Hook / Thesis
On 03/26/2026, Alphabet published a paper showing a new AI compression technique that can reduce memory requirements for large language models by more than 6x. The market reacted violently: memory names were marked down on fear that a sudden structural decline in demand for NAND could be coming. That reaction was overdone.
Sandisk (SNDK) is not a generic commodity vendor; it is a leading NAND flash supplier with a market setup and balance sheet that lets it cash in on tight supply for AI datacenters. The pullback that followed Alphabet’s announcement created a tactical buying opportunity. I’m upgrading to a buy for a mid-term trade: entry $636.77, stop $580.00, target $760.00 over a mid-term horizon (45 trading days).
What Sandisk Does and Why the Market Should Care
Sandisk develops, manufactures and sells NAND flash-based storage: SSDs, memory cards and USB flash drives. The company is trading with a market cap around $94.0 billion and sits at the center of the AI infrastructure stack because modern training and inference workloads are extremely storage- and bandwidth-hungry.
Two forces matter for investors: (1) near-term demand from hyperscalers that are buying enterprise SSDs and custom storage arrays for AI clusters, and (2) industry supply dynamics. Supply remains constrained relative to surging AI demand, which supports pricing and gross margins even amid efficiency gains in model architectures. That mismatch - constrained supply vs strong end-demand - is the reason Sandisk’s cash generation has been meaningful despite a volatile headline environment.
Hard Numbers That Support the Case
| Metric | Value |
|---|---|
| Current price | $636.77 |
| Market cap | $94,009,316,087 |
| Enterprise value | $88,092,478,281 |
| Free cash flow (last reported) | $1.449B |
| Price / Sales | 9.97x |
| EV / Sales | 9.87x |
| Price / Cash Flow | 54.72x |
| Debt / Equity | 0.06 |
| Free float (shares) | ~138.7M |
| 52-week range | $27.89 - $777.60 |
Two numbers stand out: free cash flow of $1.449 billion and enterprise value of roughly $88.1 billion. That puts FCF yield at roughly 1.6%, which is low on the surface, but FCF is real and the business is capital-light relative to other parts of semiconductors (debt/equity 0.06). The company’s balance sheet gives management breathing room to maintain operations and invest for AI product cycles while customers ramp.
Why the Alphabet Paper Is a Head-Fake for Storage
Alphabet’s compression idea matters academically and will influence engineering trade-offs, but there are three practical reasons why it’s unlikely to collapse NAND demand in the near-to-mid term:
- Model-efficiency gains take time to adopt at hyperscaler scale. Even if a technique compresses memory needs 6x in research settings, production adoption requires validation, changes to existing pipelines, and hardware qualification.
- Compression addresses model memory footprint, not raw dataset storage, hot cacheability, latency-sensitive NVMe SSD tiers or backup/replication needs. Hyperscalers will still buy low-latency enterprise SSDs for inference and training cache layers.
- Supply-side tightness for NAND persists. Analysts and markets are still factoring in constrained supply through 2028; tight supply protects pricing even if per-model memory demand grows more slowly.
Technical and Market Context
Sandisk has been volatile: a 52-week low of $27.89 (04/07/2025) — the spin-off trough — to a 52-week high of $777.60 (03/20/2026). The stock’s 10-day SMA is $698.18 and the 50-day SMA is $602.78; RSI sits near 49, neutral. Recent MACD shows bearish momentum, which explains why short-term traders sold into the Alphabet news. Short interest has been modest in days-to-cover terms (around 1 day), but daily short-volume numbers show sizable activity on high-volume days, which can amplify moves in either direction.
Valuation Framing
At a market cap near $94 billion and EV roughly $88 billion, Sandisk is priced like a premium structural growth play rather than a cyclical commodity maker. Price-to-sales near 9.97x and P/FCF above 60x reflect that the market has already priced in continued strong growth from AI infrastructure demand. That’s a high bar, and the stock can gap lower on macro or demand fears — which is exactly why the Alphabet-triggered dip matters: it lets disciplined traders buy into an above-average business at a slightly lower price.
Put differently, you're paying growth multiple premium for NAND exposure. I accept that premium for a mid-term trade because I expect the supply/demand imbalance to reassert itself and because Sandisk’s cash flows and low leverage reduce structural risk versus peers that are more capital intensive.
Catalysts (what could drive the trade higher)
- Hyperscaler re-acceleration: renewed, visible procurement from major cloud providers for AI clusters will restore confidence in enterprise SSD demand.
- Quarterly results showing sequential revenue growth and margin resilience vs consensus expectations.
- Micron/competitor commentary confirming constrained NAND supply or pushing capex timelines later than expected.
- Any supply-chain update showing continued shortages in key inputs (e.g., controller chips or wafers) that keep pricing firm.
Trade Plan (actionable)
Direction: Long
Entry: $636.77
Stop loss: $580.00
Target: $760.00
Horizon: mid term (45 trading days) — I expect the trade to play out across several catalyst windows (quarterly results, competitor commentary, and re-acceleration in hyperscaler purchases) inside this timeframe. If those catalysts take longer to show, I will reassess and either roll the position or tighten risk.
This is a tactical, conviction-backed trade: entry is near current market levels after the headline-driven volatility; the stop at $580 limits downside if the market re-prices growth expectations aggressively. The target of $760 sits below the recent $777.60 high on 03/20/2026 and reflects a reversion to momentum-driven highs rather than a multiple expansion call.
Risks and Counterarguments
- Structural demand risk: If Alphabet-style compression is adopted broadly and quickly, model-level memory demand could decline meaningfully, reducing long-term addressable demand for NAND in AI workloads.
- Inventory and cyclical risk: Hyperscalers could destock or delay purchases, triggering an inventory-driven price correction in NAND components.
- Valuation vulnerability: The market already prices strong growth — a single negative print or weak guidance could compress multiples rapidly given P/S near 10x and elevated P/FCF.
- Macroeconomic / geopolitics: Escalating geopolitical events or a macro recession could slow enterprise IT spend and cloud capex, hitting Sandisk sales more than expected.
- Execution risk: If Sandisk misses on product execution, supply agreements or margin improvement, cash flow could disappoint and the premium multiple would be at risk.
Counterargument: Technical momentum has been choppy and some sell-side targets are lower than current price. If broader memory sentiment deteriorates after follow-up research or competitor product substitutions emerge, the stock could resume a deeper correction. The trade is therefore medium-risk: the upside depends on the supply/demand narrative holding and Sandisk showing sequential strength; downside is cushioned by the stop above structural support.
What Would Change My Mind
I would abandon this long stance if one or more of the following occur within the next 45 trading days: (1) hyperscaler capex guidance shifts from continued growth to sustained contraction, (2) Sandisk reports a material sequential revenue or margin miss accompanied by inventory build commentary, or (3) multiple industry players corroborate that compression techniques are being deployed in production at scale and materially reduce hardware needs. Conversely, stronger-than-expected procurement from cloud providers or continued supply tightness would reinforce this thesis.
Conclusion
The Alphabet compression paper was an important technical advancement, but it did not change the immediate reality in the market: hyperscalers still need vast amounts of fast storage, and NAND supply remains tight. Sandisk’s combination of real free cash flow ($1.449B), low leverage and scale puts it in a favorable position to benefit from AI infrastructure spending even if model-level efficiency improves over time.
For focused traders willing to accept medium risk, buying Sandisk at $636.77 with a $580 stop and $760 target over the next 45 trading days is a rational way to play a near-term overreaction while maintaining disciplined risk control.