Memory/Storage: Challenges for Memory as AI Growth Accelerates

10 Sep 2025
Hardware & Systems
As the adoption of generative and agentic AI accelerates, the challenges for memory as a key enabler of AI/ML processing architectures continue to grow. Balancing the demands for ever greater bandwidth and capacity with the needs of power efficiency, thermal management and increased reliability is increasingly difficult. Continued advances in high performance HBM and GDDR memories, and mainstream DDR and LPDDR memories, remains a strategic industry imperative. In addition, a suite of new technologies including multiplexed modules (MRDIMM), CXL and processing in memory are needed to meet upcoming AI requirements. In this panel, we’ll discuss the evolution of memory technologies and the challenges the industry faces on the road ahead for future AI chips and systems.