Upscale AI offers Scale-Out Ethernet Systems
Product
Scale-out Ethernet systems are open fabrics built on NVIDIA Spectrum-X switch silicon and a SONiC-based NOS for AI clusters.
- Currently Offered
Scale-out Ethernet systems are open fabrics built on NVIDIA Spectrum-X switch silicon and a SONiC-based NOS for AI clusters.
Upscale AI's scale-out family builds open Ethernet fabrics using NVIDIA Spectrum-X switch silicon paired with an AI-optimized SONiC network operating system. The systems serve as scalable, low-latency fabrics for modern AI workloads across compute, accelerators, memory, and storage, and are designed for heterogeneous, multi-vendor AI data center infrastructure under a consistent operating model.
The offering bundles hardware, software, and lifecycle services and is planned for availability later this year. It is positioned as an open alternative to proprietary scale-out interconnects such as Nvidia InfiniBand.
NVIDIA InfiniBand is a high-throughput, low-latency interconnect fabric for scale-out AI and HPC clusters.
Spectrum-X combines Ethernet switches and SuperNICs for GPU communication in distributed AI training and inference.
NVIDIA InfiniBand is a high-throughput, low-latency interconnect fabric for scale-out AI and HPC clusters.
Spectrum-X combines Ethernet switches and SuperNICs for GPU communication in distributed AI training and inference.
Arista AI Networking is a portfolio for high-performance AI cluster connectivity.
CN5000 is Cornelis Networks' shipping Omni-Path fabric for AI and high-performance computing.
High-bandwidth, low-latency cloud networking for distributed AI and GPU clusters.
AuraLinks is an open-standards network fabric for scale-up and scale-out interconnect in AI data centers.
Arista AI Networking is a portfolio for high-performance AI cluster connectivity.
JetStream is d-Matrix’s I/O accelerator connecting Corsair systems for scalable data center AI inference.
CN5000 is Cornelis Networks' shipping Omni-Path fabric for AI and high-performance computing.
High-bandwidth, low-latency cloud networking for distributed AI and GPU clusters.
Cisco Silicon One is a unified networking silicon architecture.
AuraLinks is an open-standards network fabric for scale-up and scale-out interconnect in AI data centers.