Micron Boosts Memory Spend Amid AI Crunch

🔑 Enhanced Key Takeaways

• •Micron's Q2 FY2026 earnings on March 18, 2026, reported revenue beating estimates with non-GAAP gross margin at 68%, driven by full commitment of HBM production capacity for the year[3].
• •Micron began high-volume shipments of HBM4 36GB 12-high memory for Nvidia's Vera Rubin platform, offering over 11 Gb/s speeds and 2.8 TB/s bandwidth, 2.3x faster than HBM3E[3].
• •FY2026 capex raised to over $25 billion for Tongluo fab and U.S. expansions, with $200 billion long-term investment announced in February 2026 for HBM capacity starting 2027[1][5].
• •Micron qualified 1-gamma DRAM node using ASML EUV lithography in 2025, enabling higher yields and power efficiency over competitors[3].
• •New factories opened in India and expanded in New York, with HBM4e 2026 capacity fully contracted and new clients on waiting lists[2].

🛠️ Technical Deep Dive

• •HBM4 36GB 12-high (12H) stacks: speeds exceeding 11 Gb/s, bandwidth over 2.8 TB/s, 2.3x performance increase and 20% more power-efficient than HBM3E[3].
• •HBM4 48GB 16-high (16H) stacks in sampling: utilize ultra-thin die technology for higher capacity in AI accelerators[3].
• •1-gamma (1γ) DRAM node: world's most advanced process using Extreme Ultraviolet (EUV) lithography from ASML, achieving higher yields and better power efficiency[3].
• •HBM4e: core product with all 2026 production capacity contracted[2].

🔮 Future ImplicationsAI analysis grounded in cited sources

⏳ Timeline