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

Micron and Sandisk are benefiting from AI-driven memory chip shortages, with Wall Street projecting FY2027 revenue growth of 63% for Micron and about 122% for Sandisk. Both stocks trade at roughly 9x forward earnings, suggesting valuation remains reasonable despite last year’s explosive gains. The key risk is that if memory supply catches up to demand, pricing could fall and hurt results.

Analysis

The market is treating memory like a pure scarcity trade, but the second-order beneficiary is the hyperscaler capex cycle itself: every incremental dollar of AI infrastructure spend carries a hidden tax in storage and working-memory content, so the tighter the supply, the more entrenched the spend becomes. That makes MU the cleaner operating leverage play because it spans both NAND and DRAM, while SNDK is more directly exposed to a single-product bottleneck and thus has a higher beta to any short-term pricing spike.

What the consensus is missing is that this is less about near-term unit growth and more about the duration of elevated pricing before capacity comes online. Once the industry sees evidence of new wafer starts, inventory restocking will likely normalize faster than end-demand, which typically creates a sharp margin inflection 2-4 quarters before revenue growth visibly rolls over. That timing asymmetry matters: the stocks can look cheap on forward earnings right up until the market starts discounting the next downcycle.

The biggest risk is that investors are extrapolating spot-like memory economics into FY27 and beyond, but memory is notorious for mean reversion once capital spending follows margins. If hyperscaler capex pauses even modestly for one quarter, the leverage works both ways and these names can de-rate quickly despite still-strong reported growth. The favorable setup is real, but it is a trade on supply discipline, not a permanent rerating of the franchise.

From a positioning standpoint, the better expression is not a blind long but a controlled duration trade: stay long the leaders while using options or relative value to manage the eventual supply response. The market is still underpricing the possibility that the “winner” of the shortage phase is also the first to retrace when inventories normalize, especially in the more commoditized NAND stack.

Key Decisions for Investors

• Long MU vs. short a broad semis basket for 3-6 months; MU has cleaner dual-memory exposure and better optionality if DRAM remains tighter than NAND, while the short basket dampens sector beta.
• Buy MU Jan-2027 call spreads on pullbacks for a 2:1 to 3:1 payoff; the trade captures continued scarcity premium while limiting downside if memory pricing mean-reverts earlier than expected.
• Trade SNDK tactically only on strength in channel checks; trim or hedge if lead times stop extending, since single-product exposure makes it the faster de-rating candidate when supply catches up.
• Use a pairs trade long MU / short NVDA into any AI capex disappointment; if hyperscaler budgets pause, memory may still outperform compute hardware because storage demand tends to lag less aggressively than accelerator orders.
• Set a hard review trigger on any announced capacity adds from major memory producers; if visible supply expansion accelerates, reduce gross exposure by 30-50% within one earnings cycle.