Micron (MU) Q4 2026 Earnings Call Transcript

Micron reported fiscal Q4 revenue of $54.23 billion and non-GAAP EPS of $33.42, propelled by AI-related server and storage demand; operating cash flow reached $43.97 billion and adjusted free cash flow was $33.20 billion. For fiscal Q1 2027, the company guided revenue to $60 billion-$63 billion, non-GAAP gross margin to about 86.3%, and EPS to $37.15-$39.15, with more than 75% of fiscal-2027 shipments already allocated or contracted. Management expects AI and agentic-workload demand, higher HBM pricing, and constrained industry supply to support tight memory markets through 2028, while raising fiscal-2027 R&D spending by more than $1 billion and directing capex toward clean-room capacity in Idaho and Singapore.

Analysis

MU’s earnings power is becoming less levered to spot DRAM than prior cycles: contracted volume and price-band structures improve downside visibility, but also cap some upside if allocation pricing continues to outrun negotiated ceilings. The more important valuation question is whether investors capitalize current margins as cyclical peak earnings or as a multi-year scarcity rent; confirmation of sustained forward contract repricing over the next 1-3 months would support multiple expansion rather than merely higher EPS estimates.

The second-order beneficiary is NVDA: memory scarcity raises system cost, but it also protects accelerator ASPs and makes qualified HBM supply strategically valuable. The risk is that customized HBM increases customer concentration and shifts bargaining power toward NVDA over time; MU’s differentiation must translate into durable qualification wins, not simply a one-generation supply shortage. Samsung Electronics (005930 KS) and SK Hynix are the principal competitive threats because incremental HBM qualification or faster node yields would compress the scarcity premium before new clean-room capacity arrives.

Near term, the setup remains positive but crowded after guidance. The key 1-3 month catalyst is evidence that contract pricing resets remain favorable while cost declines lag ASP gains; the 6-18 month risk is demand normalization in client/mobile and AI infrastructure digestion, which would expose the operating leverage embedded in elevated margins. Treat the reported figures as management claims until the 10-Q and earnings release reconcile the unusually large cash-flow and revenue figures; a mismatch is a no-trade trigger.

Contrarian view: the market may be underestimating the extent to which higher memory prices ration lower-end PC and handset demand rather than create incremental industry revenue. If OEM unit weakness broadens beyond premium devices, MU can still post strong near-term results, but investors will discount the terminal margin earlier than consensus expects.

Key Decisions for Investors

• Maintain/add a tactical long MU only after SEC-filed results validate the reported guidance and cash-flow figures; target a 1-3 month holding period through the next pricing/contract update. Thesis is upward FY27 EPS revisions and a cyclical-to-structural rerating; exit if gross-margin guidance falls below 80% or management indicates allocation releases are accelerating.
• Express the AI-memory bottleneck with a long MU / short Samsung Electronics (005930 KS) pair over 3-6 months, sized beta-neutral. MU has greater direct exposure to high-value server memory mix, while Samsung carries more exposure to weaker consumer-memory and handset ecosystems; close the spread if Samsung demonstrates material HBM qualification gains or MU loses a major accelerator design win.
• For a lower-single-name-risk expression, buy SOXX and fund with an underweight/short XLP or broad defensives only if the portfolio needs AI-capex exposure; do not chase MU calls immediately after a large gap because implied volatility and peak-margin risk are likely elevated.
• Set alerts for hyperscaler capex guidance, NVDA platform qualification commentary, and DRAM contract-price data over the next 90 days. Any evidence of broad AI rack de-specification, falling server-memory content, or a meaningful Chinese supplier yield breakthrough would falsify the tight-supply thesis and warrants reducing MU exposure.