The Linux Foundation Projects
Skip to main content


Steering Members

Highly compute-intensive workloads – such as AI training, machine learning, and HPC simulation and modeling – require scalable and cost effective industry-wide solutions with interoperability as a top priority. In an effort to create an open Ethernet-based architecture to address the evolving needs of modern data center workloads, we are joining the Ultra Ethernet Consortium as a founding member. AMD has a long history of supporting open industry standards and we are proud to continue on that course today with the UEC.

Robert Hormuth
Corporate Vice President, Architecture and Strategy, Data Center Solutions Group, AMD

As one of the founding members of UEC, Broadcom is excited to see so many new companies join and actively contribute to the success of Ethernet for AI and HPC networking. Broadcom has long been committed to Ethernet technology, driving innovations in all aspects of the hardware and software network stacks. We look forward to working with all of the members of the growing consortium to drive Ethernet to meet the needs of next-generation AI training, AI inference, and HPC networks.

Ram Velaga
Senior Vice President and General Manager, Core Switching Group, Broadcom

To meet the demands of generative AI, it’s imperative to design our networks with a focus on supercomputing scale and efficiency. The Ultra Ethernet Consortium aims to create an open, scalable, and cost-effective networking solution based on Ethernet, delivering the performance and capabilities required to support high-performance workloads. The widespread use and interoperability of Ethernet offers customers the necessary performance to handle diverse, data-intensive tasks such as simulation and AI model training and fine-tuning. Parallelized computing solutions offer the optimal performance, reliability, and sustainability to accommodate a sharp growth in data volume and increased complexity of models that will arise as AI adoption continues.

Barry Davis
Senior Director, High Performance Interconnects, Hewlett Packard Enterprise

As we continue to build our global infrastructure and develop the systems of the future, we appreciate the work that UEC is leading. Meta is committed to an open source environment because we know that technologies such as ethernet have shown time and again that open standards prevail. As we move towards the next phase of high performance computing for AI & other use cases, the scale of systems may move beyond what was previously imagined and efforts such as UEC will be key to the industry’s success.

Nic Viljoen
AI Systems Engineer, Meta

Oracle is proud of its involvement with UEC! The complete UEC stack represents the first industry-wide high performance transport in over 20 years, that builds on the massive ecosystem of Ethernet and which is widely prevalent in modern data centers. It helps us to offer our AI/ML, Database and HPC customers the benefits of modern RDMA technology that is scalable, reliable and highly performant.

Partha Kundu
Architect, Oracle Cloud Infrastructure, Oracle

General Members

Sanjay KumarVice President, Product Management & Marketing, Arrcus, Inc.

AI will have a transformative impact on everything we do, and needs a new network to help realize its full potential. Ethernet can deliver this next gen network, one that is reliable, accessible, scalable and cost-effective. UEC is playing a pivotal role in bringing this vision to reality by bringing together a community to help solve the key challenges and develop standards to drive wide scale adoption. Arrcus is commited to this mission and excited to be part of this journey.

Jason KeiVP, AI Infra Products, VNET

We believe Ethernet will continue to be the most viable, robust, and scalable solution to high-performance and mission-critical workloads in the Generative AI era. As a leader in data centers and high-performance computing infrastructure, we are committed to working with the UEC community to accelerate the development of the next-generation Ethernet-based high-speed network for the explosive growth of AI data traffic.

Shrijeet MukherjeeChief Development Officer & Co-Founder of Enfabrica

Normalizing the attributes of Ethernet networks for large scale, high-performance, distributed compute has been a critical objective for Enfabrica. Implementing step-function changes in networking has been a challenge for the industry as it needs more than just technology that meets a specification – it requires thoughtful improvement in the interfaces for fabric devices, applications, management, and in the overall distributed systems architecture. We are excited to be an active contributor to Ultra Ethernet Consortium, and to work alongside our industry colleagues to reach solutions grounded in deep technical excellence that further the value of Ethernet to modern AI and HPC networks.

Alessandro CurioniIBM Fellow, Vice President Europe and Africa and Director of IBM Research Zurich

Efficient data communication has powered large-scale distributed compute systems for decades and is now serving the ever-more demanding workloads of the AI era. Driven by the converging infrastructure requirements of next generation High-Performance Computing and generative AI, both in-cloud and on-premise, a converged high performance networking technology is inevitable, and it should be based on an evolution of cost efficient, ubiquitous Ethernet. We support the UEC's open collaboration, which will enhance Ethernet to meet the stringent performance, scalability, and reliability requirements to come.

Mansour KaramGVP of CRDC Product Management, Juniper Networks

Juniper Networks proudly leads the way in innovation, championing the mission of the Ultra Ethernet Consortium to propel high-speed networking advancements that facilitate the AI revolution underway. Through collaboration with this important organization, we strive to cultivate a collective vision and develop the capabilities for next-generation Ethernet solutions that ensure seamless connectivity with unparalleled scalability, performance, and reliability.

Venkat PullelaChief of Technology, Networking, Keysight Technologies

Unprecedented growth in the adoption of generative AI is testing the limits of existing data center networks. AI and ML workloads require fine tuning at the system level across all layers from Ethernet to collective communications to optimize the job completion time (JCT). The Ultra Ethernet Consortium has established a strong platform that brings together leaders to collaboratively design solutions and address this issue head-on. As a market leader in network testing products, Keysight is committed to joining the UEC and sharing its expertise by taking an active role in the UEC, co-leading Compliance & Testing, Performance and Debug working groups.

Nick KucharewskiSVP and GM Network Switching BU, Marvell

Like the UEC, Marvell believes the future of the AI cloud at scale will rely on standards-based Ethernet enabling low-latency fabrics. UEC’s collaborative efforts to define an end-to-end standard aligns with the Marvell commitment to drive innovation for next-generation AI fabrics.

Aniket KhoslaVP of Wireline Product Management, Spirent Communications

Communications networks worldwide rely on Ethernet technology, and testing how the Ethernet fabric of new AI environments performs has proved to be an enormous challenge. As data centers reimagine their architectures for AI, testing is becoming ever more critical. As a leader in Ethernet testing, Spirent is committed to contributing to UEC community efforts to deliver an Ethernet-based open, interoperable, high performance network architecture to meet the growing demands of AI workloads at scale.

John KoeterSenior Vice President of Product Nanagement and Strategy for IP, Synopsys

Ethernet standards are continually evolving to support the insatiable demand for higher speeds and more bandwidth in compute-intensive data center applications. As an active contributor to the Ultra Ethernet Consortium and leading provider of silicon-proven and trusted Ethernet IP solutions, Synopsys is helping to drive adoption of the next generation of the Ethernet interconnects to unlock 1.6Tbps rates that address the stringent performance, power, and latency requirements of advanced data center chips.

Contributor Members

Dr. Ethan HuCEO of Accelink

Accelink is thrilled to be a part of and actively contribute to the Ultra Ethernet Consortium (UEC). The pioneering work undertaken by UEC holds immense importance in facilitating efficient communication for emerging applications such as Artificial Intelligence and Machine Learning, which are experiencing exponential growth and demanding ever-increasing bandwidth capacities. As a prominent leader in optical communication solutions, Accelink is committed to leveraging our expertise and resources in high-speed optics technologies and advanced package platforms to ensure the successful adoption of UEC's groundbreaking outcomes.

Amir ElbazFounder & CEO of BE Networks

As we work to navigate the intricate technical challenges our enterprise and service provider customers face, our goal in the UEC is to forge synergies that surpass individual limitations. At BE Networks, our commitment lies in leveraging our unique industry expertise with other consortium members to deliver a next-generation architectural roadmap for the simplified design, deployment, and ongoing operational management of Ethernet-based data center and cloud network fabrics.

Lior KhermoshCTO of NeuReality

Ethernet serves as the bedrock of scalable Datacenter Networking. Facilitating its seamless extension to support and scale adjacent AI Inference via the UEC is crucial for efficient and sustainable generative AI, LLMs, and of course, conventional AI applications. NeuReality's AI-centric inference system architecture, featuring our Network Addressable Processing Unit (NAPU) directly connected to Ethernet, independent from CPU, paves the way for affordable AI in the world's Data Centers. Committed to open standards and ecosystem development, we actively engage with the UEC community to drive the evolution of next-generation Ethernet Networking for efficient and scalable AI inference for all.

Andrew GarrettSoftware Architect at Salience Labs.

As AI workloads drive the need for higher bandwidth and higher performance networks, optical switches and interconnect offer a compelling route to meet future demands. Standards such as UEC are essential to bring an ecosystem together to solve the challenges of next generation interconnect. Salience Labs brings to the consortium decades of experience in Silicon Photonics to build low loss, high bandwidth optical switches for Data Centre AI.

Sage ZouVice President of Tencent Cloud

Congratulations to UEC on all the remarkable milestones achieved. We firmly believe that AIGC will propel computing and networking into a new era. On behalf of Tencent's IDC and Network Team, we would like to collaborate with the members of UEC to foster network innovation and shape a promising future for Ethernet technology.

Rajagopal KrishnaswamyVice President Marketing & Business Development, XSightLabs

The AI/ML tsunami has further intensified the spotlight on the need for high bandwidth, low latency & efficient interconnect fabrics. UEC’s vision to drive a unified standard that builds on the strengths and leverages the broad ecosystem of Ethernet is strongly aligned with Xsight Labs’ vision to deliver end-to-end connectivity for the datacenter, cloud & AI clusters. UEC’s efforts on this front we believe will lay the foundation for next wave of innovation that will help drive better network efficiency, higher compute resource utilization and lower TCO.

Subscribe to the UEC Newsletter!

Get the latest announcements, event info, and the community news in your inbox.