Micron Says Memory Supply Will Be Much Tighter in 2028. Sandisk's Contracts Limit How Much It Can Gain.

Sandisk shares fell 23% in Q3 to roughly $1,740, despite remaining up about 629% year-to-date, as NAND pricing growth and the company’s revenue expansion have decelerated from exceptionally high levels. Micron expects NAND supply-demand conditions to remain tight through 2028, supporting the durability of Sandisk’s earnings, but Sandisk’s fiscal Q1 revenue-growth guide is only 15%-20% sequentially and gross-margin guidance of 83%-85% suggests profitability has plateaued near record levels. Long-term contracts are expected to cover about two-thirds of fiscal 2028 bit sales, providing downside price floors but limiting upside through price ceilings; at roughly 8x forward earnings versus 19x for the S&P 500, the stock is cheaper but still carries post-shortage earnings risk.

Analysis

The key valuation distinction is not NAND tightness but incremental participation: SNDK's growing contracted mix converts a cyclical spot-price business into a lower-volatility cash-flow stream, while simultaneously limiting upside convexity. MU retains more favorable exposure to industry pricing and has additional earnings levers outside NAND, making it the cleaner vehicle for a further shortage extension over the next 1-3 months. SNDK should therefore trade less on spot NAND momentum and more on whether investors assign a higher multiple to its protected earnings floor.

The second-order implication is that sustained allocation supports NAND suppliers' pricing discipline but delays unit growth for SSD/controller ecosystems. Silicon Motion (SIMO) and Phison (8299.TW) could face shipment constraints even if content per device rises; conversely, equipment names such as LRCX, KLAC and AMAT do not necessarily benefit until suppliers decide that contracted demand justifies capacity additions. A prolonged shortage can be negative for those capex names if producers prioritize returns and avoid the historical overbuild cycle.

Contrarianly, the market may be underpricing the embedded downside protection in SNDK's contract floors, not merely overpricing the capped upside. That thesis requires verification of the actual pricing corridors, customer concentration, volume flexibility and counterparty remedies; without those disclosures, an apparently cheap earnings multiple may simply reflect uncertainty around normalized post-contract economics. Falsify the durability case if SNDK guides gross margin below 80%, reports meaningful contract-volume deferrals, or if MU's NAND pricing commentary turns flat-to-down before the next two earnings cycles.

Key Decisions for Investors

• Run a beta-adjusted 1-3 month long MU / short SNDK pair rather than an outright NAND short: MU has broader memory-cycle optionality, while SNDK's contract ceilings should suppress relative earnings upgrades if spot pricing remains tight. Cover the short leg if SNDK discloses materially wider price bands or contract economics that support earnings above current consensus through fiscal 2028.
• Do not initiate a core SNDK long before the next earnings release unless management quantifies contract floors, ceilings and customer concentration. Upgrade to a 6-18 month long only if contracted cash flows demonstrably protect an 80%+ gross-margin profile through a moderate NAND correction; this would support multiple expansion from a perceived peak-earnings valuation.
• Use SIMO and NAND-controller suppliers as a negative read-through watchlist, not a short recommendation, over the next quarter. A widening gap between NAND supplier revenue growth and controller unit shipments would confirm allocation is shifting economics upstream; improving controller volumes would challenge that thesis.
• For semiconductor-equipment exposure, prefer neutral/underweight LRCX, KLAC and AMAT versus memory suppliers until announced NAND wafer-capacity additions accelerate. The catalyst to reverse this stance is concrete 2027 capacity guidance from major NAND producers rather than continued ASP strength alone.