nVent and AMD team up on cooling stack for AMD Helios RackScale AI systems

nVent and AMD said they are collaborating to add a combined cooling solution to AMD’s Helios RackScale rack-level AI platform, targeting what they describe as a more fully integrated system for large AI workloads.

Under the collaboration, nVent will contribute liquid and air cooling capabilities to the AMD Helios RackScale approach. The companies’ goal is to reduce the friction between compute and infrastructure layers by pairing AMD’s rack-scale design with an engineered cooling stack from nVent.

The announcement also points to openness within the broader industry ecosystem. The companies plan to unveil a first look at an open-source version of the cooling solution at the Open Compute Project (OCP), a community that develops and shares standardized infrastructure designs for data centers. The companies did not provide in the announcement which specific elements will be open-sourced or how the reference design will be licensed.

AMD’s Helios RackScale platform is positioned as a building block for AI data centers, designed to bring more of the system-level configuration under one architecture. Cooling is a central part of that equation because AI workloads typically drive high and sustained thermal demand, and operators often manage cooling as a separate procurement and integration problem.

Beyond the general description of adding liquid and air cooling, the companies did not disclose in the post any detailed technical parameters, such as target thermal limits, fluid specifications, rack densities, or power and airflow profiles. They also did not state whether the cooling will be delivered as a complete rack-level assembly, as modular components, or as a set of design guidelines for system integrators to implement.

The announcement did not include timing for broad availability, pricing, or production commitments. It also did not name specific data center customers, integrators, or deployment sites that would validate the design. For now, the key near-term milestone appears to be the public first look at OCP, where the partners plan to demonstrate the direction of the combined solution.

For the AI infrastructure supply chain, cooling partnerships like this are increasingly important. As organizations build out clusters for training and inference, they often face a tradeoff between custom integration and faster deployment. A rack-scale platform that pairs compute architecture with a defined cooling approach can help align performance targets, operational procedures, and serviceability expectations.