RBC Capital reiterates Micron stock rating on supply outlook

Micron forecast a stronger fiscal first quarter and delivered fiscal Q4 2026 adjusted EPS of $33.42 on $54.23B of revenue, above consensus estimates of $31.16 and $50.45B, respectively. AI-driven memory demand—especially high-bandwidth memory—remains robust through 2028, with more than 75% of fiscal 2027 shipments secured and 26 supply-chain agreements covering over 35% of revenue through 2030. RBC reiterated its Outperform rating and $1,500 price target, citing tightening DRAM supply-demand conditions, expected fiscal 2027 margin expansion, and potential majority free-cash-flow buybacks after Micron reaches its cash target.

Analysis

The investable read-through is not simply higher AI-memory volumes: contracted capacity shifts MU’s earnings profile from spot-DRAM cyclicality toward a scarcity-rent model, which can justify a higher trough multiple if contract terms include enforceable minimums rather than nonbinding reservations. HBM wafer allocation also removes supply from conventional DRAM, creating a second earnings lever in mobile and PC memory; this benefits MU and SK Hynix, while Samsung’s broader commodity-memory exposure makes it a less pure scarcity trade. The key diligence item is the share of agreements containing fixed prices or price caps—floor pricing protects downside but could surrender upside if HBM pricing remains tight.

Near term, the market is likely to focus on the modest gross-margin friction and whether buybacks begin on schedule; after a parabolic move, even a strong print can sell off if the implied pricing trajectory does not rise. Over 1-3 months, HBM qualification progress, customer concentration, and capex guidance from MU, SK Hynix and Samsung will determine whether consensus extends the upcycle. Over 6-18 months, new supply and any faster-than-expected yield improvement are the central risks: memory cycles reverse when industry capex responds to peak returns, not when demand first weakens.

Consensus may be underweight the possibility that AI systems become memory-bandwidth constrained, making HBM content growth more durable than accelerator unit growth. Conversely, the bullish framing likely overstates visibility if supply agreements are primarily capacity reservations; a slowdown in hyperscaler capex could leave customers legally able to defer volumes. The thesis is falsified by sequential HBM pricing erosion, inventory days rising for two consecutive quarters, or a material upward revision to industry DRAM bit-supply growth.

Key Decisions for Investors

• Maintain/add MU exposure only on post-earnings volatility rather than chase strength; use a 6-12 month horizon and size against a defined downside of a 15-20% multiple reset if gross-margin expansion or contracted-volume disclosure disappoints.
• Express the memory-scarcity thesis as long MU / short Samsung Electronics (005930 KS) on a 3-6 month horizon, subject to borrow and liquidity review. MU has greater operating leverage to high-value memory, while Samsung has more exposure to a normalization in commodity DRAM/NAND; close if Samsung demonstrates sustained HBM yield and customer qualification gains.
• Use a staged MU call spread, financed only after confirming implied volatility is below post-results realized volatility: buy 6-9 month moderately out-of-the-money calls and sell strikes 20-30% higher. This preserves upside to estimate revisions while limiting premium at a point where valuation sensitivity is elevated.
• Set an earnings diligence trigger: reduce exposure if management cannot quantify binding take-or-pay terms, pricing collars, and customer concentration for its contracted capacity. Treat the agreements as lower-quality revenue visibility until those details are independently supported.
• Monitor WDC and Sandisk as a secondary read-through rather than direct substitutes: stronger DRAM economics can divert industry capital away from NAND, but NAND longs require separate evidence of supply discipline and pricing recovery.