HBM4E delivers a stable pin speed of 14 Gbps, scalable to 16 Gbps, and provides up to 3.6 TB/s per stack, aiming to boost performance for large language models and next-generation AI systems.

At the MSST conference, Peak:AIO and Los Alamos National Laboratory launched Lattice, an open-source pNFS metadata server designed to eliminate metadata bottlenecks in large-scale AI and HPC storage systems.

Kioxia will showcase its contributions to HPE Spaceborne Computing missions, including projects aboard the International Space Station and future lunar exploration efforts, underscoring the role of SSDs in space computing.

The high demand for HBM memory continues due to limited production capacity and long-term supply contracts with hyperscale data centers.

The architecture aims to address underutilization of GPU compute by addressing the metadata bottleneck that constrains AI workloads, with AI infrastructure requiring ultrafast, parallel access to massive datasets.

Samsung has begun shipping 12-layer HBM4E samples to major global customers, following earlier mass production of HBM4, with a 48-GB 12-layer configuration and plans to expand to 32-GB (8-layer) and 64-GB (16-layer) options to meet rising demand from AI workloads.

Kioxia announces involvement in HPE platform deployments, highlighting PCIe 5.0 Enterprise and Datacenter standards (EDSFF and 24G SAS), with products aimed at higher performance, density, and power efficiency across enterprise servers, digital storage, mobile computing, and modern data centers.

Lattice is a Linux-based, user-space metadata server architecture that splits the metadata control plane into four layers (Protocol State Plane, Lattice Core, MD Catalog Authority, Data Server Control Plane), enabling elastic metadata services that can scale from a single server to over 1,000 metadata servers.