As AI and accelerated computing systems continue to scale in size and complexity, performance alone is no longer enough. Modern infrastructure must deliver both high-speed connectivity and simple, reliable management at scale. True scalability depends on ensuring that powerful accelerator fabrics can also be efficiently monitored, secured and managed across multi-vendor environments.
This is where the collaboration between the UALink Consortium and DMTF becomes important. Through DMTF’s Alliance Partner program, the two organizations have agreed to a formal work register that establishes technical collaboration and alignment between their respective standards efforts.
While each organization focuses on a different layer of the technology stack, together they create a more complete foundation for scalable AI systems. UALink enables fast, efficient data movement between accelerators based on open standards and DMTF provides the standards required to discover, monitor, configure and secure those systems consistently across vendors.
UALink is an open, high-performance interconnect designed specifically for accelerator-to-accelerator communication. By delivering low latency and high bandwidth, UALink allows GPUs and AI accelerators to operate as a unified compute resource, critical for large models and demanding AI workloads.
At the management layer, DMTF focuses on open standards for system management. Specifications such as Redfish®, PLDM®, and SPDM provide consistent mechanisms for hardware discovery, telemetry, configuration and security. These standards enable operators to manage complex, heterogeneous infrastructure environments with greater visibility and operational efficiency.
The alliance between UALink and DMTF is designed to bridge these capabilities. Its goals include enabling a more holistic management experience for accelerator fabrics, promoting adoption of both organizations’ standards among member companies and expanding UALink’s participation within DMTF’s Alliance Partner program and working groups. The collaboration will also introduce new Redfish modeling requirements to support UALink devices and fabrics, helping ensure that next-generation interconnect technologies can be seamlessly integrated into standardized management frameworks.
As AI infrastructure continues to evolve, open ecosystems and cross-organizational collaboration will be essential. By aligning high-performance interconnect innovation with mature, open system management standards, UALink and DMTF are working together to support scalable, secure and operationally efficient AI deployments.
To learn more about this alliance, the work register detailing the partnership is available here.