A Once-in-a-Decade Buying Opportunity: Sandisk and Micron Shares Are Dirt Cheap and Look Primed to Skyrocket

The article highlights that Micron and Sandisk are benefiting from sustained memory price strength tied to the AI data center boom, with Sandisk up nearly 600% YTD 2026 and Micron up nearly 250%. It argues both stocks trade cheaply versus forward earnings—Sandisk at ~7x forward earnings and Micron at ~6x—and could re-rate materially if tightness persists into 2027 (Micron expects market tightness to last until after 2027). The piece also notes supply is not expected to meaningfully increase until late 2027–2028, which may prolong the pricing backdrop even as new capacity comes online.

Analysis

This is a classic operating-leverage setup, but the better risk/reward is not symmetric. MU has the cleaner claim on scarcity because DRAM tied to accelerator builds is harder to substitute than NAND, so each incremental bit of supply tightness should flow disproportionately into EPS and FCF. SNDK is the more reflexive trade: pure NAND exposure means stronger beta on the way up, but also a faster multiple reset if customers slow buys or channel inventory normalizes.

The second-order loser is not just hyperscalers; it is the whole AI hardware stack that must fund memory inflation before revenue catches up. That can compress deployment cadence at OEMs and cloud buyers, while helping memory vendors defend pricing through 2027 if the allocation game persists. Near term, the market will trade quarterly commentary on contract duration and inventory days; a single hint that customers are pre-buying beyond demand can extend the rerate.

Contrarian take: the cheap P/E may be a cyclical mirage, not a bargain. Memory stocks usually look optically inexpensive right before the market starts discounting peak earnings, so the key falsifier is flattening ASPs or rising inventories over the next 1-2 quarters. If that happens, the multiple can compress faster than earnings grow, especially in SNDK where the move is already fully extended.

Key Decisions for Investors

• Long MU on any 3-5% pullback; 3-6 month horizon. Prefer MU over SNDK because DRAM scarcity has the longer duration and less substitution risk. Target: add if forward P/E stays under ~8x while estimates continue rising; stop if management signals 2027 ASP deceleration or inventory builds.
• Avoid chasing SNDK after the vertical run; if already long, trim into strength and use only on post-earnings pullbacks. 1-3 month risk/reward is worse than MU because NAND is the faster mean-reverting leg of the cycle.
• Relative value: long MU / short SNDK into the next earnings window. The thesis is that MU's mix is more durable, while SNDK has more multiple vulnerability if the market starts pricing peak-cycle earnings. Hedge ratio should be volatility-weighted; cover if NAND pricing remains tighter than expected.
• Watch hyperscaler capex disclosures over the next quarter as the key catalyst path. If memory spend is still rising as a percentage of AI infrastructure budgets, the trade can run for months; if cloud capex growth slows materially, take profits quickly because memory pricing will lag the slowdown only by one to two quarters.
• Set an alert on memory ASPs and inventory days at MU/SNDK; any two consecutive quarters of softer pricing is the main thesis breaker. That would argue for de-risking the entire memory basket rather than trying to pick a bottom.