Cornelis Launches CN5000: Industry Leading AI and HPC Scale-out Network

June 3rd, 2025

By Cornelis Networks

Outperforms alternative networks with 30% higher HPC application performance, 2X greater message rates, and 6X faster AI collective communication

Wayne, PA June 03, 2025 – Cornelis Networks, a leading independent provider of intelligent, high-performance networking solutions, today launched the CN5000 family, the industry's highest-performing end-to-end networking solution purpose-built to maximize AI and HPC performance, supporting deployments with up to 500,000 endpoints. Engineered to address compute underutilization and scaling challenges, CN5000 delivers the industry's most advanced lossless data transfer and congestion avoidance capabilities—enhancing infrastructure performance and efficiency across AI and HPC environments.

"Networking should do more than just move data quickly – it should unlock the full potential of every compute cycle," said Lisa Spelman, CEO of Cornelis. "That’s the performance we are offering customers with the CN5000 – a new breed of network-led application acceleration for AI and HPC applications where our scale-out network becomes a force-multiplier for performance at any scale."

The World's Highest-Performance Scale-Out Network for AI & HPC

CN5000 enables AI and HPC applications to achieve faster and more predictable job completion times, and greater computational efficiency by minimizing congestion events and maintaining highest throughput under load. For HPC workloads, CN5000 outperforms InfiniBand NDR with 2X higher message rates¹, 35% lower latency¹ and up to 30% faster simulation times² driven by its advanced, congestion-free architecture. For AI applications, CN5000 delivers 6X faster collective communication³ compared to RoCE, driven by Omni-Path's lossless architecture.

AI at Scale: CN5000 delivers near-linear scaling of training performance for large language models and unlocks advanced inference with extended reasoning enabling higher-quality outputs and tackling previously intractable problems.
Faster, More Predictable HPC: CN5000 delivers up to 30% faster execution for compute intensive HPC workloads like computational fluid dynamics (CFD), climate modeling, and seismic simulations with higher sustained throughput under real-world conditions compared to InfiniBand NDR.
Universal Interoperability: CN5000 seamlessly interoperates with GPUs, CPUs, and accelerators, from AMD, Intel, NVIDIA, and more. Fully supports OpenFabrics software stacks for truly vendor-neutral deployments.

CN5000 Product Family Overview

The CN5000 family is a comprehensive 400G scale-out networking solution that includes:

SuperNICs: Single and dual-port, air- and liquid-cooled
Switches: 48-port switch and modular 576-port Director Class switch configurations, air- and liquid-cooled
OPX Software Suite: Open-source, industry-standard host and network management software
Cabling: Passive and active copper and active optical options for efficient, high-density connectivity

This fully integrated, end-to-end solution delivers maximum performance, scalable efficiency, and seamless deployment across diverse workloads. CN5000 Product Family Link | Cornelis Networks

Omni-Path Architecture Overview

Powered by the Omni-Path architecture, CN5000 delivers lossless and congestion-free data transmission with credit-based flow control and dynamic fine-grained adaptive routing. Designed for maximum performance, reliability, scalability, and data integrity with dynamic lane scaling and optimized link protection. The Omni-Path architecture is driving advanced features into the Ultra Ethernet Consortium today to bridge the performance gap between traditional networks and the demands of scale-out computing environments.

Strategic Innovation Roadmap

CN5000 is the first step in Cornelis' long-term vision to revolutionize performance in scale out computing. The upcoming CN6000 (800G) will unify Omni-Path and RoCE-enabled Ethernet for double the bandwidth and seamless cross-paradigm integration. CN7000 (1.6T) is a revolutionary platform integrating Ultra Ethernet Consortium standards with Cornelis' proven Omni-Path architecture, establishing a new performance frontier for the most demanding AI and HPC environments.

Availability

CN5000, Cornelis' breakthrough AI and HPC scale-out networking solution, begins shipping to customers this month and will be broadly available starting Q3 2025 from all major OEMs.

About Cornelis

Cornelis Networks delivers high-performance, scale-out networking solutions that accelerate AI and HPC workloads. Built on the powerful Omni-Path architecture, Cornelis technology enables lossless, congestion-free networking that reduces training time, improves inference, and maximizes compute utilization. From foundation model training to complex climate modeling and real-time analytics, Cornelis' solutions power the most demanding workloads across commercial, academic, government and cloud environments. With a focus on performance, scalability, and efficiency, Cornelis helps organizations achieve faster insights and greater return on infrastructure investments. Learn more at cornelisnetworks.com.

Contact

PR contact for Cornelis Networks

Matt Stubbs

mstubbs@voxuspr.com

Voxus PR

Performance Configuration Details

2x Greater Message Rate and 35% Lower Latency details: Intel MPI 2021.15, Intel(R) MPI Benchmarks 2021.9. Tests performed on 2 socket AMD EPYC 9334 32-Core Processor. Turbo enabled with acpi-cpufreq driver. Rocky Linux 9.5 (Blue Onyx). 5.14.0-503.14.1.el9_5.x86_64 kernel. 24x16GB, 384 GB total, Memory Speed: 4800 MT/s. Cornelis OPX 12.0.0.0.17. NVIDIA NDR InfiniBand: Mellanox Technologies MT2910 Family [ConnectX-7]. MQM9700-NS2F Quantum 2 switch. 2M passive copper cables. UCX as packaged in hpcx-v2.23.
Ansys Fluent. Tests performed on 2 socket AMD EPYC 9755 Eng Sample. Turbo enabled with acpi-cpufreq driver. Rocky Linux 9.5 (Blue Onyx). 5.14.0-503.14.1.el9_5.x86_64 kernel. 24x32GB, 768 GB total, Memory Speed: 5600 MT/s. Cornelis OPX 12.0.0.0.17., Open MPI 5.0.7. NVIDIA NDR InfiniBand: Mellanox Technologies MT2910 Family [ConnectX-7]. MQM9700-NS2F Quantum 2 switch. 2M passive copper cables. UCX and Open MPI 4.1.7rc1 as packaged in hpcx-v2.23.
Cornelis simulations of all-to-all traffic on industry-standard simulation framework (SST) modeling CN5000 with credit-based flow control and standard 400G Ethernet with priority flow control.

Broad OEM Partner and Ecosystem Adoption

"Altair is excited to drive the next wave of breakthroughs with Cornelis. The CN5000's advanced network architecture delivers measurable performance improvements for our simulation workloads, reducing time-to-solution for our customers. The significant reduction in communication bottlenecks is particularly valuable for our large-scale Radioss simulations, where network performance directly impacts business outcomes. We're pleased to support this powerful advancement in AI and high-performance computing." - Eric Lequiniuo, SVP Radioss Development and Altair Solver

"We're excited to see how our deep collaboration with Cornelis Networks —across multiple generations built using Alphawave's High-Speed Connectivity IP portfolio and Custom Silicon Platform—has delivered breakthrough performance, efficiency, and scale in high-performance networking," said Mohit Gupta, Executive Vice President and General Manager, Custom Silicon & IP, Alphawave Semi. "CN5000 exemplifies how strong semiconductor partnerships can unlock transformative capabilities for tackling the world's most demanding AI and HPC challenges."

"At AMD, we are committed to customer choice and a robust ecosystem of standards-based enterprise solutions that deliver the performance and scalability needed for today's AI and HPC workloads. We're thrilled to collaborate with Cornelis Networks as a member organization of the Ultra Ethernet Consortium driving next-generation computing and networking innovations." - Robert Hormuth, Corporate Vice President, Data Center Solutions Group, AMD

"The CN5000 networking solution delivers targeted performance to accelerate Ansys simulations at scale," said Wim Slagter, Sr. Director Partner Programs at Ansys. “By reducing communication bottlenecks, Cornelis Networks' next-generation interconnect helps our customers achieve faster time-to-solution with greater cost efficiency."

"Aspen Systems is excited to strengthen our partnership with Cornelis Networks as we implement their networking solutions across our education, government, and commercial customers. The new CN5000 line represents a significant advancement in the Omni-Path Architecture that our customers have been eagerly anticipating. We expect this technology to enhance the performance of our HPC clusters and enable more powerful solutions for demanding workloads. We look forward to bringing these innovations to market through our collaboration with Cornelis Networks." - Mako Furukawa, Senior Sales Engineer, Aspen Systems

"Atipa Technologies sees rapidly growing demand for integrated HPC and AI clusters that deliver exceptional performance at scale. The CN5000 provides the advanced networking capabilities our customers need for their most demanding workloads. We're enthusiastic about incorporating these innovations into our solutions, as they enable us to design more powerful systems without the traditional bottlenecks of conventional networks. The CN5000 strengthens our ability to differentiate our offerings in this competitive marketplace." - Bart Willems, Chief Technology Officer, Atipa Technologies

"At Intel, we're excited to see Cornelis launch CN5000—a high-performance, end-to-end networking solution that broadens customer choice. With its unique Lossless and Zero Congestion architecture, CN5000 helps maximize the compute potential of Intel Xeon 6 processors and AI accelerators, enabling customers to achieve faster time-to-solution across the most demanding of HPC and AI applications. We're equally enthusiastic about Cornelis' contributions to the Ultra Ethernet Consortium, helping shape the future of high-performance networking." – Ronak Singhal, Senior Fellow, Intel

"We're excited to see Cornelis Networks raising the bar in high-performance networking with CN5000. As system integrators focused on delivering optimized HPC and AI solutions, we recognize the critical importance of advanced interconnect technologies for application performance. The CN5000 shows great promise in addressing key networking challenges our customers face. We look forward to exploring how this innovation can enhance the solutions we deliver to the research, scientific, and enterprise computing communities." - Dominic Daninger, VP of Engineering, Nor-Tech

"Penguin Solutions has been working with Cornelis for many years, and we're excited to see the CN5000 come to market delivering updated, enhanced technology from the Omni-Path roots. The inclusion of a liquid cooled and OCP Open Rack (ORV3) compatible switch design in the product roadmap will be particularly valuable to our HPC customers using any of Penguin's 21-inch OCP-optimized servers. Our customers are adding AI to traditional HPC workloads so the CN5000's core latency and message rate improvements will be appreciated too, whether doing Beowulf clustering with our ICE ClusterWare™ software platform or solving for the demanding communication patterns of modern AI training." - Phil Pokorny, Chief Technology Officer, Penguin Solutions

"PSSC Labs is excited to integrate the CN5000 fabric solution into our high-performance computing systems. As demand for efficient AI and HPC clusters continues to accelerate, our customers increasingly require networking solutions that eliminate performance bottlenecks. The CN5000's advanced capabilities align perfectly with our commitment to delivering optimized, purpose-built computing systems through our PowerWulf clusters and NZO Cloud offerings. We anticipate this technology will enable our customers to achieve significantly better application performance and scaling efficiency." - Alex Lesser, VP Sales, PSSC Labs - NZO Cloud

"Fast GPUs alone don't move the needle - application performance depends on a rock-solid foundation where high-throughput storage and high-speed networking work in sync. With Cornelis' CN5000 fabric and Quobyte's parallel file system, organizations get a tightly integrated data path that scales effortlessly, handles mixed I/O patterns and eliminates the latency and congestion challenges typical of legacy stacks. It's the performance backbone for next-gen AI and HPC environments." - Bjorn Kolbeck, Co-Founder and CEO, Quobyte

"We are pleased with our recent collaboration with Cornelis in developing and delivering the CN5000, which supports the advancement of scale-out networking for AI and HPC through TSMC's advanced manufacturing process," said Lucas Tsai, Vice President of Business Management at TSMC North America. “This partnership represents the convergence of robust manufacturing excellence and high-performance network design, enabling the scale, efficiency, and reliability essential for next-generation AI and HPC infrastructure."

"At VDURA, performance and scalability are non-negotiable—and that's exactly what the Cornelis CN5000 delivers. Its lossless, high-bandwidth network unlocks the full potential of our next-gen storage platforms, enabling faster insights and seamless AI and HPC integration at scale" - Samantha Clarke, VP, Strategic Partnerships, VDURA