Flash storage, recently separated from Western Digital. | Analysis date: 2026-03-13
Leopold holds ~$250M in SanDisk. SanDisk was spun off from Western Digital in early 2025 and designs/sells NAND flash memory. AGI score of 0 is likely wrong (scored before separation). The real thesis: AI training and inference require massive storage. SanDisk at $662/share ($97.7B market cap) trades at 9.6x book value. Leopold is betting on storage demand explosion from AI.
SanDisk designs and sells NAND flash storage: enterprise SSDs, client SSDs, consumer flash drives and cards. Separated from Western Digital in Feb 2025. ~$7.4B revenue (XBRL FY2025), rapidly growing to ~$8.9B (TTM). 11,000 employees. Currently unprofitable (net loss $1.6B in FY2025) as it ramps up post-separation, but gross margins improving to ~35%. Has $5B goodwill from WD legacy.
| Metric | FY2025 |
|---|---|
| Revenue | $7.4B |
| Net Income | -$1.6B |
| Operating Cash Flow | $84M |
| Capital Expenditures | $204M |
| Operating Income | -$1.4B |
| Gross Profit | $2.2B |
| Balance Sheet (Latest FY) | Value |
|---|---|
| Total Assets | $13.0B |
| Cash & Equivalents | $1.5B |
| Property, Plant & Equipment | $619M |
| Goodwill | $5.0B |
| Long-Term Debt | $1.8B |
| Stockholders Equity | $9.2B |
| Tangible Book Value | $4.2B |
| Tangible Book / Share | $29.08 |
| Shares Outstanding | 145,000,000 |
Leopold Aschenbrenner holds ~$250M in SNDK. AI storage demand thesis.
AI training requires petabytes of fast storage for datasets and checkpoints. Inference at scale needs NVMe SSDs for model loading and caching. As AI scales from millions to billions of users, storage demand grows non-linearly. NAND flash prices are recovering. SanDisk is positioned for a massive upcycle as AI drives demand. Revenue could double to $15-20B.
SanDisk is freshly separated from WD and currently unprofitable due to separation costs and NAND cycle trough. As the cycle recovers and separation costs fade, margins should normalize to 15-20% net margins. At 9.6x book, the stock is priced for the recovery, but if AI demand creates a super-cycle, it could be cheap.
NAND flash is largely a commodity. SanDisk competes with Samsung, SK Hynix, Micron, and Kioxia. Pricing power is limited. Currently unprofitable. The $5B goodwill is questionable. Tangible book is only ~$4.2B vs $97.7B market cap. 23x tangible book is expensive.
At $662/share and $97.7B market cap, SNDK trades at ~13x trailing revenue. For a commodity NAND company, this is aggressive. If the AI storage thesis does not play out, the stock could easily halve. There is no asset floor here.
| Current Price | $661.62 |
| Bull Case Price (by 2030-2035) | $2,000 |
| 10x Entry Price (Bull Case / 10) | $200 |
| Current Price vs 10x Entry | $661.62 ABOVE entry zone |
Rationale: If AI storage creates a super-cycle: $20B revenue at 18% margins = $3.6B earnings at 40x = $144B market cap = ~$975/share. Aggressive: $2,000. Entry for 10x: $200. Need major NAND downturn + post-IPO hangover.
SNDK at $661.62: Trading well above the 10x entry zone (would need a significant pullback for 10x potential).
AGI Score 0/10 — Moderate AGI impact. Leopold holds this position.
Floor estimate: Tangible book value per share is $29.08. Current price is well above tangible book: downside protection comes from cash flows, not assets.
Data sources: SEC EDGAR XBRL, yfinance, 10-K filings. Analysis date: 2026-03-13.