to do this almost since NAND was invented. Good luck.
· Management sees demand growth in calendar 2025 to be very low double digits (I hear 11%) with supply growth below that. Then they see demand growth “mid-double-digits” (I think they mean mid-teens) in 2026 with supply continuing to be below that. I am shocked at their confidence predicting supply and demand dynamics in the memory market going out nearly a year and a half.
· Demand is believed to be “consistent” currently and in the second half of calendar 2025. The company believes inventories at their customers have normalized.
· Several analysts asked questions on why they see the market will be undersupplied for several more quarters. I’m not the only one who is wondering where the company’s confidence comes from on this.
I will hand it to the executive team at SanDisk. They have called an historic shot and are sticking with that prediction. They forecasted again that the NAND market will remain undersupplied for the rest of 2025 and into 2026. They first made this prediction one quarter ago. So far, they were a little optimistic on pricing and a little pessimistic on bit shipments. These two are related, so maybe they over shipped bits a bit and took the hit on price. They also believe customer inventories are at normal levels now. My view is the NAND market is close to balanced right now. If that is the state, I think it is nuts to predict it will remain undersupplied for almost six more quarters. I think the company is spun up on their newfound independence after the split from WD and they now believe they will be the ones to overturn forty-plus years of commodity semiconductor memory history. They will underutilize their fabs with such skill that they will stave off the next downturn. Other NAND manufacturers, including YMTC, will show a similar level of prognosticating skill. Also, SanDisk’s current and future products will be so strong that they will evade commoditization. I am short these beliefs as I am short Micron’s stock. I think NAND pricing will be headed down by the end of the year, and probably by early in the fourth calendar quarter. This is a push-out in my view from ninety days ago, when I thought the fall in pricing would continue unabated. I think it has slowed and may actually go up for a quarter or two, but I still think the NAND market is unhealthy and will be oversupplied before the end of 2025.
– S. Hughes (short MU)