NVIDIA’s Spectrum-6 reaches gigascale AI factories

NVIDIA Spectrum-6 is arriving across gigascale AI factories as networking becomes a central constraint for large training and inference systems. The 102.4-terabit-per-second Ethernet switch system delivers 2x the capacity of previous-generation systems and anchors the next generation of NVIDIA Spectrum-X Ethernet as part of the NVIDIA Vera Rubin platform.

CoreWeave, Microsoft, Nebius, SpaceXAI and Tesla will be among the first infrastructure builders to bring in Spectrum-6. CoreWeave, Microsoft and Nebius are also among the first providers expected to deploy Vera Rubin-based infrastructure with Spectrum-6, extending access to developers, startups and enterprises.

The platform is designed for AI workloads that require accelerators to exchange data continuously and stay synchronized. Spectrum-6 combines with NVIDIA ConnectX-9 SuperNIC and is engineered alongside the NVIDIA Vera CPU, Rubin GPU, NVLink 6 Switch, BlueField-4 DPU and Spectrum-6 Ethernet switch, with support for pluggable and co-packaged optics, liquid cooling, open network operating systems and multiple RDMA transport models.

NVIDIA says Spectrum-X Ethernet delivers up to 1.6x higher AI networking performance than off-the-shelf Ethernet and sustains up to 95% network efficiency across deployments exceeding 100,000 GPUs. Hardware-accelerated Spectrum-X multiplane topologies reduce the number of switches required for data centers by 1.7x, while Spectrum-X Ethernet Photonics adds 5x higher power efficiency and 10x improved mean time between incidents.