Western Digital vs. Sandisk: The Better Way to Play the Hard-Drive Side of the AI Storage Boom

Sandisk reported fiscal Q4 2026 revenue of $8.9 billion, up 372% year over year, and non-GAAP EPS of $39.25 versus $0.29 a year earlier, supported by NAND flash demand from AI data centers. It guided fiscal Q1 revenue to roughly $10.5 billion, up 453%, EPS to about $45, and gross margins of 83%-85%; management expects the NAND market to approach $500 billion in 2027. Western Digital also posted strong results, with Q4 sales up 44% to $3.8 billion, EPS more than doubling to $3.56, and Q1 revenue guided to $4.1 billion, but the article favors Sandisk for higher AI-memory upside while noting its greater exposure to NAND pricing cycles.

Analysis

The key investment question is not AI-storage demand but whether current NAND economics can survive the inevitable supply response. SNDK's profitability now embeds an unusually favorable pricing/spread environment; even a modest normalization in NAND contract pricing would create disproportionate EPS downside because fixed-cost absorption and gross margin are at cyclical highs. The relevant confirmation data over the next 1-3 months are enterprise-SSD contract prices, hyperscaler inventory commentary, and announced wafer-capacity additions from Samsung, SK Hynix and Kioxia—not management's TAM framing.

WDC offers a cleaner risk-adjusted expression of capacity-constrained nearline HDD demand. HDD supply is effectively a two-player market with STX, and qualification cycles make substitution difficult for hyperscale customers; this should support pricing for several quarters even if AI server capex moderates. A second-order beneficiary is STX, whose earnings revisions may lag WDC despite nearly identical industry structure; conversely, NAND-heavy vendors face a sharper inventory correction if cloud customers shift from buildout to utilization.

Consensus appears to be extrapolating SNDK's peak earnings rather than discounting a memory-cycle reversal. The stock's extreme trailing appreciation makes it vulnerable to a "beat-and-raise but lower pricing" reaction, while WDC can rerate on durability of free cash flow rather than further margin expansion. Over 6-18 months, AI inference may favor high-capacity HDDs for colder data lakes, while flash retains the higher-growth but more commoditized performance-storage opportunity.

Key Decisions for Investors

• Initiate a 3-6 month relative-value position: long WDC / short SNDK, dollar-neutral. The thesis is margin-duration divergence rather than an outright AI slowdown; target 15-25% relative outperformance. Exit if SNDK sustains gross margin above 80% while NAND contract prices continue rising for two consecutive quarters.
• Add STX on weakness as a secondary HDD oligopoly exposure, preferably versus SOXX to isolate storage pricing from broad semiconductor-beta risk. Reassess if hyperscalers signal reduced nearline-HDD exabyte shipments or if WDC/STX pricing discipline breaks through incremental supply concessions.
• Do not chase SNDK outright after the parabolic move. For investors requiring upside exposure, wait for independent confirmation of enterprise SSD pricing and buy defined-risk call spreads only after a post-results consolidation; the missing input is SNDK's valuation versus normalized, mid-cycle earnings.
• Set alerts around NAND spot/contract-price declines and capex announcements by Samsung, SK Hynix, Micron (MU), and Kioxia. A sequential NAND-price decline or accelerated capacity additions would be the earliest catalyst to increase the WDC/SNDK relative short.