There is no clean way to contextualize what SanDisk has done. A stock that priced a secondary public offering of shares sold by Western Digital at $38.50 per share in June 2025 is now trading near $1,959 as of June 17, 2026. That is not a typo. That is a 5,000%-plus move in roughly a year, making SNDK one of the most aggressive price appreciation stories in semiconductor market history.
The stock hit a new 52-week high of $2,167 on June 16 before pulling back. Wall Street price targets range from $1,000 on the low end to $2,900 from Cantor Fitzgerald’s C.J. Muse, with Bank of America at $2,100 and Mizuho at $2,200. The dispersion tells you everything about how divided the street actually is.
Why the Numbers Are Genuinely Strong
This is not a story being held up by hype alone. SanDisk reported Q3 FY2026 revenue of $5.95 billion, a 251% increase year-over-year, with gross margin in the high-70% to low-80% range. The company also disclosed three signed “New Business Model” (NBM) multi-year supply agreements in the quarter totaling $42 billion in minimum contractual revenues (for those three Q3 agreements).
- Q3 FY2026 Revenue: $5.95B (up 251% YoY)
- Gross Margin: high-70% to low-80% range
- EPS (Q3): $23.41 (non-GAAP)
- Contracted Backlog: $42B minimum contractual revenues (for the three Q3 NBM agreements)
The core business logic is straightforward. NAND flash storage and enterprise SSDs sit at a critical chokepoint in AI infrastructure. Training large language models and running inference workloads at scale requires massive, fast, persistent storage. SanDisk is a pure-play NAND company in a supply-constrained market where demand from hyperscalers is accelerating.
Slight tangent worth noting: I could not verify the claim that Lynx Equity specifically named SanDisk as a direct beneficiary of SpaceX’s xAI and a “Terafab” project. Without a primary, on-the-record source, that should be treated as unconfirmed and not part of the core thesis.
The Part People Are Skipping
Polymarket apparently labeled SNDK “the most overbought stock in history” as of June 16. That is a crowd signal, not a fundamental one, but it is worth sitting with for a moment.
The trailing P/E is near 67x. That is the number that should make anyone pause. A company whose rally is driven by genuine fundamentals can still experience a 30-40% momentum-driven correction without the underlying business deteriorating at all. The multiple contracts before the thesis breaks.
The next major fundamental checkpoint is the next earnings report, which some market calendars currently list as August 24, 2026. Before that date, Western Digital has disclosed exchange transactions involving shares of SanDisk that are expected to close on June 22, 2026, which could add near-term technical noise to the price action.
Bull / Base / Bear
- Bull: AI inference workloads keep scaling, new customer wins (hyperscalers) push 2027 bookings to capacity, pricing holds above 2026 levels. SNDK trades toward $2,500+.
- Base: Demand remains strong but momentum cools into the June 22 exchange-related close. Stock consolidates in the $1,700-$2,100 range through the next earnings report.
- Bear: Samsung or SK hynix new NAND capacity comes online faster than expected, pricing breaks, multiple compression accelerates. A 35-40% drawdown is plausible without the business actually failing.
What Investors Should Watch
Gross margin trajectory is the single most important number heading into the next print. If margins compress materially, the pricing power thesis starts to weaken. Watch also for any indication that hyperscaler capex guidance is being revised downward in upcoming earnings calls. That is the domino that would matter most.
Beyond that, the June 22 Western Digital exchange-related close is a near-term event risk. It is structural, not fundamental, but the mechanics could create temporary supply-side pressure on the shares.
The bull case for SNDK is still alive. The question is whether the stock at $1,959 gives you the same risk-reward as it did at $400. It does not. Traders and long-term investors need entirely different exit frameworks here.
For informational purposes only.
