Marvell unveils Structera X memory expanders for AI inference

Marvell Technology announced the second-generation Marvell Structera X 3406 and Marvell Structera X 3506 CXL memory expanders, targeting cloud, enterprise and AI infrastructure providers building more efficient, scalable and higher-performance systems. The chips are designed for hyperscale datacenters and demanding data center workloads.

The company positioned the launch around rising memory pressure from AI and ML systems. Larger LLMs, expanding context windows and growing KV-cache requirements are increasing memory demands beyond what traditional server architectures can meet. Marvell said memory has become a critical constraint on AI performance and infrastructure efficiency.