MU may be underestimated for the following reasons: Micron states that HBM capacity will be largely sold out/locked for 2025-2026, with HBM3E pricing mostly locked until 2026, and HBM4 is in progress. The HBM TAM is projected to grow from about ~$35-36B in 2025 to ~$100B by 2030; the share of HBM in DRAM continues to rise. SK hynix leads in HBM share, with Samsung quickly catching up (12-Hi HBM3E certified by NVIDIA), and Micron has significantly closed the generational gap and is entering top customers over the past two years. Policy and localization advantages (U.S. expansion + subsidies reduce geopolitical and compliance risks) 2025: HBM3E rapidly ramps up; Q4 single-quarter HBM revenue nearing $2B (annualized ~ $8B); FY25 Capex ≈ $13.8B, heavily reinvesting in DRAM/HBM front-end; HBM capacity will be largely sold out in 2025. 2026: Management guidance indicates FY26 Capex will be higher than FY25; HBM capacity will be fully booked within months in 2026; HBM3E pricing is mostly locked until 2026, with HBM4 in sampling/negotiation; deep collaboration with TSMC to advance HBM4/4E. Currently at a PE of 20 times, forward PE expected to be below 15 times. nfa dyor