EP 163: Breaking the Memory Wall: Micron’s Strategy for the AI Era

AI-driven demand for memory and storage has fundamentally altered the semiconductor industry, shifting it away from traditional cyclicality toward a new, higher baseline of growth. Memory now serves as a critical strategic asset for overcoming bottlenecks in both model training and inference. As AI workloads evolve, the industry faces a "memory wall," necessitating innovations like high-bandwidth memory (HBM) and expanded storage capacities to manage complex context windows and concurrent agentic AI tasks. Jeremy Werner, leading the core data center business at Micron, highlights the importance of deep co-design with hardware partners and the development of high-density storage solutions, such as 245-terabyte SSDs, which significantly improve gigabyte-per-watt efficiency. Despite massive investments in new fab capacity, the industry remains supply-constrained, as the rapid acceleration of AI capabilities continues to outpace current production infrastructure.