Schneider Electric and AMD have unveiled their first co-engineered reference design for Helios, enabling rack-scale AI clusters up to 10.4 MW with 246 kW per rack and advanced liquid cooling to achieve a target PUE around 1.12. The design, built around AMD Instinct MI455X GPUs, 6th Gen AMD EPYC CPUs, Pensando Vulcano NICs and ROCm software, provides a validated, scalable blueprint that streamlines planning and deployment of high-density AI environments. It supports greenfield deployments and multi-cluster retrofits, allowing operators to scale capacity progressively while managing power and cooling demands. Liquid cooling sits at the center of the architecture, capable of removing up to 84% of heat per rack to sustain higher densities and reduce energy use. In parallel, Schneider Electric is expanding its AI-data center roadmap with Nvidia collaboration to produce publicly available reference designs for AI deployments, underscoring a broader industry push toward standardized, efficient, and scalable data-center solutions for AI workloads.