Why Sandisk Stock Popped Again Today

Melius Research initiated Sandisk at Buy with a $1,350 price target, implying about 36% upside from the stock's recent level near $990. The note argues AI-driven memory demand and unusually high margins could persist through the end of the decade, supporting a higher valuation multiple. The article also notes Sandisk's earnings inflection after three years of losses, with analyst forecasts calling for $41.75 EPS in 2026 and $107 in 2027 before a cyclical pullback.

Analysis

The market is not just pricing Sandisk as a beneficiary of AI memory demand; it is implicitly debating whether this is a true scarcity cycle or a classic semiconductor inventory cycle in disguise. The second-order tell is that if memory ASPs and margins stay elevated longer than consensus expects, the entire AI hardware stack gets repriced upward because memory has been the most visibly cyclical leg of the bill of materials. That creates a relative winner-set for suppliers with leverage to AI buildout but less direct exposure to front-end compute capex.

The key risk is timing mismatch: the equity can rerate immediately while fundamentals may only validate over several quarters. If hyperscaler capex moderates, or if NAND/HBM capacity additions come through faster than expected, the market will compress the multiple before reported earnings roll over. In that scenario, the downside is not necessarily a collapse in earnings, but a de-rating from “strategic scarcity” back to “late-cycle component,” which can cut 20-30% off the stock even with flat-to-up profits.

The contrarian read is that investors may be over-fixated on peak-cyclicity and underappreciating duration: the AI memory bottleneck is more about qualification cycles, packaging constraints, and customer concentration than generic semiconductor supply. If that is right, the winners extend beyond Sandisk into adjacent memory/test/packaging ecosystems, while weaker second-tier memory vendors remain structurally disadvantaged because they lack scale to win long-duration supply agreements. This is a multiple story first, earnings story second.

Key Decisions for Investors

• Go long SNDK on pullbacks, not breakouts, using a 3-6 month horizon; target a re-rating toward the analyst upside case, but keep a hard stop if the stock loses its post-initiation momentum and AI capex commentary weakens.
• Pair trade: long SNDK / short a more cyclical semiconductor name with less AI duration exposure (e.g., INTC) for a 6-12 month relative-value expression; the thesis is memory scarcity vs. legacy semiconductor reset, with downside if AI demand broadens but memory pricing normalizes.
• Buy medium-dated call spreads on SNDK to express upside while limiting multiple-compression risk; this is best if you expect continued positive revisions over the next 1-2 earnings prints.
• Watch NVDA and the broader AI capex tape as a confirmation signal; if Nvidia commentary or hyperscaler spending inflects down, reduce SNDK exposure quickly because the stock is likely trading ahead of visible earnings.
• Consider a basket long in AI infrastructure beneficiaries with less direct cycle risk than SNDK, funded by shorts in names with weaker pricing power; the idea is to capture the AI duration theme without taking full commodity-memory beta.