Hardware
NVIDIA NVLink Fusion Brings NVHBM to Next-Generation AI Infrastructure
NVIDIA’s NVLink Fusion integrates its custom NVHBM memory into the MGX rack‑scale architecture, enabling hyperscalers to deploy large numbers of custom XPUs with higher bandwidth, reduced silicon area
NVIDIA’s NVLink Fusion integrates its custom NVHBM memory into the MGX rack‑scale architecture, enabling hyperscalers to deploy large numbers of custom XPUs with higher bandwidth, reduced silicon area, and lower power consumption than standard HBM4e. The new memory delivers up to 30 % more throughput per stack, cuts physical support area by up to 67 %, and uses 15 % less HBM power, resulting in an overall XPU performance boost of roughly 30 % at rack scale. This technology supports next‑generation AI workloads that demand larger models, higher compute density, and efficient data‑center deployment.
Related
- NVIDIA NVLink: The Scale-Up Network for AI Factories
- How the NVIDIA Vera Rubin Platform is Solving Agentic AI’s Scale-Up Problem
Source: NVIDIA Developer | 2026-08-26