The global AI computing arms race is pushing the memory chip market into a multi-year supply shortage. Management from Micron Technology stated at the JPMorgan Annual Technology Summit that supply constraints for HBM, DRAM, and NAND memory chips are expected to extend well beyond 2026. The core driver is the robust demand for high-performance memory from AI applications, while supply is constrained by technological bottlenecks that hinder rapid capacity expansion.
In a subsequent investment report, JPMorgan cited this view, stating that after hearing from Micron's management, the firm has greater confidence in the multi-year bull market logic for the AI memory segment. The supply-demand gap is difficult to reverse in the short term, presenting a clear signal for investors in the memory industry.
**Supply Expansion Faces Structural Constraints**
According to the JPMorgan report, Micron indicated that the causes of the tight memory supply are structural, not merely a cyclical capacity issue. Specifically, supply growth is limited primarily by two factors. First, performance improvements for new-generation memory chips are gradually narrowing, meaning the room to expand effective supply through technological iteration alone has significantly diminished. Second, the increasing die sizes of new-generation HBM chips are reducing the number of chips produced per wafer, further weakening supply elasticity. Additionally, while the introduction of extreme ultraviolet (EUV) lithography technology helps improve the manufacturing precision of advanced-node DRAM, it also imposes new constraints on the speed of production ramp-up and costs.
Given these combined factors, even if memory manufacturers actively expand capacity, it will remain difficult to quickly fill the supply gap created by AI demand in the short term.
**HBM4 Ramp Accelerates; 1-gamma Node to be Highest-Volume Node**
Regarding product progress, Micron disclosed several key details. Management stated that, driven by strong demand from AI applications, the company's 1-gamma process node is expected to be its highest DRAM wafer output node ever. HBM memory chips, which are created by vertically stacking and packaging multiple DRAM modules, are widely used in AI GPUs. Micron is continuously integrating EUV lithography into the mass production process for the 1-gamma node.
In terms of HBM product iteration, Micron revealed that the production ramp for HBM4 is twice as fast as that for HBM3. Mass production ramp-up for the next-generation HBM4E product is expected to begin in 2027, with initial samples utilizing DRAM modules based on the 1-gamma node.
**AI Inference Demand Expansion Boosts SSD Market Share**
Beyond HBM and DRAM, Micron also noted that the evolution of AI workloads is creating new growth opportunities for its solid-state drive (SSD) business. Management explained that the continuous expansion of AI context windows and the rapid growth of inference workloads are driving demand for high-capacity, high-performance storage, allowing the company to gain market share in the SSD segment.
Notably, Micron emphasized that its strategy is not to offer standardized off-the-shelf products but to collaborate deeply with customers to develop customized storage solutions for specific application scenarios. This model helps strengthen customer loyalty and enhances product premium potential.