Sandisk: Brace Yourself, The Ride Won't Be Easy

Sandisk (SNDK) is rated a speculative buy as AI-driven NAND demand accelerates. The company reported record Q4 2026 revenue near $9B, up 372% YoY, with non-GAAP gross margin expanding to 84.6% on stronger pricing. Visibility is improving: NBM contracts now guarantee $16.5B in commitments, with 50% of FY27 production secured and expectations to reach ~2/3 by FY28.

Analysis

This is less a one-quarter earnings story than a regime shift in how the market should value SNDK: if a meaningful share of output is now locked into multi-year commitments, the business starts to behave more like contracted infrastructure than a pure spot-memory cyclical. That matters for multiple expansion because the market typically pays up for visible capacity scarcity when it believes supply is sticky and customer switching costs are high.

The second-order winner is the broader NAND ecosystem: equipment vendors and adjacent storage suppliers should see follow-through if peers are forced to defend share with capex or pricing discipline. The loser is hyperscalers and server OEMs if storage inflation persists, but the pressure is likely small versus GPU spend; the bigger effect is on project timing, where enterprises may delay lower-priority inference deployments or optimize architectures to use fewer high-end drives per rack. In other words, the demand shock can show up first in reorder patterns, not in headline data-center construction.

The main risk is that this looks like peak-cycle economics dressed up as AI secular growth. If supply normalizes over the next 6-12 months, margins can compress faster than revenue, and the equity can derate even while shipments stay solid. The thesis is falsified if spot NAND pricing softens, if peer capex ramps, or if the next contract roll-off shows customers pushing back on take-or-pay economics; those are the signals that the current scarcity premium is being arbitraged away.

Contrarian view: the market may be underappreciating how much of the current upside is already priced in as a scarcity narrative, while underappreciating the probability that AI storage demand is lumpy and inventory-driven. If the stock has already moved as a 'quality AI' name, the better expression may be relative value rather than outright chase.

Key Decisions for Investors

• Long SNDK on 3-6 month horizon, but only on pullbacks or after consolidation; use a 15-20% trailing stop because the stock is vulnerable if margin expectations normalize faster than revenue growth.
• Pair trade: long SNDK / short WDC for 1-3 months to isolate the market’s willingness to pay for pure-play AI storage visibility versus mixed exposure; thesis breaks if WDC re-rates on stronger enterprise demand or if NAND pricing cools.
• If options liquidity allows, express upside with a 3-6 month call spread rather than outright equity; the asymmetry is positive if contract visibility keeps improving, but implied volatility is likely elevated after a name like this.
• Watch MU and the broader semiconductor equipment complex (AMAT, LRCX, KLAC) as confirmation signals: if they start guiding tighter supply or stronger memory capex, the NAND tightness thesis extends 6-18 months; if not, take profits on SNDK strength.
• Set an alert for any evidence of customer pushback on multi-year commitments or inventory build in channel checks; that would be the cleanest early warning that the AI-demand narrative is being front-loaded rather than sustained.