Sandisk's AI Supercycle Is Just Getting Started

Q2 revenue was $3.0B (+61% YoY) with gross margins at 51.1%, and management guided Q3 margins to 65–67% driven primarily by pricing expansion. Data center demand jumped 64% sequentially, while PCIe Gen5 SSD qualifications and BiCS8 QLC ramp support a multi-quarter growth pipeline. The industry remains supply-constrained (mid-to-high teens bit growth vs ~60%+ data center demand), underpinning late-cycle pricing strength.

Analysis

Supply inelasticity plus a big demand pulse creates a multi-quarter window of pricing optionality for NAND suppliers rather than a one-off beat — capex and process-node capacity responses take 12–24 months, so positive EBITDA leverage will front-run volume growth. That front-loading amplifies equipment and specialty-IP vendors’ order visibility (controllers, etch/deposition tools), creating a stretch where equipment vendors can re-rate on order cadence even if unit bit growth normalizes later.

But the ramp of higher-density QLC technologies is a double-edged sword: it lowers $/TB and can stimulate incremental demand, yet it also accelerates bit-supply elasticity once yields and controller firmware mature. If QLC mix share rises meaningfully over 2–4 quarters, ASP per usable TB could decelerate versus headline bit growth, pressuring gross margins for suppliers that can’t monetize migration to higher value tiers.

Second-order winners include controller/firmware specialists and backend test/pack players who see higher BOM complexity and tighter qualification cycles; losers are legacy HDD vendors and storage OEMs with thin NAND inventory pass-through where component cost volatility depresses OEM margins. The critical catalyst window is 1–3 quarters for demand confirmation (customer bill-of-materials and server OEM orders), and 3–12 months for potential supply relief — either could flip the tape.

Consensus risk is asymmetric: the market may underprice the speed at which customers adjust mix to cheaper QLC or the pace at which newer fabs add capacity, meaning current optimism is vulnerable to a supply-side re-rating. Conversely, if hyperscalers continue to prioritize capacity over unit-cost and maintain bookings, upside could be larger than headline models assume because of step-function contract renewals and buy-through clauses.

Key Decisions for Investors

• Long MU (Micron) equity, 6–12 month horizon — overweight into any sub-10% pullback. R/R: +50–70% upside if pricing/mix holds; -30–40% downside if supply ramps faster than expected. Consider pairing with short-dated call sales to fund LEAP purchases (calendar spread) to monetize elevated IV.
• Long KLAC or LRCX (equipment exposure), 9–18 months — allocate to capture durable capex backlog and higher ASP per tool. R/R: 25–40% upside with lower probability of cyclic downside because tool replacement/upgrade cycles are sticky; hedge with 5–10% position sizing.
• Relative pair: long MU / short NTAP (NetApp) or PSTG (Pure Storage), 3–9 months — captures memory supplier margin tailwind versus OEMs that face input-cost variability and slower pass-through. Target 2:1 upside/downside; tighten on signs of OEM contract re-pricing or hyperscaler destocking.
• Event hedge: buy 3–6 month puts on a major cloud provider (AMZN or MSFT) sized as 5–10% portfolio protection — tail insurance against demand shock from capex pullback that would quickly depress NAND pricing. Cost is small insurance versus concentrated memory exposure.