Can Sandisk be empty now?

One of the craziest stocks in the market recently is not Nvidia, not Tesla, but a company many previously looked down upon—SanDisk, a “USB drive seller.”

SanDisk has surged 40 times in a year, from $30 in May 2025 to $1,562 now, with a market cap of 231 billion yuan, a P/E ratio of 46 times, and the recent price rally seems quite aggressive. I’m pondering whether this wave of SanDisk’s rise is about to end, and whether this position offers a relatively high value for risk-reward?

1. SanDisk’s main business:

SanDisk mainly deals with storage—NAND Flash (solid-state memory). Its main products include: consumer storage (USB drives, memory cards), SSDs, enterprise SSDs, NAND Flash chips (for device manufacturers).

The current AI wave has led giant companies to massively expand AI data centers, and AI servers demand “memory” several times or even dozens of times more than regular servers. Because AI training requires huge amounts of data, first the data must be stored, then fed into AI training.

2. Can production capacity meet demand?

So, this demand is undoubtedly real. My question is, can they ramp up capacity so quickly? Since market cap has surged, if performance delivery falls short, isn’t that just a bubble?

The results I found indicate they are indeed falling behind. In the fiscal third quarter of 2026 (ending April 3, 2026), the earnings call explicitly stated NAND demand continues to outpace supply, with inventory turnover days dropping from 135 to 115 days. This tight balance is expected to continue until the end of 2026.

Therefore, companies like SanDisk are trying to extend capacity through technological iterations (higher-layer NAND, larger capacity 256TB SSDs) and a joint venture with Kioxia until 2034 to secure capacity. But in the short term, it remains a seller’s market, with prices and gross margins at high levels (Q3 gross margin once approached 80%).

3. Recent orders

On April 30, 2026, when the financial report was released, SanDisk announced:

• Signed five long-term supply agreements (up to 5 years), with three contracts signed in the third quarter totaling at least $42 billion in minimum contract revenue.

In the fourth quarter, two more agreements were added, totaling five agreements with financial guarantees exceeding $11 billion (including customer prepayments, collateral, and breach compensation mechanisms, far exceeding traditional orders).

Because wafer fab construction takes 2-4 years and involves hundreds of billions of dollars, many customers are locking in orders early to secure prices, as memory prices have risen too fast, and they worry about not being able to get supplies later.

Who would have thought that a memory market, once considered a “junk cycle” stock, would now become the foundation of the AI industry?

4. Financial analysis

The latest financial report shows that in the first quarter of 2026, revenue was 33.4B yuan, far exceeding the estimate of 1.7 billion yuan last year, a surge of 340%. Based on estimates for the next three quarters, revenue would be 81 + 94 + 100 = 53B yuan, so the total estimated revenue for 2026 is 59 + 275 = 32.3B yuan. Given the orders announced by the CEO totaling 42 + 11 = 53 billion yuan, the current estimate appears lower than the CEO’s announced order book.

Let’s calculate the P/E ratio for 2026:

Using the first quarter net income of 3.6 billion and revenue of 5.9 billion, the ratio is 61%. If we assume the CEO’s revenue estimate of 530 billion yuan, then net income would be 530 * 61% = 323 billion yuan. With a market cap of 2,310 billion yuan, the P/E ratio is 7 times! So, its dynamic P/E ratio is only 7, indicating a complete undervaluation. No wonder SanDisk is rushing like crazy. If future earnings continue to be good, it might not be a bad idea to hold more!

5. Does SanDisk have risks?

The answer is yes. During the internet bubble, storage stocks also experienced a bubble. When demand didn’t meet expectations and capacity kept expanding, the market could collapse. NAND crashed in 2000. Currently, giants are fighting for AI market share, fearing falling behind, so they are aggressively expanding. The AI market is limited, and with so many people, some will exit once they lose. Will infrastructure demand follow the same pattern? I think it’s highly likely, just not yet fully reflected.