XCENA launches MX1 memory lineup for hyperscale AI inference

XCENA introduced MX1, a production memory lineup designed to address capacity, utilization and data-movement limits affecting AI inference at scale. The company is showcasing the lineup at FMS 2026: The Future of Memory and Storage, being held Aug. 4-6 at the Santa Clara Convention Center.

The launch targets growing pressure from generative AI workloads, including larger models, longer context windows and expanding KV cache footprints. XCENA says high-bandwidth memory remains costly and capacity-limited, while scaling with additional servers can leave DRAM stranded and increase total cost of ownership. Built around CXL, MX1 is positioned as a way for operators to scale memory more efficiently alongside compute, with demonstrations using Intel Xeon 6 platforms for memory-intensive AI workloads.