NVIDIA Spectrum-X Ethernet Networking Platform

The NVIDIA Spectrum-X Ethernet platform is the world’s first end-to-end, AI-optimized Ethernet platform for multi-tenant generative AI factories. It combines purpose-built Ethernet switches and SuperNICs to deliver high effective bandwidth, low latency, and strong performance isolation for hyperscale AI infrastructure—all built on standards-based Ethernet.

NVIDIA Spectrum-X Ethernet is designed for cloud service providers and large enterprises running multi-tenant AI workloads at hyperscale. It supports environments ranging from large enterprise deployments to massive cloud AI factories with hundreds of thousands of GPU connections.

Spectrum-X Ethernet uses advanced telemetry-based congestion control to provide noise isolation between tenants. Built-in performance isolation ensures each tenant’s AI workloads run predictably and consistently, improving service quality across simultaneous AI jobs.

Adaptive routing is a feature where the Spectrum-X Ethernet switch and SuperNIC work in tight coordination to dynamically route traffic, enabling the highest effective bandwidth and network resiliency for AI fabrics. It is specifically targeted at lossless networking to prevent performance degradation at scale.

Spectrum-X Ethernet scales to hundreds of thousands of GPU connections in efficient two-tier leaf/spine topologies. With Spectrum-X Multiplane technologies, networks can scale up to 128,000 GPUs in two tiers—64x more than single-plane networks—by splitting each GPU’s SuperNIC traffic across two or more independent network planes.

Spectrum-XGS Ethernet is an extension of the Spectrum-X platform designed for “gigascale” AI, enabling multiple distributed data centers—potentially separated by hundreds of kilometers—to function as a single AI super-factory. It introduces a modified congestion control algorithm that accounts for inter-facility distance, and it accelerates multi-site NCCL collectives by 1.9x compared to standard approaches.

The newest generation (for NVIDIA Vera Rubin AI factories) is based on the Spectrum-6 ASIC, offering 102.4 Tb/s per switch chip with 200 G SerDes—double the bandwidth per lane versus the prior generation. The SN6600 offers 128 ports of 800 G OSFP connectivity, while the SN6800 packs 512 ports of 800 G in a 5U form factor.

The Spectrum-X Ethernet Photonics switches use silicon photonics (co-packaged optics) technology to deliver 10x greater mean-time-between-interruptions (MTBI), 5x longer uptime, and 5x better power efficiency compared to traditional pluggable transceiver-based switches. They use the ultra-dense MMC-16 optical fiber form factor for efficient switch footprint usage.

The Spectrum-X Ethernet platform supports three SuperNICs across generations: the BlueField™-3 (for NVIDIA Hopper-era systems, optimized for RDMA/RoCE and multi-tenant isolation), the ConnectX™-8 (800 Gb/s total throughput via 2×400 G, PCIe Gen6, for NVIDIA Blackwell systems), and the ConnectX-9 (1,600 Gb/s per GPU via 4×200 G SerDes, PCIe Gen6, for NVIDIA Vera Rubin NVL72 systems). All SuperNICs offload collective communications and provide telemetry for adaptive routing.

NVIDIA NetQ™ provides flow-level visibility, tracing performance from the GPU to the SuperNIC, and mapping per-hop behavior across switch ports and RoCE queues. For deployment, NVIDIA DSX Air simulates a full AI factory ecosystem, allowing customers to validate hardware and partner offerings to accelerate time-to-first-token and ensure day-one performance at scale.