JPMorgan cautious on Sandisk, sees risk of earnings normalization

JPMorgan initiated coverage of SanDisk with a Neutral rating and a $235 price target, noting the stock’s strong year-to-date rally limits further multiple expansion. The bank acknowledged SanDisk’s exposure to the AI-driven enterprise SSD supercycle and its cost advantage via the Kioxia JV, but emphasized materially smaller AI/datacenter exposure versus peers and warned current pricing looks like a cyclical peak with planned capacity additions in 2027+ risking margin pressure—characterizing the risk/reward as balanced.

Analysis

Market structure: Near term (next 3–9 months) the winners are low-cost NAND suppliers that can pick up hyperscaler eSSD wallet share (SanDisk via Kioxia JV) and larger peers with broader datacenter optionality (e.g., WDC, MU) as investors rotate into AI exposure. Losers will be smaller, higher-cost suppliers and OEMs whose SKU mix is heavy on consumer SSDs if NAND ASPs roll over; JPMorgan’s call flags 2027 capacity additions as a 20–40% downside risk to cyclical margins. Cross-asset: a visible NAND-led tech re-rating would tighten IG credit spreads for large suppliers but widen junk spreads for specialist smaller names; commodity and FX moves should be muted except for capex-driven copper/silicon demand on multi-quarter horizons.

Risk assessment: Tail risks include accelerated Chinese capacity adding 30%+ incremental supply (high impact, low prob) or a hyperscaler AI budget pause that cuts procurement >15% QoQ. Immediate (days) risks are sentiment/earnings reactions; short-term (0–6 months) risks are ASP volatility ±15% and inventory swings; long-term (2027+) risks are structural margin erosion from capacity. Hidden dependencies: hyperscaler inventory-days, cross-licensing with Kioxia, and NAND mix (QLC vs TLC) can shift gross margins by 300–500bps.

Trade implications: For a 3–6 month tactical trade favor a modest long SNDK (1–3%) to capture cyclical price momentum but monetize with 30–60 day call overwrites or sell-call spreads to protect against mean reversion. Consider a relative-value pair long WDC (2–3%) / short SNDK (2%) to express rotation into broader datacenter optionality; target 8–12% relative outperformance and cut if relative moves exceed -8% in 6 weeks. For multi-year hedge, buy 18–24 month SNDK put spreads (tight cost) sized 0.5–1% of portfolio to guard against 2027 capacity risk.

Contrarian angles: Consensus underestimates timing risk — the market may already price a near-term cyclical peak; therefore short-duration options sellers can collect premium but beware of earnings shocks. Historical parallels: 2017–19 NAND cycle shows 12–18 month rallies followed by 30–50% trough-to-peak mean reversion when capacity steps up, implying any long position >6 months requires protection. Unintended consequence: overly bullish positioning in SNDK could compress volatility and leave long-only funds exposed when capacity announcements materialize, amplifying downside.

Key Decisions for Investors

• Establish a tactical 1–3% long position in SNDK for a 3-month horizon to capture cyclical upside, immediately sell 30–60 day call spreads (e.g., buy 1–2% OTM calls and sell near-ATM calls) to cap upside and collect premium; liquidate if NAND ASPs fall >10% QoQ or SNDK share price drops 12% from entry.
• Implement a 2% long WDC / 2% short SNDK pair trade to express rotation toward broader datacenter optionality, target 8–12% relative return in 3–6 months; close the pair if WDC underperforms SNDK by >8% within 6 weeks or if WDC guidance recasts downward.
• Buy an 18–24 month SNDK put spread (buy longer-dated 15–25% OTM puts, sell deeper 30–40% OTM puts) sized 0.5–1% notional as insurance against 2027 capacity-driven margin erosion; increase hedge if Kioxia or Chinese OEM announce +20% incremental capacity.
• Reduce consumer SSD/retailer exposure by 1–2% of portfolio weight and reallocate to select datacenter-exposed names (e.g., WDC, MU) that derive >30% revenue from hyperscalers; review hyperscaler inventory reports and monthly NAND ASP prints for rebalancing decisions.