Samsung is developing an 8-layer HBM4E for NVIDIA's NVHBM.

Huo Xing Finance reports that, on August 28, according to the Seoul Economic Daily, Samsung Electronics is developing an 8-layer HBM4E product at NVIDIA’s request for use in NVIDIA’s custom high-bandwidth memory, NVHBM. This design reduces the number of stacked layers compared to Samsung’s original 12-layer and 16-layer proposals. NVIDIA has set a target speed of 17 to 18 Gbps, approximately 20% higher than the 14.4 Gbps of Samsung’s early HBM4E samples. The 8-layer architecture lowers the complexity of wafer thinning, precise stacking, and backend packaging, easing yield pressures and enhancing supply capacity. NVHBM is expected to be first deployed in the Rubin Ultra AI GPU, slated for release next year. The GPU interconnect scale of Rubin Ultra will expand from the current 72 to up to 576 GPUs, compensating for reduced per-GPU memory capacity through increased high-speed GPU collaboration. Samsung possesses end-to-end capabilities in DRAM, logic chip, and base die design and manufacturing, enabling it to offer a one-stop solution for custom HBM. Industry insiders believe that Samsung has already demonstrated high transmission speeds with HBM4; as NVIDIA shifts its competitive focus from high stacking to higher transmission speeds, Samsung’s related capabilities may attract increased attention.