半导体 - 传统存储器定价能力再度提升-Greater China Semiconductors-Old Memory Upping Pricing Power Again
Micron TechnologyMicron Technology(US:MU)2026-02-05 02:22

Summary of Conference Call on Greater China Semiconductors Industry Overview - The focus is on the semiconductor memory market, particularly old memory types such as NOR, MLC NAND, and DDR4 - The analysis indicates a significant shift in supply-demand dynamics, with expectations of reduced support from mainstream memory vendors for older memory types [1][3] Key Points Memory Supply-Demand Dynamics - The memory supercycle is expected to further squeeze support for old memory types, with expectations that commodity memory supply will be fully sold by 2026 [3] - Forecasts indicate a 26% undersupply for DDR4 in 2026, up from a previous estimate of 20%, driven by reduced support from Korean vendors [4] - MLC NAND is projected to face a 30%+ undersupply in 2026, with Samsung, Kioxia, and Micron potentially reducing supply [5] - NOR memory is expected to see an 8% undersupply in 2026, an increase from 5%, due to declining supply from US vendors [6] Pricing Power and Market Sentiment - There is a bullish sentiment on old memory, with analysts recommending buying on dips [1][3] - Pricing for DDR4 and MLC NAND is expected to rise, with significant price hikes noted for smartphone products (60-70% QoQ) and server DDR5 pricing increasing by 100% QoQ [59] - The pricing for consumer NAND is anticipated to increase by 30-40% QoQ, while eSSD pricing may rise by 50-70% QoQ [59] Company-Specific Insights - Price targets for several companies have been raised, including: - Nanya Technology Corp: NT$298.00 to NT$348.00 - Winbond Electronics Corp: NT$130.00 to NT$155.00 - Powerchip Semiconductor: NT$77.00 to NT$88.00 - Macronix International: NT$93.00 to NT$121.00 - GigaDevice Semiconductor: Rmb355.00 to Rmb414.00 [7] AI and Memory Demand - The shift towards AI, particularly Agentic AI, is driving massive demand for DRAM and NAND memory, necessitating higher memory capacity and performance [34][35] - The memory requirements for AI inference are increasing, with projections indicating that AI inference demand will account for significant portions of global memory supply in 2026 [49] Market Risks and Considerations - The potential for supply shortages poses risks, particularly for automotive applications [5] - The overall DRAM industry demand is expected to grow by 20% YoY in 2026, driven by strong order demand from cloud service providers [59] Additional Insights - The analysis highlights the importance of memory in AI development, noting that memory is becoming a critical bottleneck as AI systems evolve [29][33] - The report emphasizes the need for architectural and software-level memory efficiency to address the increasing memory demands of AI applications [29][33] This summary encapsulates the key insights and projections regarding the semiconductor memory market, highlighting the implications for pricing, supply-demand dynamics, and the impact of AI on memory requirements.

半导体 - 传统存储器定价能力再度提升-Greater China Semiconductors-Old Memory Upping Pricing Power Again - Reportify