Cornelis CN6000 Networking Platform Delivers Industry-leading 1.6 Billion Messages per Second at 800G for AI and HPC
Brings more than a decade of supercomputer-scale networking and consecutive first-silicon successes to mainstream AI;
Press Release Disclaimer: This is a press release distributed through the XPR Media network. It has not been independently verified by our newsroom.
![]()
Cornelis today announced the Cornelis® CN6000™ family, its next-generation 800G networking platform engineered to address the demanding communication requirements of AI and high-performance computing (HPC). Drawing on Cornelis’ HPC networking heritage, CN6000 is designed to reduce networking bottlenecks in parallel workloads, from large-scale scientific simulation to AI training and disaggregated inference, helping organizations make better use of their compute investments. With CN6000, Cornelis takes the fabric technology proven in leadership-class supercomputers to the standards-based Ethernet that runs mainstream AI. OEM and system partners across the AI and HPC ecosystem are planning validation and support for CN6000.
The CN6000 family supports Omni-Path and RoCEv2 Ethernet SuperNICs, along with the CN6000 Omni-Path switch. With 800G connectivity and a choice of networking fabrics, customers can select the infrastructure that best fits their workloads and existing environments. Omni-Path delivers Cornelis’ highest performance, and the Ethernet SuperNIC gives Ethernet-first customers a standards-based option. CN6000 is Cornelis’ second consecutive product generation to achieve first-silicon-to-production success, bringing SuperNICs and switches to customers for evaluation today and to partner systems in Q1 2027.
“Every minute an accelerator waits on the network is capacity our customers paid for and didn’t use,” said Lisa Spelman, CEO of Cornelis. “We built CN6000 for the communication patterns that make parallel AI and HPC workloads so demanding. Omni-Path delivers the highest message rate in the industry with lossless delivery at scale, and our new Ethernet SuperNIC brings that same parallel computing expertise to the standards-based fabrics where most AI infrastructure runs. That combination of performance and choice is what our customers need as they scale training, inference, and scientific computing.”
Designed for Demanding AI and HPC
CN6000 is built to keep accelerators working. The Omni-Path SuperNIC doubles the bandwidth of CN5000 to 800G and delivers 1.6 billion messages per second, the industry’s highest. Its sub-microsecond MPI latency is 21% lower than NVIDIA InfiniBand XDR at the same 800G speed.¹ For tightly coupled HPC codes that exchange millions of small messages, that gap shortens every communication step. The SuperNIC’s multi-plane and multi-rail support lets customers build fabrics that isolate traffic and stay up through link failures. In fabrics built with the CN6000 Omni-Path switch, adaptive routing spreads traffic across every available path to reduce congestion and keep data moving efficiently as workloads scale. For Ethernet environments, the new Ethernet SuperNIC brings low-latency RoCEv2 to industry-standard Ethernet switches. Ultra Ethernet support will follow through software updates.
The CN6000 Family
The new CN6000 family includes:
- Omni-Path SuperNIC: 800G connectivity, 1.6 billion messages per second, sub-microsecond MPI latency, multi-plane and multi-rail support, and credit-based lossless delivery. A dedicated hardware Root of Trust with SPDM 1.4 attestation and post-quantum (ML-DSA) signatures verifies firmware before the SuperNIC joins the system.
- CN6000 Omni-Path switch: A 1U switch completes the 800G fabric with 48 ports of 800G, adaptive routing and native multi-plane support.
- Ethernet SuperNIC: Low-latency RoCEv2 connectivity with industry-standard Ethernet switches, plus Ultra Ethernet support through software updates.
- Open software ecosystem: Support for NCCL, RCCL, libfabric, MPI, and OpenSHMEM on Omni-Path, using the same software and host driver as CN5000.
Broad Ecosystem Momentum
AI and HPC system partners across the industry support Cornelis networking today. CN6000 supports AMD, Arm-based, and Intel CPU platforms and AMD, NVIDIA, and Qualcomm accelerators. OEM and system partners planning validation and support include Dell Technologies, Hyve Solutions, Lenovo, NEC, Penguin Solutions, Supermicro and more.
AMD
This week, Cornelis announced expanded collaboration with AMD to help HPC and AI for Science customers design, validate and deploy systems pairing AMD EPYC™ CPUs and AMD Instinct™ GPUs with Cornelis® networking. The program combines AMD compute platforms and engineering with the Cornelis fabric, so customers get more choice in compute, systems validated on their own applications, and a direct path to buy and deploy through the OEMs and system integrators they already use. Cornelis is validating CN6000 with AMD EPYC 9006 Series CPUs and next-generation AMD Instinct GPU platforms, with OEM systems expected beginning in Q1 2027. Learn more at cornelis.com/company/newsroom.
Intel
Intel and Cornelis continue to validate and optimize Xeon platforms with Omni-Path for AI and HPC customers. Systems built on that work run in production today at leading national labs and institutions, including the Texas Advanced Computing Center (TACC). The companies are now extending that collaboration to CN6000 and next-generation Xeon platforms.
“Intel Xeon and Cornelis Omni-Path have run together in production HPC systems for years, and our joint customer labs and workload optimization work show up in the results those systems deliver,” said Jeff McVeigh, Vice President & GM of Data Center Strategic Programs, Intel. “CN6000 brings an 800G fabric to Xeon 6 platforms over PCIe Gen6 and supports Ethernet and Omni-Path. It fits the open, standards-based infrastructure Intel is building and will give our customers performance and deployment flexibility. We’re proud to partner with Cornelis to deliver exceptional performance on our customers’ most demanding workloads.”
Lenovo
Cornelis and Lenovo have an established partnership across AI and HPC. CN5000 is qualified across Lenovo ThinkSystem™ V3 and V4 servers and supported in Lenovo EveryScale, with joint reference architectures for CAE and multi-node HPC on Neptune™ liquid-cooled systems. Lenovo is now qualifying Cornelis with its new AMD EPYC-powered Lenovo ThinkSystem servers to support networking for demanding AI and HPC workloads.
“AI and HPC are converging around increasingly large, distributed workloads that demand the right combination of compute, networking and energy efficiency,” said Martin Hiegl, Executive Director, Infrastructure Solutions Group, Lenovo. “Lenovo is bringing these technologies together to help customers build high-performance AI and HPC environments that can scale from advanced simulation to AI-driven research and discovery. Lenovo ThinkSystem infrastructure provides a high-performance foundation for the compute, storage and networking capabilities customers need to accelerate time to insight while maximizing infrastructure efficiency.”
Penguin Solutions
Penguin Solutions designs, builds, and operates AI and HPC solutions for commercial, research, and government customers. The company plans to validate and support CN6000 in those systems.
“The network largely determines how much of their compute investment our customers put to work,” said Phil Pokorny, Chief Technology Officer at Penguin Solutions. “As AI and HPC workloads continue to evolve, high-performance, low-latency networking is essential to maximizing infrastructure performance and utilization. At Penguin Solutions, we design and build complex AI and HPC environments that bring together compute, networking, storage and software to meet demanding workload requirements. Innovations like Cornelis’ CN6000 help us deliver the performance, scalability and reliability our customers need to maximize the value of their infrastructure investments.”
Supermicro
Supermicro has validated Cornelis CN5000 across its FlexTwin™, BigTwin™ and Hyper™ platforms and plans to extend that work to CN6000.
Availability
The Cornelis CN6000 family, including SuperNICs, switches and software is available from Cornelis today for evaluation and early deployments, with systems from OEM and system partners beginning in Q1 2027. Cornelis will demonstrate the CN6000 family at SC26 in Chicago, Nov. 15–20, at booth #1419.
Product details are available at cornelis.com/products/cn6000.
About Cornelis
Cornelis delivers high-performance, scale-out and scale-up networking solutions that accelerate AI and HPC workloads. 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 www.cornelis.com.
¹ Based on Cornelis internal testing. 8-byte MPI Ping-Pong latency using Intel MPI, CN6000 Omni-Path SuperNIC vs. NVIDIA InfiniBand XDR at 800 Gb/s: 0.95 µs vs. 1.2 µs.
FAQ
What is Cornelis announcing with CN6000? Cornelis is announcing the launch of CN6000, its next-generation 800G networking platform for AI and HPC. The CN6000 family includes Omni-Path and Ethernet SuperNICs, the CN6000 Omni-Path switch, and software.
What makes CN6000 different? CN6000 offers Omni-Path and RoCEv2 Ethernet SuperNICs. The Omni-Path SuperNIC delivers 1.6 billion messages per second, the industry’s highest, and 21% lower MPI latency than NVIDIA InfiniBand XDR at 800G. The Ethernet SuperNIC supports RoCEv2 and Ultra Ethernet through software updates.
Who will offer CN6000? CN6000 is planned for validation and support across Dell Technologies, Hyve Solutions, Lenovo, NEC, Penguin Solutions, Supermicro and more.
Does the CN6000 Ethernet SuperNIC work with my existing switches? Yes. It interoperates with industry-standard Ethernet switches.
Which switch does the CN6000 Omni-Path SuperNIC use? The CN6000 Omni-Path switch, which adds adaptive routing across the fabric. CN6000 Omni-Path SuperNICs are also designed to run in existing CN5000 Omni-Path fabrics to provide deployment flexibility for customers.
View source version on businesswire.com: https://www.businesswire.com/news/home/20261112417616/en/
Media gallery

