A Billionaire Investor Just Increased His Exposure to Memory Makers Sandisk and Micron. Should Investors Buy the Stocks?

Sandisk and Micron are benefiting from surging NAND/DRAM prices driven by AI-related demand and supply constraints, with Citi citing NAND storage prices up 186% this year and SSD prices rising even faster. Both companies are also beginning to sign 3- to 5-year deals for the first time, improving visibility and potentially lifting earnings floors; valuations remain low at 6.5x and 7.5x forward P/E on fiscal 2027 estimates. The article is broadly constructive on the memory cycle, though it emphasizes these stocks remain cyclical and require monitoring.

Analysis

The important signal is not just tighter supply; it is that the memory industry is experimenting with contract duration discipline for the first time. That changes the cash-flow regime from pure spot beta to something closer to a semi-annuity, which should compress downside volatility and justify higher terminal multiples if the market believes capacity rationality is enforceable. The second-order effect is that capex allocation across the broader semiconductor stack may finally favor returns over share gain, which could keep supply growth muted for longer than the traditional cycle model assumes.

The underappreciated winner is the ecosystem that consumes memory-heavy AI workloads, not just the memory vendors themselves. As inference scales, memory intensity rises faster than compute intensity, so the bottleneck shifts toward bandwidth and storage economics; that supports continued pricing power in HBM/DRAM and NAND, but also raises the total cost of AI deployment for hyperscalers and enterprise buyers. Over time, that can force model providers to optimize architectures, which could slow near-term demand elasticity for leading-edge GPU deployments while reinforcing demand for memory suppliers.

The main risk is that current pricing strength induces a supply response with a lag of several quarters, not weeks. Because memory profitability is so sharply levered, even a modest capex reset can create a future overshoot in supply that the market won’t price until margins peak; this is the classic trap in cyclical semis. The current setup works best over the next 6-12 months, but the market is likely underestimating how quickly sentiment can reverse once contract pricing stabilizes and inventory normalization begins.

The contrarian view is that the market may be paying too much attention to the improving floor and not enough to the ceiling on earnings quality. If the new long-term deals primarily lock in volume rather than true price discipline, the apparent de-cycling could be temporary, and consensus may be extrapolating a structurally higher multiple too aggressively. That argues for owning the better balance-sheet and execution stories, but staying tactical rather than treating either name as a secular compounder.

Key Decisions for Investors

• Long MU vs short a basket of cyclically exposed semiconductor names for the next 6-9 months; thesis is that DRAM/HBM pricing and contract visibility improve the earnings floor faster than peers can re-rate, with downside protected by current low multiples.
• Buy SNDK on pullbacks only, using a 3-6 month horizon; upside is continued NAND pricing and margin expansion, but position size should be smaller than MU because pure-play NAND remains the more reflexive part of the cycle.
• Sell covered calls against existing MU/SNDK longs into strength over the next 1-2 quarters; implied volatility should remain elevated, and premium harvest helps offset the risk of a late-cycle reversal.
• Avoid chasing NVDA as a second-order beneficiary here; if memory costs keep rising, GPU system-level economics become less attractive, which can create a relative headwind even if unit demand stays strong.