Sandisk Stock Is Up More Than 3,700% From Its 52-Week Low. Is the Memory Rally Still Investable, or Is This Stock Priced for Perfection?

SNDK is the cleanest way to express the AI storage bottleneck, but the market is already paying up for a very rare combination: cyclical earnings at what could be peak operating leverage, yet a still-modest headline multiple. The key mechanism is not just better demand; it is duration. If hyperscaler storage demand is truly multi-year, then forward estimates can keep ratcheting up and the current P/E can stay deceptively low even after a large rerating.

The second-order winner is likely the rest of the NAND ecosystem, but with less torque than SNDK because the market will assume the purest AI exposure deserves the premium. WDC still has leverage to a healthier memory tape, yet it lacks the same post-spin narrative and may trade more like a mixed-quality legacy storage asset. On the customer side, higher storage costs are manageable relative to GPU budgets, so this is more likely to slow procurement pacing at the margin than to break AI spending outright.

The real risk is that investors are confusing revenue visibility with pricing immunity. Multiyear contracts reduce near-term revenue shock, but they do not eliminate the classic memory-cycle failure mode: once supply catches up or hyperscaler inventories normalize, ASPs can roll over quickly and the multiple can compress before the earnings base visibly breaks. Over the next 1-3 months, watch for guide revisions, backlog commentary, and any sign that NAND price strength is decelerating; over 6-18 months, the thesis is falsified if AI capex growth slows materially or storage intensity per workload proves lower than expected.