Marvell Launches AI Infrastructure Memory Solution Portfolio, Targeting Agentic AI Inference Bottlenecks

Data infrastructure semiconductor giant Marvell Technology announced the launch of a new generation portfolio of memory solutions for AI infrastructure, covering server-level AI storage, rack-level CXL memory expansion and pooling, and multi-cabinet optical interconnect shared memory, aimed at addressing the growing memory capacity and bandwidth bottlenecks during Agentic AI inference. Marvell stated that as AI model scales expand, context windows lengthen, and KV Cache demands grow, traditional tightly coupled compute and memory architectures are limiting AI inference efficiency. Through memory disaggregation, memory resources can be scaled more independently from compute resources, improving GPU utilization and reducing data movement latency.