AMD
“The release of UALink’s second-generation specification marks an important milestone for open, high-performance AI infrastructure. By advancing capabilities such as In-Network Computing while also delivering the organization’s first management and chiplet specifications, the consortium is laying the foundation for scalable, interoperable accelerator fabrics. AMD is proud to support UALink’s progress toward an open ecosystem that is accelerating the industry toward a future where intelligent compute fabrics are open, composable, and optimized for AI at scale.”
Kurtis Bowman, Director of Architecture and Strategy, AMD
Astera Labs
“Hyperscalers can benefit from In-Network Compute to further improve performance, standardized chiplets to quickly enable UALink integration into their accelerators, and management tools that integrate seamlessly into existing infrastructure. These new UALink specifications deliver all three. They also demonstrate a core advantage of open standards: the entire ecosystem moves faster than any single vendor can alone—and that velocity matters when you’re deploying at the pace of AI innovation.”
Chris Petersen, VP, CTO Office, Astera Labs and Board Member, UALink Consortium
Google Cloud
Google welcomes the ratification of the latest UALink specifications as an important step forward in establishing an open, high-performance foundation for next-generation AI infrastructure. The introduction of UALink Manageability—powered by gNMI and Redfish—is essential for delivering the scale, reliability, and interoperability required by modern AI workloads.
Amber Huffman, Principal Engineer, Google Cloud
Synopsys
“As AI workloads scale across large accelerator clusters, the interconnect bandwidth, latency, and efficient memory access have become core system requirements. Open scale-up interconnects such as UALink are critical to meeting these needs by enabling high performance, efficiency, and multi-vendor interoperability. As an active member of the UALink Consortium, Synopsys brings deep interface IP expertise along with proven 224G and security IP to help enable first-pass deployment of the UALink ecosystem, reducing integration risk and accelerating time to market.”
Priyank Shukla, Director of Product Management at Synopsys
UnifabriX
“UnifabriX supports UALink as a fundamental enabler of open, memory centric scale up for AI accelerators. The UALink 2.0 Specification advances the industry’s ability to build flexible, high performance AI systems while preserving interoperability and long term architectural choice.”
Ronen Hyatt, UnifabriX, CEO and Chief Architect
View the full document: https://ualinkconsortium.org/wp-content/uploads/2026/04/UALink_2.0-Statement-of-Support_FINAL.pdf