Micron Capacity Bet Reflects Rising AI Memory Demand

Micron completed its US$1.8B acquisition of PSMC’s P5 Tongluo site and announced a Phase 2 build adding ~270,000 sq ft of cleanroom space to start by end of FY2026, with retrofit shipments expected in FY2028. Shares jumped to $445.35 in premarket (+$19.22, +4.51%) on March 16, lifting YTD performance to ~+49.3% and one‑year gains to >350%. Q1 FY2026 revenue was $13.64B with net income $5.48B (EPS $4.78 vs est. $3.96); market cap ~ $479.6B, trailing P/E 40.47 and forward P/E 13.30, and Wedbush recently raised its PT to $500, underscoring bullish analyst sentiment and likely meaningful supply-side implications for DRAM/HBM markets.

Analysis

The market is treating AI memory as a durable moat; the less-obvious consequence is a two-stage value chain uplift where capital goods and local Taiwan ecosystem participants capture outsized cashflow before wafer-level revenues normalize. Memory equipment vendors (deposition, etch, metrology) will see multi-year order phasing that is far less lumpy than wafer output, so their revenue visibility and margins should lead DRAM makers’ earnings re-rating by 6–24 months.

A supply wave that materially increases HBM/leading-edge DRAM capacity will compress ASPs once ramps reach steady-state; the critical inflection is not 'build' but 'first high-yield, qualified production' — which is typically a 24–48 month path and where schedule slippage or yield misses create the largest option value. Geopolitical frictions and export controls act as asymmetric tail risks: they can truncate a multi-site buildout or force vendor re-sourcing, quickly repricing regional valuations.

Near-term market moves will be dominated by sentiment and event-driven flows, but the real value unlock is multi-year and binary — successful qualification expands addressable AI spend, while missed timelines trigger sharp multiple contraction. That creates distinct tradeable windows: capture equipment/manufacturing beta on the capex lead, hedge revenue cyclicality with short-duration protection on memory stocks, and avoid unconstrained long-duration exposure to a single memory name without explicit hedge funding the time-to-production risk.

Key Decisions for Investors

• Long MU via Jan-2028 LEAP calls (buy MU 01/19/2028 350C) size to 1–2% NAV: asymmetric upside if qualification and yield milestones meet market expectations; trim to half at 2x and use a 30% trailing stop to protect against rapid ASP-driven re-rating reversals.
• Buy a 9–15 month call spread on Lam Research (LRCX) or Applied Materials (AMAT) (e.g., 9–12 month 1x-2x call spread) to capture multi-quarter capex realization with defined premium outlay; target >2x payoff if capex cadence accelerates, max loss = premium.
• Pair trade: Long MU (40–60% notional) / Short a short-dated DRAM cyclicity hedge (buy MU puts 3–6 months out or short a memory ETF) sized to cap downside to ~10–15% NAV — protects against an inventory digestion or ASP collapse while keeping upside participation.
• Sell near-term volatility and buy longer-dated optionality: implement a calendar/diagonal on MU (sell 1–3 month calls around company events, buy 12–18 month calls) to collect rich front-month IV and retain long convex exposure to successful ramps.
• Take partial profits and reallocate to Taiwan/Taiwan-supplier exposure (small overweight to industrials/suppliers listed in the US) if MU rallies >30% from current levels — capture supply-chain carry while de-risking single-name concentration risk.