Our first year in review: preparing for what comes next. A letter from the chair
Hello NoLimit community,
As we approach the foundation's first anniversary, I want to take a moment to thank all of our members for the work, the ideas, and the energy you continue to bring. Whether you have been here since October or joined this summer, 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: Architecture 1.0
This was a milestone year. In May we released Architecture 1.0. It was an ambitious target completed in a very short time, and it would not have happened without the dedication of our working groups, editors, and contributors.
Architecture 1.0 is more than a technical document. It represents interoperability, standardization, and readiness for real-world use. We have already seen encouraging signs of that through early ecosystem activity, including new hardware from members such as Broadcom, AMD, Keysight, Nokia, Arista, and Eviden. It is always a risk to enumerate members, because you always run the risk of forgetting someone. We also paused to celebrate achievements within the community, such as Torsten Hoefler receiving the ACM Prize in Computing, a recognition of the kind of technical excellence that continues to shape our direction.
Since the release, the momentum has continued. Members are advancing important updates for the next iteration of the architecture: flexible congestion management, enhanced congestion signaling, improved small-message performance, an optimized scale-up transport, and in-network collective operations. These efforts are not always visible outside the working groups, but they are essential to keeping the architecture ready for deployment at scale.
Our community kept growing as well. This year we welcomed new member companies across every facet of data center AI infrastructure, and we stayed closely connected with partner organizations such as SNIA, OCP, IEEE, and NVM Express. That broader involvement shows how far the foundation's influence now reaches beyond its immediate circle.
What's next
Looking to the coming year, we are focused on two priorities.
First, helping the industry implement the architecture with confidence. It needs to be more than correct on paper. It needs to be usable and trusted by the engineers who work with it. Supporting implementations and helping members put NoLimit into practice will be a major focus.
Second, education. Interest in the foundation is growing quickly, and the people who want to adopt the architecture need the right tools and guidance. Clear documentation and practical examples will help everyone understand how to use it effectively. Expect those guides over the coming months, and please tell us if there is something specific you would like covered.
Both priorities reinforce the same idea: NoLimit is not 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 this year. We have reached an important milestone together, and the next phase of the work is already taking shape. I am grateful for your continued support and looking forward to all we will achieve in the year ahead.
Best regards,
Everett Caldwell
Chair, NoLimit Foundation