Sandisk Stock Was the Biggest Winner in the First Half of 2026. What's Next for the Second Half?

SanDisk (SNDK) is up ~800% year-to-date in 2026 and is benefiting from a NAND memory shortage that could persist beyond 2027, with Micron expecting “tight conditions” to continue beyond that window. Management/industry demand tailwinds are reinforced by accelerating data-center buildouts, with Nvidia projecting $3T–$4T in annual spend by 2030. Wall Street projects fiscal Q4 growth of 336% and fiscal 2027 growth of 122%, and the article argues the stock’s forward P/E of ~11x is inexpensive enough to allow further upside (potentially doubling) even without repeating the prior surge.

Analysis

SNDK is trading less like a normal memory supplier and more like a scarcity asset, which is why the key question is duration, not just current demand. The upside still exists if the market keeps underestimating how long supply can lag hyperscaler storage demand, but the easy multiple expansion is probably behind it; from here, returns depend on estimate revisions and margin capture, not rerating alone. MU should benefit too, but SNDK has the cleaner operating leverage to NAND pricing because it is more directly exposed to the tightness.

The more interesting second-order effect is that rising NAND costs act like a tax on AI infrastructure buildouts. If SSD economics keep worsening, hyperscalers can slow some storage-heavy deployments, push more budget toward compute/HBM, or stretch replacement cycles in enterprise IT; that is a subtle medium-term headwind for the broader semiconductor complex, even if it does not bite immediately. NVDA remains the primary beneficiary of AI capex, but this kind of input-cost inflation can create friction at the margin if data-center BOMs keep moving higher.

The contrarian view is that a multi-year shortage can still be a cyclical peak if the supply response arrives faster than consensus expects. The main reversal catalysts are capex inflection, shorter tool lead times, or management commentary that 2027 tightness is no longer intact. Near term, momentum likely remains intact; over 6-18 months, the trade becomes about who keeps revising earnings higher fastest, not who looks cheapest on static forward P/E.

Key Decisions for Investors

• Stay long SNDK on pullbacks, not breakouts: accumulate on 10-15% drawdowns over the next 1-3 months, with the thesis invalidated if management starts signaling margin deceleration or supply normalization before mid-2027.
• Express the relative-value view with a small long SNDK / short MU pair for 3-6 months if the market keeps rewarding pure-play NAND leverage; cut the pair if MU's commentary shows NAND tightness broadening into a stronger memory supercycle rather than remaining SNDK-specific.
• Use SNDK call spreads instead of outright stock if implied volatility remains elevated; the best risk/reward is a defined-risk upside capture into the next earnings cycle, not chasing spot after an 800% run.
• Treat NVDA as a secondary hedge, not a direct beneficiary: if data-center storage inflation starts to slow capex conversion, reduce broad semi exposure rather than the AI compute leg first.
• Set a watch item on capex and supply commentary from MU/SNDK: if either management team indicates meaningful capacity coming onstream by 2H27, take profits on SNDK longs because the scarcity multiple can compress quickly.