Semi-custom AI infrastructure with industry-proven AI scale-up performance.
Overview
NVIDIA NVLink™ Fusion delivers industry-leading AI scale-up and scale-out performance with NVIDIA technology plus semi-custom ASICs or CPUs. With a robust NVIDIA partner ecosystem, hyperscalers can build an ASIC hybrid AI infrastructure with NVIDIA NVLink and AI factory technologies, including rack-scale architecture, NVIDIA GPUs or NVIDIA Grace™ CPUs, co-packaged optics (CPO) switches, ConnectX® SuperNICs, BlueField® DPUs, and Mission Control.
Benefits
Unlocking the full potential of AI factories hinges on the need for swift, seamless communication among every accelerator. NVIDIA NVLink 1.8 TB/s interconnect can scale up to 72 accelerators to unleash AI performance.
The NVIDIA ecosystem partners, including ASIC designers, CPU, and IP providers, and OEM/ODMs enable hyperscalers to deploy custom silicon with NVIDIA NVLink Fusion with the fastest time to market.
As hyperscalers already deploy full NVIDIA rack solutions, NVLink Fusion allows heterogeneous silicon offerings while standardizing around a common rack design, accelerating AI factory deployment and simplifying management.
NVIDIA maintains an annual roadmap rhythm ensuring the highest performing NVLink and full rack architecture performance for NVLink Fusion adopters.
Platform
NVIDIA NVLink and NVLink Switch Chip enables 130 TB/s of bandwidth in one 72-accelerator NVLink domain (NVL72) and delivers 4X bandwidth efficiency with NVIDIA Scalable Hierarchical Aggregation and Reduction Protocol (SHARP)™ FP8 support.
NVIDIA NVLink-C2C extends the industry-leading NVLink technology to a chip-to-chip interconnect. This enables the creation of a new class of integrated products with NVIDIA partners, built via chiplets, allowing NVIDIA GPUs or CPUs to have a high-bandwidth coherent connection with custom silicon.
Learn more about the capabilities of NVIDIA NVLink Fusion.