AI factories now span tens of thousands of GPUs—and soon, millions—operating as a single distributed computing engine. To keep accelerators fully utilized, AI workloads demand deterministic latency, lossless throughput, stable iteration times, and the ability to scale not only within a data center, but across multiple sites.

NVIDIA networking delivers the full-stack fabric that makes this possible, combiningNVIDIA NVLink™ scale-up,NVIDIA Quantum InfiniBand andSpectrum-X™ Ethernet scale-out, Spectrum-XGS multi-data-center scale-across,NVIDIA® BlueField® DPU andDOCA™ for infrastructure services, and next-generationsilicon-photonics platforms to power the world’s most advanced AI factories.

Spectrum-6 Ethernet Switches Scale NVIDIA Rubin-Based AI Factories

Networking Purpose-Built for AI

The NVIDIA Spectrum-X Ethernet platform is purpose-built for AI, delivering 1.6x higher network performance than off-the-shelf (OTS) Ethernet. Spectrum-XGS Ethernet scale-across technology enables high-performance AI fabrics that span racks, buildings, and entire campuses.

With ultra-low latency, in-network computing, and next-generation silicon-photonics CPO systems, the NVIDIA Quantum InfiniBand architecture provides the industry’s fastest, most scalable networking for AI and HPC.

The NVIDIA BlueField platform combines high-speed networking with powerful computing to accelerate data center infrastructure, unlocking new levels of performance, efficiency, and security for the world’s AI data centers.

NVIDIA networking delivers a full-stack, accelerator-optimized fabric that minimizes tail latency, stabilizes iteration times, and keeps GPUs fully utilized across training and inference.

NVIDIA networking enables gigascale growth with high-radix switching, co-packaged optics, and Spectrum-XGS Ethernet technology that unify facilities into one high-performance AI fabric.

Data visualization and isolation are essential security functionalities for every data center, and NVIDIA networking solutions enhance security, simplify data center automation, and provide complete visibility into your network.

Scalable, modern network operations frameworks such as NetQ, UFM, and DOCA use network telemetry to deliver AI-powered cyber intelligence and analytics to ensure your AI network fabric operates smoothly.

Build digital twins of data center networks with NVIDIA Air—a cloud-hosted virtual environment for design, testing, and validation of network provisioning, automation, and security policies.

NVIDIA accelerated networking enables efficient infrastructure with energy-efficient design, optimized connectivity through co-packaged optics, and intelligent in-network processing.

NVIDIA Spectrum-X Ethernet Photonics switch systems deliver 5x improved power efficiency and uptime. New NVIDIA Inference Context Memory Storage Platform with NVIDIA BlueField-4 storage processor to fuel inference for AI reasoning.

NVIDIA BlueField-4 powers the NVIDIA Inference Context Memory Storage Platform, a new class of AI-native storage infrastructure designed to accelerate and scale gigascale agentic inference.

Hyperscalers adopt NVIDIA open networking platform to drive gigascale AI data center performance.

Spectrum-XGS Ethernet unifies distributed data centers into AI super-factories with 1.9x higher NVIDIA Collective Communications Library (NCCL) performance.