Sandisk and Micron Dominated the First Half of 2026. Which Is the Best Buy Now?

Micron (+325% YTD) and SanDisk (nearly +900% YTD) are benefiting from AI-driven demand for DRAM and NAND, with data center capex projected at $3T–$4T annually by 2030. Street estimates call for Micron revenue growth of 78% next fiscal year and SanDisk 122%, with forward P/E still below typical large-cap tech levels (Micron ~8x, SanDisk ~13x vs mid-20s). The article argues there’s “room for upside,” projecting SanDisk could double and Micron could triple if prices re-rate, despite ongoing supply-demand tightness.

Analysis

The clean takeaway is not that memory is cheap, but that the market is re-rating the duration of the upcycle. In the next 1-3 months, the main beneficiaries are the names with the tightest exposure to AI server BOM inflation and the most operating leverage to pricing—MU first, then SNDK—while the second-order loser is hyperscaler capex efficiency: every incremental dollar of memory spend raises the hurdle for software and networking returns, which can eventually compress enthusiasm for adjacent AI names.

The more interesting issue is that the move is already behaving like a scarcity trade, so the next leg depends on whether pricing power survives the first wave of supply response. If DRAM and NAND remain firm into the next two earnings prints, the multiples can keep expanding; if not, these stocks can de-rate very quickly even with decent unit growth because memory earnings are highly sensitive to ASPs, not just shipment volumes.

Contrarianly, the market may be underestimating how much of the easy upside is already in the stocks after the year-to-date move. SNDK has the cleaner pure-play scarcity story, but that also makes it more vulnerable if NAND supply normalizes first; MU is better diversified but could underperform on any DRAM-specific soft patch. Over 6-18 months, the key falsifier is not demand rhetoric but visible inventory builds, weaker guide for cloud capex, or a turn in contract pricing.

I would frame this as a cyclical momentum trade, not a structural compounder thesis. The best risk/reward is to own pullbacks, not chase strength, and to hedge with a basket short of broader AI infrastructure where memory inflation eventually leaks into margins.

Key Decisions for Investors

• Buy MU on a 5-10% pullback; target a 3-6 month trade into the next two earnings cycles, with thesis invalidation if DRAM pricing or gross margin guide rolls over.
• Prefer SNDK only as the higher-beta expression of the same theme; use smaller sizing than MU because single-segment exposure makes it more vulnerable to a NAND price snapback within 1-2 quarters.
• Pair trade: long MU / short a basket of AI capex beneficiaries with weaker pricing power (e.g., NVDA or AMD on strength) to isolate memory scarcity from broader AI exuberance; hedge if hyperscaler capex growth slows.
• Set a watch item on cloud capex and memory contract pricing; if either shows sequential weakening, reduce exposure immediately because the re-rating can unwind faster than the fundamental slowdown.
• If already long the names, consider selling upside calls against positions into momentum spikes to monetize elevated implied volatility while preserving core exposure.