Micron and Sandisk Are Up Big During the Past Year. Can the Run Continue?

Micron and Sandisk are benefiting from an AI-driven memory chip shortage, with demand from data center build-outs pushing prices and earnings higher. Wall Street expects FY2027 revenue growth of about 63% for Micron and 122% for Sandisk, while both stocks still trade near 9x forward earnings. The article is constructive on fundamentals and valuation, but notes that any easing in supply-demand tightness could pressure prices and shares.

Analysis

The market is still underpricing how long memory tightness can persist once hyperscaler capex becomes embedded in procurement plans. The second-order winner is not just MU and SNDK, but the entire memory ecosystem: substrate, test equipment, packaging, and enterprise SSD controller suppliers should see pricing power lag the commodity move by a quarter or two, creating a window where upstream vendors re-rate before memory ASPs peak.

What matters is that this is a classic supply response trap. Memory makers can announce capacity, but meaningful wafer output and qualification for data-center grade parts take multiple quarters, so the earnings inflection can stay intact well into 2027 even if end-demand moderates. The real risk is not a near-term slowdown; it is a delayed oversupply event in late 2027/2028 once every producer chases the same margin pool, which historically compresses multiples before spot prices fully roll over.

Consensus is focused on valuation being cheap, but the more interesting point is that cheap forward P/E is a poor signal in an upcycle because earnings are being mechanically inflated by commodity scarcity. That makes these names attractive tactically, but fragile strategically: if guidance stops stepping up, the multiple can compress even while current fundamentals look fine. NVDA and GOOGL are beneficiaries through capex intensity, but they are also the canaries—if hyperscaler spending is rephased, memory demand would be one of the first pockets to soften.

The contrarian setup is that the best risk/reward may be in expressing the thesis through a barbell rather than outright beta. Long memory can work for the next 2-4 quarters, but the cleaner asymmetry may be selling downside in the form of put spreads into any post-earnings volatility, while fading late-cycle optimism with a longer-dated short against broader semis once the market starts pricing in eventual capacity normalization.

Key Decisions for Investors

• Maintain tactical longs in MU and SNDK for the next 3-6 months, but trim into strength above consensus FY27 numbers; risk/reward remains favorable while pricing power is still rising, but upside is increasingly tied to a narrow supply narrative.
• Pair trade: long MU / short a semis basket with less direct memory exposure (e.g., SOXX via ETF structure) over 2-4 quarters; this isolates the scarcity premium while reducing exposure to a broader semiconductor multiple reset if AI capex growth normalizes.
• Use options to express upside with defined risk: buy 6-9 month call spreads in SNDK rather than stock, targeting another 20-30% upside with limited premium outlay; the stock is likely to remain volatile around supply commentary.
• Watch for a capacity-announcement inflection in H2 2026; if industry capex steps up faster than expected, use that as a signal to hedge or exit 30-50% of the long exposure, since the trade can unwind quickly once supply visibility improves.
• Do not chase NVDA or GOOGL on this headline; they are beneficiaries but not the cleanest expression. Use them only as sentiment indicators for data-center spend durability, not as primary longs on the memory thesis.