Is This AI Winner Still a Buy After a 500% Run?

Since its spinoff from Western Digital earlier in 2025, SanDisk shares have surged roughly sixfold from $36 to $220 driven by AI hyperscaler demand for NAND. In the quarter ended Oct. 3, 2025 revenue rose 23% year-over-year and 21% sequentially; GAAP EPS fell 33% year-over-year but rose over fourfold sequentially, and results beat estimates with guidance well above expectations. Management guided current-quarter revenue of $2.55–$2.65 billion (midpoint $2.6B) and non-GAAP EPS $3.00–$3.40 (midpoint $3.20), implying sequential revenue growth of ~12.6% and EPS growth of ~162%; Wall Street now pegs FY26 EPS at $12.81 and FY27 at $20.21 while the stock trades at ~17x forward earnings. Analysts cite multi-quarter capacity constraints for memory as a tailwind and note potential upside from further beats, S&P 500 inclusion, or M&A interest, supporting the case for additional upside into 2026.

Analysis

Market structure: SanDisk (SNDK) is a near-term beneficiary of asymmetric NAND pricing vs. limited supply — hyperscalers (Google/Azure/Meta) and server OEMs win from denser, cheaper storage per AI model while NAND suppliers (SNDK, WDC) capture outsized margin expansion. Expect pricing power to persist for several quarters: Morgan Stanley’s view implies capacity shortfall 2–6 quarters out, supporting ASPs and supplier free cash flow through mid-2026. Cross-asset: stronger memory cash flow narrows tech credit spreads, supports high-yield tech debt, and should modestly lift TWD/KRW vs. USD on better export receipts; HDD commodity and oil effects are second-order.

Risk assessment: Key tail risks include rapid hyperscaler inventory digestion or aggressive capacity additions that could compress NAND ASPs by >20% within 3–8 quarters, and geopolitical/export controls that reroute demand or suppliers. Near-term (days–weeks) volatility centers on the Feb 2026 earnings print; short-term (months) hinges on successive beat-and-raise quarters; long-term (2026–27) depends on FY27 EPS realization ($20.21 consensus) and potential M&A. Hidden dependency: SNDK’s margin upside is concentrated in a few large buyers — any change in hyperscaler procurement cadence is a high-leverage risk.

Trade implications: Tactical: prefer a calibrated long exposure to SNDK (2–5% of portfolio) funded by trimming frothy, high-P/E AI software names. Use capital-efficient option structures: buy one-year LEAP calls (Jan/Jun 2027) or sell 90-day 15% OTM cash-secured puts to collect premium and secure entry. Pair trade: long SNDK vs. short WDC or a broad semiconductor ETF (e.g., SOXX) to isolate NAND-specific upside while hedging beta; target hedge ratio 0.7–1.0.

Contrarian angles: Consensus underestimates cyclicality — memory historically re-rates up quickly then reverses when capex ramps (DRAM cycle 2017–19 parallel). The S&P 500 inclusion and takeover chatter can create crowding; if one large buyer reduces orders, multiple compression could erase >30% quickly. Actionable edge: require two consecutive beat-and-raise quarters before scaling beyond 5% position and cut if forward EPS revisions roll >10% down or NAND ASPs fall >10% QoQ.

Key Decisions for Investors

• Establish a 2–3% portfolio long position in SNDK ahead of Feb 2026 earnings; scale to 5% only after two consecutive beat-and-raise quarters. Set a hard stop: reduce to 1% or exit if next-quarter revenue or EPS guidance misses midpoint by >5% or non-GAAP EPS < $2.80.
• Execute a paired relative-value trade: long SNDK (2%) and short WDC (1.5%) to isolate NAND pricing upside vs. legacy HDD/other exposures; rebalance if the spread narrows by >20% or if SNDK EPS revisions exceed +15%.
• Use options to express asymmetric upside: buy SNDK Jan 2027 LEAP at-the-money calls (allocate up to 1% notional) OR sell 90-day cash-secured puts 15% OTM to collect premium and obtain entry; close if implied vol drops >30% or if puts are assigned and revenue guidance misses.
• Reduce exposure (by 3–5%) to high-P/E software/AI names (non-infrastructure) and redeploy into memory suppliers (SNDK, consider selective NVDA/MSFT exposure) to capture infra-led earnings re-rating; reassess in 90 days based on NAND ASP indices and hyperscaler capex commentary.
• Monitor three specific indicators over the next 60–120 days before adding size: 1) Bloomberg NAND/flash ASP index (watch for >5% QoQ moves), 2) hyperscaler capex commentary (Google/Azure/Meta reports), and 3) SNDK sell-side EPS revisions (add if consensus +10%+ for FY27; trim if revisions -10%+).