Sandisk Stock Is Down 30% From Its Peak: Is the Memory Winner Still a Screaming Buy Before September?

Sandisk (SNDK) is up nearly 600% in 2026 but is still ~30% below its peak after a mid-year profit-taking pullback. The article attributes the surge to ongoing NAND memory pricing strength driven by AI data-center demand, while warning that the memory market is cyclical and prices should fall once the supply shortage eases (the shortage not expected to be resolved in 2027). Valuation is highlighted as relatively cheap at 7.6x forward earnings versus many AI peers above 20x, implying potential upside if earnings remain strong and demand holds.

Analysis

The real alpha here is not just that SNDK can print near-term earnings beats; it is that constrained NAND supply acts like a hidden tax on the AI infrastructure stack. Over the next 1-3 quarters, that should support SNDK’s gross margin leverage and keep peers like WDC and the memory-heavy portions of SMH elevated, while creating margin pressure for OEMs and server integrators that cannot fully pass through storage costs. If the shortage persists into 2027, the market may start to re-rate NAND vendors less like cyclical commoditized hardware and more like quasi-capacity assets with pricing power.

The second-order loser is the customer side of the AI buildout: hyperscalers can still buy GPUs, but higher storage cost per rack can slow deployment economics, favor more selective capex, and encourage substitution toward higher-value compute per byte stored. That is modestly negative for NVDA at the margin over 6-18 months, not because demand vanishes, but because the total system bill of materials rises and procurement teams optimize around ROI. Any sign of inventory normalization or new capacity coming on faster than expected would likely compress SNDK’s multiple before it hits earnings.

The consensus may be underestimating how fragile the bull case is to a single data point: quarterly NAND ASPs and customer lead times. At 7.6x forward earnings, the stock is discounting either a fast cycle turn or skepticism that peak margins persist; that creates upside if the shortage is sticky, but also makes the name vulnerable to sharp de-rating on the first miss. The key falsifier is not macro noise but evidence of weakening order book quality, easing contract prices, or channel inventory rebuilds over the next 1-2 earnings prints.

Key Decisions for Investors

• Long SNDK on pullbacks, sized for 1-2 earnings catalysts over the next 3-6 months; target is multiple expansion toward higher-quality AI hardware comps if ASPs and margins hold, but cut risk if management commentary implies easing contract pricing.
• Pair trade: long SNDK / short WDC for the next 1-2 quarters to express the pure-play scarcity premium, since SNDK should carry more convexity if NAND remains tight while WDC is more exposed to cycle skepticism.
• Small tactical hedge: short NVDA or buy downside protection on NVDA into the next AI capex read-through if storage inflation starts to show up in hyperscaler budgets; this is a second-order hedge, not a primary thesis.
• Set an alert on the next SNDK earnings release for any commentary on lead times, inventory days, or customer mix; a clear inflection in any of those metrics would be the best trigger for either adding to or exiting the long.