Collaboration Enables Scalable, High-Performance Architectures Across Edge, Core, Transport, Data Center, and AI Infrastructure
Arrcus, the hyperscale networking software company and a leader in core, edge, and multi-cloud routing and switching infrastructure, and UfiSpace, a leading provider of open disaggregated networking hardware, today announced a strategic partnership to deliver tightly integrated, best-of-breed networking solutions purpose-built for the demands of AI infrastructure. As enterprises and service providers race to support AI-driven applications and services, they face the challenge of building a network fabric that is highly scalable, cost-effective, and optimized for distributed AI workloads. From inferencing at the edge to training clusters in hyperscale data centers, the demands on network infrastructure continue to rise.
Read More on CIO Influence: AI-Augmented Risk Scoring in Shared Data Ecosystems
The collaboration between Arrcus and UfiSpace addresses these needs with a disaggregated, high-performance architecture powered by Arrcus’ ACE-AI platform and UfiSpace’s open hardware portfolio, leveraging the latest merchant silicon, including Broadcom’s Tomahawk 5, Trident 4, and Jericho 3-AI chipsets.
“AI infrastructure demands a fundamentally new approach to networking—one that’s open, composable, and built for extreme scale,” said Shekar Ayyar, CEO and Chairman at Arrcus. “Our partnership with UfiSpace brings together the best of software and hardware innovation to deliver powerful, agile, and cost-efficient solutions for distributed AI networking.”
The joint solution suite is designed for:
- AI inferencing at the edge, with low-latency performance
- High-throughput data pipelines in metro and transport networks
- Scalable AI training clusters in private and hyperscale data centers
- Multitenant AI fabrics with deterministic segmentation and isolation
- Advanced application and network monitoring capabilities
At the core of this collaboration is ACE-AI, Arrcus’ intelligent networking stack built specifically for AI workloads. With its modular architecture and deep silicon programmability, ACE-AI enables:
Also Read: AppDirect Appoints Carl Emond as the General Manager of ITCloud
- Two architectures to build highly scalable modern GPU clusters: IP CLOS fabric for horizontal scale and latency and Virtualized Distributed Routing (VDR) architecture for a fabric based load balancing and congestion control
- High-bandwidth, low-latency networking for large-scale model training
- Build a lossless Ethernet fabric with RoCEv2, PFC, ECN, and Load Balancing
- Dynamic routing and congestion control across GPU clusters
- Real time telemetry and monitoring of the hardware such as watchdog, buffers, counters and integration to orchestration frameworks
In addition, the solution includes ArcIQ, Arrcus’ real-time network visibility and analytics platform, providing end-to-end observability, intent-based telemetry, and closed-loop automation across the entire AI infrastructure—from edge to core.
“UfiSpace is committed to enabling the next generation of open, scalable infrastructure,” said Vincent Ho, CEO at UfiSpace. “By working closely with Arrcus, we’re delivering AI-optimized solutions that empower cloud providers, enterprises, and telecoms to deploy intelligent networking fabrics tailored to AI workloads.”
The Arrcus and UfiSpace partnership reflects a shared vision for disaggregated, cloud-native networking that is agile, secure, and optimized for high-performance solutions. The companies will continue to expand their joint solution portfolio to support evolving customer needs in areas such as AI networking fabrics, cloud edge services, and multicloud routing. The Arrcus and UfiSpace partnership delivers unprecedented flexibility, performance, and control—ushering in a new era of AI networking at global scale
[To share your insights with us, please write to psen@itechseries.com]
