Upscale AI Supercharges Open, Heterogeneous Scale-Out AI Clusters with NVIDIA Ethernet Switch Silicon

Upscale AI to Deliver an Open, Interoperable AI Fabric with NVIDIA Spectrum-X Ethernet and AI-Optimized Systems & Software

Upscale AI announced its plan to deliver open, scale-out Ethernet systems. These systems will be built upon industry-leading NVIDIA Spectrum-X Ethernet switch silicon and Upscale’s AI-optimized SONiC software. This enables AI infrastructure operators to deploy scalable, low-latency fabrics optimized for modern AI workloads, while supporting heterogeneous architectures across compute, accelerators, memory, and storage.

As part of this initiative, Upscale AI has joined the NVIDIA Partner Network, working closely with NVIDIA and its ecosystem on reference architectures and validated designs to accelerate the deployment of large-scale AI data center networks. Upscale AI’s planned NVIDIA-powered solutions strengthen its position as a pure-play provider of AI-native networking infrastructure.

Streamlining AI Infrastructure at Scale

To reduce complexity at scale, Upscale AI is fueling the adoption of open, interoperable Ethernet networking for heterogeneous AI clusters. This new AI fabric blueprint for NVIDIA Spectrum-X Ethernet-based systems, centered on Ethernet interoperability and designed for diverse compute environments, enables customers to deploy multi-vendor, production-grade AI infrastructure while maintaining a consistent operating model.

Upscale AI’s AI optimized, scale-out systems built on NVIDIA Spectrum-X Ethernet switch silicon are powered by an enterprise-grade, focused SONiC network operating system with end-to-end support. By integrating ASIC-native telemetry with deterministic, lossless Ethernet behavior and industry-standard networking workflows, these systems deliver predictable performance, operational simplicity, and reliability at scale.

This full-stack approach enables high-speed data movement, workload isolation, and scalable orchestration across heterogeneous environments, while preserving the flexibility of open-source management.

Upscale AI plans to bring its Spectrum-X Ethernet based scale-out systems to market later this year, targeting AI data centers building diverse, multi-vendor infrastructure. Delivered as fully supported, end-to-end solutions, these offerings combine hardware, software, and lifecycle services to accelerate deployment, simplify operations, and enable long-term AI infrastructure evolution.

Also Read: AiThority Interview With Arun Subramaniyan, Founder & CEO, Articul8 AI

“NVIDIA Spectrum-X Ethernet switch silicon is setting a new standard for Ethernet-based AI performance,” said Barun Kar, CEO of Upscale AI. “Pairing this technology with our purpose-built systems and AI-optimized SONiC software allows us to deliver the best of both worlds: an open, highly scalable and interoperable architecture with operational simplicity.”

“To lead in the trillion-parameter model era, scalability and efficiency are paramount,” said Gilad Shainer, SVP at NVIDIA. “We look forward to collaborating closely with Upscale AI as the team leverages the NVIDIA Spectrum-X Ethernet platform to help companies build the world’s most advanced open AI infrastructure.”

“As AI infrastructure evolves toward increasingly heterogeneous architectures, scalable and operationally sound Ethernet fabrics are becoming essential,” said Alan Weckel, Co-Founder and Technology Analyst at 650 Group. “By combining NVIDIA Spectrum-X Ethernet switch silicon with a full-stack approach that integrates systems, software, and support, Upscale AI is addressing a key gap in making open, scale-out AI networking practical for large scale AI deployments.”

“As scale-out Ethernet becomes the foundation for AI infrastructure, organizations are looking for solutions that combine openness with operational maturity,” said Sameh Boujelbene, Vice President at Dell’Oro Group. “By delivering a fully integrated stack built on NVIDIA Spectrum-X Ethernet switch silicon-based systems and SONiC, Upscale AI is addressing key requirements around scalability, supportability, and long-term operability for heterogeneous AI data center networks.”

Also Read: Cheap and Fast: The Strategy of LLM Cascading (Frugal GPT)