Sandisk: Analyst Day Changes The Way I Value The NAND Cycle (Upgrade)

SanDisk’s NBMs underwrite $93.9B of minimum revenue, with more than half of fiscal 2027 bits already covered. Management expects about two-thirds of fiscal 2028 bits under multiyear agreements, which could dampen SanDisk’s historical NAND earnings volatility. The datacenter mix has risen from ~12% to 38% as AI inference drives structural NAND demand.

Analysis

The real signal here is not near-term revenue, it’s a step-change in cash-flow visibility. If a larger share of output is contractually spoken for, the market should stop valuing SNDK as a pure spot-price proxy and start assigning a lower earnings discount rate, which is the setup for multiple expansion rather than just higher EPS. That rerating would most likely come at the expense of more cycle-sensitive memory names that still trade on peak/trough assumptions, especially peers with less contracted enterprise exposure.

The datacenter mix shift suggests AI is adding a durable storage layer, not just a one-time compute burst. Inference workloads tend to create recurring demand for read-intensive capacity, caching, and model-logging, which supports steadier bit demand over 6-18 months even if consumer memory stays weak. Second-order winners are enterprise SSD ecosystem players and hyperscale storage vendors; second-order losers are consumer/mobile NAND businesses that remain tied to slower replacement cycles and price elasticity.

The key risk is that ‘contracted’ does not equal ‘immune’: if hyperscalers pause capex, they will renegotiate volume or push out takes, and the visible backlog can become a false floor. Over the next 1-3 months, the catalyst is whether management can convert contract visibility into margin guidance that actually narrows estimate dispersion; if not, the stock can give back gains once the headline fades. What would falsify the thesis is any sign that datacenter mix stalls, renewal rates soften, or ASP assumptions remain too dependent on a tight spot market.

Key Decisions for Investors

• Initiate a modest long SNDK position on pullbacks, with a 3-6 month horizon; thesis is multiple expansion from reduced NAND volatility rather than a near-term EPS beat.
• Prefer a defined-risk call spread on SNDK over outright common if chasing the rerating narrative; use it to express upside while limiting damage if contract visibility is already priced in.
• Pair trade: long SNDK / short WDC or MU for relative-value exposure to contracted vs. more cyclical memory economics over the next 1-3 months.
• Set an exit alert if the next guidance cycle fails to show margin stabilization or if datacenter mix growth decelerates; that would imply the market is overvaluing the backlog.