Sandisk Unveiled an $14 Billion Buyback and Warren Buffett Has Something to Say About It

Sandisk authorized a $14 billion share-repurchase program after buying back $4.5 billion of stock in fiscal Q4, signaling management's confidence in future cash generation despite the stock's 548% gain in 2026. Q4 revenue reached nearly $9 billion, up 372% year over year, while data-center revenue rose 437% and approached one-quarter of sales amid AI-driven NAND SSD demand. The article argues that high-margin NAND pricing, hyperscaler capex, and multiyear contracts could justify the buyback if shares remain below intrinsic value, though memory-sector cyclicality remains a key risk.

Analysis

The authorization is most consequential as a capital-allocation constraint, not as a valuation certificate. In a memory upcycle, management can mistake peak-cycle cash generation for durable free cash flow; deploying a large fraction of liquidity near the top reduces the balance-sheet flexibility needed when NAND pricing resets. The key diligence item is the funding mix: buybacks funded from internally generated FCF are accretive, while debt-funded repurchases would amplify downside through a cyclical earnings trough.

Near term, SNDK may benefit from mechanical support and a broader investor base focused on EPS growth rather than commodity-memory cyclicality. Over the next 1-3 months, the more relevant catalyst is disclosed execution pace versus average repurchase price, alongside enterprise SSD pricing, bit shipment growth, inventory days, and gross-margin guidance. A slow execution rate would indicate the authorization is principally signaling; aggressive execution after the rally raises the risk that management is shrinking float at unfavorable cycle-adjusted economics.

The underappreciated second-order effect is industry supply discipline. If SNDK prioritizes capital returns over capacity expansion, it supports NAND pricing and benefits Kioxia and Samsung Electronics' memory operations; it is less directly favorable to equipment suppliers such as LRCX, AMAT, and KLAC if industry wafer-fab-equipment budgets remain constrained. For NVDA, sustained storage bottlenecks can support data-center build urgency, but a hyperscaler capex pause would simultaneously weaken SSD demand and unwind the scarcity premium.

Consensus appears to treat AI-linked NAND demand as structurally insulated from the memory cycle. The falsifier is not a single weak quarter: it is sequential deterioration in enterprise SSD ASPs, rising channel inventory, or a material cut to hyperscaler capex plans. Those signals would likely compress both earnings estimates and the multiple, with the buyback unable to offset the de-rating.

Key Decisions for Investors

• Do not chase SNDK directionally until the next filing confirms net cash/debt, trailing FCF, and actual repurchase execution. Treat a debt-funded program or repurchases materially above management's stated intrinsic-value framework as a negative governance/cycle signal.
• For a 1-3 month tactical expression, consider a small long SNDK / short SOXX pair only after evidence of continued enterprise-SSD ASP strength and meaningful executed repurchases; this isolates company-specific capital-return and mix upside from broad AI-semiconductor beta. Exit if NAND pricing turns sequentially negative or SNDK cuts gross-margin guidance.
• Maintain a watch alert on LRCX, AMAT, and KLAC: confirmation that NAND producers are redirecting cash to buybacks rather than capacity should be a 6-12 month headwind to memory WFE order expectations, despite near-term AI optimism.
• Use MU as the cleaner liquid hedge for a long SNDK position, but size conservatively: MU's DRAM exposure means the spread can widen sharply if HBM/DRAM remains tighter than NAND. Reassess the pair at each memory-pricing update and hyperscaler earnings cycle.