Samsung has begun shipping 12-layer HBM4E samples to major customers. The devices target AI accelerators and data-centre systems with 48GB stacks, 14Gbps pin speed, and up to 3.6TB/s bandwidth.
Infineon has joined NVIDIA’s MGX ecosystem for AI rack power. The collaboration supports 800VDC architectures using silicon, silicon carbide, and gallium nitride technologies to improve power delivery in high-density AI data centres.
Rambus has introduced a DDR5 9600 client memory module chipset for CUDIMM, CQDIMM, and CSODIMM designs in future AI PCs, workstations, and notebooks.