NVIDIA has identified CoreWeave, Microsoft, Nebius, SpaceXAI, and Tesla as the first organizations preparing to deploy its Spectrum-6 Ethernet platform, which is designed for the upcoming Vera Rubin generation of AI infrastructure. The company said the new networking platform will begin arriving in gigascale AI factories as customers prepare systems capable of connecting hundreds of thousands of GPUs.
Spectrum-6 combines a 102.4 Tbps Ethernet switch with ConnectX-9 SuperNICs as part of NVIDIA’s Spectrum-X Ethernet platform. NVIDIA said the system supports both pluggable and co-packaged optics, liquid cooling, open network operating systems, and AI-optimized traffic management designed to improve the performance of large-scale collective communications across massive GPU clusters.
NVIDIA said Spectrum-X can deliver up to 1.6x higher AI networking performance than conventional Ethernet while sustaining up to 95% network efficiency across deployments exceeding 100,000 GPUs. The company also said its multiplane topology reduces the number of required switches while Spectrum-X Photonics improves both network power efficiency and reliability.
Analysis 🌐
The list—CoreWeave, Microsoft, Nebius, SpaceXAI, and Tesla—reads like a who’s who of next-generation AI infrastructure builders and provides an early indication of where NVIDIA expects the largest Vera Rubin AI factories to emerge.
The announcement also underscores that networking has become a competitive differentiator for AI infrastructure. As GPU clusters scale beyond 100,000 accelerators, the efficiency of Ethernet fabrics increasingly determines overall system utilization, making advanced switching technology as critical as the compute platform itself.
In March, NVIDIA used GTC 2026 to introduce its Vera Rubin platform, positioning networking—not just compute—as the core enabler of next-generation AI factories. The system integrates seven new chips into a tightly coupled architecture spanning GPU, CPU, NIC, DPU, switching, and storage, with the NVIDIA Spectrum-6 Ethernet and ConnectX-9 SuperNIC forming the backbone of a unified, high-performance data plane. The design reflects a shift toward fully disaggregated yet tightly synchronized infrastructure, where east-west traffic, memory access, and inference pipelines depend on deterministic, low-latency networking at scale.
At the system level, NVIDIA organizes Vera Rubin into five rack-scale building blocks, all interconnected through Spectrum-X Ethernet or Quantum-X800 InfiniBand fabrics. The Spectrum-6 SPX Ethernet rack anchors this architecture, delivering high-throughput, low-latency connectivity optimized for AI traffic patterns such as all-to-all communication, gradient exchange, and KV cache distribution. NVIDIA also introduces co-packaged optics within the Spectrum-X photonics platform, targeting up to 5x improved optical power efficiency and 10x greater resiliency compared to traditional pluggable optics. These advances directly address one of the central bottlenecks in large-scale AI clusters: sustaining high GPU utilization across thousands of nodes without network-induced stalls.
The broader Vera Rubin stack extends networking principles into every layer of the AI factory. ConnectX-9 SuperNICs provide high-speed, programmable interfaces for GPU and CPU nodes, while BlueField-4 DPUs offload infrastructure services such as storage virtualization and security. The new BlueField-4 STX storage rack introduces a distributed, network-attached memory tier optimized for key-value cache handling, using NVIDIA DOCA Memos to accelerate inference workflows by up to 5x. Together, these elements create a fabric-centric architecture where compute, storage, and memory behave as a single, network-coherent system, enabling large-scale agentic AI workloads to operate with higher efficiency and lower cost per token.
