HBM Supply Curve Gets Steeper, But Still Can’t Meet Demand

Micron raised its HBM total addressable market dramatically, now forecasting $100B in HBM revenue by 2028 (vs prior 2030), implying a 2025–2030 HBM pool of roughly $555B and materially larger downstream GPU/XPU and system revenue opportunities. In Q1 FY2026 Micron reported $13.64B revenue (+56.7%), operating income $6.14B (+2.8x), net income $5.24B (+2.8x), DRAM revenue $10.81B (+68.9%) and NAND $2.74B (+22.4%); datacenter revenue was $7.66B (+55.1%). Management raised FY2026 capex by $2B to $20B, accelerated fab and packaging timelines (Idaho, New York, Singapore) and now expects server unit growth in calendar 2025 in the high teens, signalling sustained strong pricing and supply-driven profitability for memory suppliers.

Analysis

Market structure: Micron (MU) is the clear near-term winner — higher ASPs and a re‑rated HBM TAM (from $35B in 2025 to $100B by 2028 implies >40% CAGR) give MU pricing power; Nvidia (NVDA) and hyperscalers gain indirectly via expanded accelerator spend while legacy low‑end memory vendors and hyperscaler margins face pressure. Supply/demand is tight: Micron reports DRAM cost/bit +20% QoQ and is accelerating $20B FY26 capex, implying material capacity rebalancing but with 12–36 month lead times for fabs and packaging.

Risk assessment: Tail risks include a demand shock (AI spending cutbacks) that could wipe out 30–50% of expected HBM upside within 6–12 months, geopolitical export controls that remove Chinese demand (high probability to materially change revenue mix), and operational execution risk from capex delays (Idaho/NY ramps slip into 2028–2030). Near term (days) expect sentiment-driven volatility; short term (weeks–months) earnings/booking updates will reprice; long term (3–5 years) outcomes hinge on Micron securing >20–25% HBM share and successful packaging scale.

Trade implications: Favor asymmetric long exposure to MU over 6–12 months while hedging execution/regulatory tail risk via options or relative shorts. Consider capitalizing on equipment/supply chain beneficiaries (SEMICON equipment names/ETF) for a 12–24 month upcycle. Use option structures (vertical call spreads on MU) to limit downside if IV spikes ahead of earnings.

Contrarian angles: Consensus may understate concentration risks in HBM assembly/packaging and overstate sustained pricing; Micron’s aggressive capex could compress FCF if prices normalize — history shows DRAM booms often reverse in 12–18 months. The market may be underpricing a scenario where hyperscalers vertically integrate or alternative memory architectures reduce HBM content per node, which would cap long‑term revenue growth.

Key Decisions for Investors

• Establish a 2–3% portfolio long position in MU (ticker MU) sized to target ~35% upside over 6–12 months; place a hard stop-loss at -18% to protect against an AI capex slowdown or inventory glut.
• Implement a pair trade for 3–6 months: long MU (2% notional) / short NVDA (1–1.5% notional) to express memory outperformance vs compute, exit if MU/NVDA relative outperformance reaches +20% or after 6 months.
• Execute an options hedge: buy a MU 9‑month call spread (buy ATM, sell 25% OTM) sized to risk 0.75–1% of portfolio to capture upside while limiting premium; roll or close on quarterly earnings beats/misses.
• Rotate ~3% from large‑cap software toward semiconductor equipment exposure (SOXX or LRCX/AMAT) for 12–24 months to play the upstream capex cycle; exit if Micron weekly server bookings fall below 10% YoY growth guidance for calendar 2025 or HBM ASP declines >15% QoQ.