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 DRAM and NAND pricing driven by AI demand and supply constraints, with Citigroup forecasting NAND storage prices up 186% this year. Both companies are also starting to sign unprecedented 3- to 5-year supply deals, which should improve visibility and raise earnings floors. Despite the strong run, both stocks still screen cheaply at forward P/E multiples of 7.5x for Sandisk and 6.5x for Micron based on fiscal 2027 estimates.

Analysis

The market is starting to re-rate memory from a pure spot-priced commodity into a quasi-contractual cash-flow stream. That matters because if 3-5 year supply agreements become standard, the earnings multiple should expand even if end-market unit growth stays lumpy: the real change is lower earnings volatility, not just higher near-term ASPs. The second-order winner is not just the memory makers, but the downstream ecosystem that depends on predictable allocation—hyperscaler procurement, server ODMs, and storage controllers should see fewer surprise shortages and less emergency pricing.

The more interesting trade is that AI has likely lengthened the cycle rather than eliminated it. HBM demand is soaking up wafer capacity that would otherwise normalize DRAM supply, so any recovery in conventional memory is now running into a structurally tighter base. That creates a multi-quarter window where both DRAM and NAND can stay disciplined; the risk is that the market extrapolates this into a permanent regime just as capex budgets re-accelerate and Chinese supply catches up on a 12-18 month lag.

Consensus is still treating memory like a boom-bust beta trade, but the underappreciated shift is pricing power with contracts. If true, the correct framing is not "are we late?" but "how much of the margin uplift is now durable enough to justify a higher trough multiple?" That said, the asymmetry cuts both ways: once customers lock in volumes, any demand wobble will show up first in deferred orders rather than price—so the downturn may be later, but when it comes, it could be sharper in share price because positioning is now crowded.

Tepper’s participation is a signal, but not a timing tool; the better read is that large, patient capital is willing to underwrite a longer cycle than the street expects. The key catalyst to monitor over the next 1-2 quarters is whether contract language, volumes, and prepayment structures continue to improve; if they do, these names deserve a step-up in valuation. If not, the market could quickly revert to viewing current margins as peak-cycle earnings.

Key Decisions for Investors

• Long MU vs short a basket of high-beta semi names over 3-6 months: MU has the cleanest benefit from HBM-driven tightness, while the short leg helps isolate memory pricing duration rather than broader AI capex beta.
• Buy SNDK on pullbacks and hold 6-12 months, but size modestly: the pure-play NAND setup has the highest operating leverage if contract pricing sticks; risk is faster supply normalization, so use a disciplined stop if gross margin expansion stalls.
• Consider a pair trade long SNDK / short WDC for 1-2 quarters: SNDK has cleaner exposure to the NAND re-pricing upside, while WDC carries more legacy complexity and less convexity to tightening pricing.
• For momentum accounts, sell out-of-the-money puts on MU with 60-90 day tenor only if implied vol remains elevated; the thesis is that earnings revisions can keep the stock supported, but the trade should be hedged because memory names can gap down hard on any supply chatter.
• Watch for a reversal trigger: if lead times flatten and capex guidance inflects higher, reduce longs aggressively—memory can lose 20-30% of its gain quickly once the market prices in a mid-cycle peak.