Hello UEC Community,
As we close out 2025, I want to take a moment to thank all our members for the work, the ideas, and the energy you continue to bring to this consortium. Whether you have been with UEC from the beginning or joined us this year, your contributions are what make this community thrive. We set out to build the best Ethernet-based high-performance network for AI and HPC workloads, and because of you, we are taking confident steps toward that goal.
A Landmark Year for UEC: Specification 1.0
This year was truly a milestone for us. In June, we officially released the UEC 1.0 Specification. It was an ambitious goal, completed in a very short amount of time. It would not have happened without the dedication of our working groups, editors, and contributors across the consortium.
UEC 1.0 is more than a technical document. It represents interoperability, standardization, and readiness for real-world use. We’ve already seen encouraging signs of this through early ecosystem activity, including new hardware from (as of this writing) UEC members like Broadcom, AMD, Keysight, Nokia, Arista, and Eviden (it’s always a risk to enumerate members; you always run the risk of forgetting someone!). We also took a moment to celebrate notable achievements within our community, such as Torsten Hoefler receiving the 2024 ACM Prize in Computing, a recognition that reflects the kind of technical excellence that continues to shape UEC’s direction.
Since the release, the momentum has continued. Members have been advancing several important updates for the next iteration of the spec, including flexible congestion management mechanisms, enhanced congestion signaling, improvements for small-message performance, optimized scale-up transport, and in-network collective operations. These efforts may not always be visible outside the working groups, but they are essential to ensuring that UEC remains ready for deployment at scale.
Our community has also continued to grow. This year, we welcomed 27 new member companies across all facets of data center AI infrastructure. We have stayed closely connected with partner organizations such as SNIA, OCP, IEEE, and NVM Express. This broader ecosystem involvement shows just how important UEC has become beyond our immediate circle.
What’s Next for 2026
As we move into 2026, we are focusing on two important priorities.
First, we want to help the industry implement UEC with confidence. The specification needs to be more than “just” correct on paper. It needs to be usable and trusted by the engineers who work with it. Supporting implementations and helping members put UEC into practice will be a major focus in the year ahead.
Second, we are investing in education. Interest in UEC is growing quickly, and we need to make sure that the people who want to adopt it have the right tools and guidance. Clear documentation and practical examples will help everyone understand how to use UEC effectively. Stay tuned in 2026 for these guides, which we hope will make the standard even more approachable. Please don’t hesitate to let us know if there’s anything in particular of interest that we can cover!
These priorities reinforce the idea that UEC is not just a collection of features. It is an end-to-end system, built for real workloads and real environments. Our job now is to help the world use it with clarity and confidence.
In Closing
Thank you again for everything you have done throughout 2025. We have reached an important milestone together, and the next phase of our work is already taking shape. I am grateful for your continued support, and I am looking forward to all we will achieve in 2026.
Best regards,
J Metz, Ph.D.
Chair, Ultra Ethernet Consortium
