Sandisk's Rally May Be Far From Over

Sandisk reported $5.95 billion in revenue, up 97% sequentially, alongside exceptional gross margins of 78.4%. Data center revenue surged 233% quarter-over-quarter as AI workloads drive demand for larger, more storage-intensive architectures. The company also secured five multiyear agreements covering over one-third of FY2027 output and more than $11 billion of guarantees, supporting future revenue visibility.

Analysis

This is less a single-quarter beat than a re-rating event for the storage value chain. When a supplier can lock in a meaningful slice of future output with guaranteed economics, the market usually stops pricing it like a cyclical component vendor and starts treating it like a capacity-constrained infrastructure asset. The spillover is negative for downstream OEMs and cloud buyers that were expecting memory pricing leverage to remain cheap; the bargaining power has likely shifted from customers to producers sooner than the consensus expected.

The second-order winner is the entire AI storage stack, not just the company itself. Larger model training and inference footprints increase the dollar content per server rack, which should support suppliers of enterprise SSDs, controllers, and high-performance NAND-related components, while pressuring data center operators to absorb a structurally higher storage bill. That dynamic also creates a subtle margin squeeze for hyperscalers if they cannot offset with higher utilization or software monetization, making storage a real cost line rather than a rounding error.

The main risk is that the market extrapolates peak scarcity too far out the curve. The setup is vulnerable if supply additions, customer inventory digestion, or a moderation in AI capex emerge over the next 2-4 quarters; in that case, the guaranteed demand story becomes less scarce and the multiple can compress faster than earnings improve. Longer term, the bull case stays intact so long as AI architectures continue to become more storage-heavy, but the near-term trade is most exposed to any sign that customer concentration is forcing renegotiation pressure or that the company is overearning versus normalized cycle economics.

Consensus may be underestimating how much of this is a pricing-power story versus pure volume growth. The move is likely underdone if investors still model memory as mean-reverting commodity exposure; if instead this is an AI-enabled capacity regime shift, the earnings power is more durable and the stock can re-rate on forward margin visibility, not just revenue acceleration. The contrarian caution is that the best time to own these names is often before the market fully recognizes tightness; once contract structure and guarantee language become widely accepted, the easy multiple expansion may already be mostly captured.

Key Decisions for Investors

• Long SNDK on pullbacks over the next 1-3 weeks; use strength to build rather than chase, with a 3-6 month target tied to continued forward guide revisions and downside capped by the newly visible contract floor.
• Pair trade: long SNDK / short a representative AI hardware beneficiary with lower pricing power over the same 1-2 quarter horizon; thesis is that storage scarcity is becoming more monetizable than general AI hardware demand.
• Add exposure to the broader storage/flash supply chain for 3-6 months; the strongest risk/reward is in names most levered to ASPs and contract repricing, where even modest upside revisions can drive outsized multiple expansion.
• Buy limited-risk upside via call spreads in SNDK for the next earnings cycle; structure to capture another leg of re-rating while defining risk if AI spend pauses or customer inventory unwinds.
• Trim or avoid shorting data-center capex proxies into this tape unless there is clear evidence of margin pressure; the better short is not AI demand itself, but companies with weak pass-through economics and heavy storage intensity.