Supermicro Expands Rack-Scale AI Leadership with AMD Helios Platform

The AMD Helios platform is designed specifically for the era of agentic AI, enabling organizations to handle demanding tasks such as Sovereign AI and large language model (LLM) training, inference, and fine-tuning. This innovative platform is equipped with advanced technologies that promise unmatched efficiency and scalability, making it a game-changer in the rapidly evolving AI landscape.

Charles Liang, President and CEO of Supermicro, emphasized the transformative potential of the Helios platform, stating, "With DCBBS, Supermicro is redefining what is possible in the data center by shifting from traditional server design to a complete rack-scale architecture." He highlighted the synergy between Supermicro's Data Center Building Block Solutions® (DCBBS) and AMD's Instinct™ MI455X GPU architecture as the key to delivering unprecedented AI performance and improved power efficiency.

Ravi Pendekanti, Corporate Vice President of AMD's Data Center Solutions Business Group, echoed Liang’s sentiments, noting that the future of AI will hinge not just on computing power but on the efficiency of deployment and scalability. He stated, "AMD Helios delivers an open, rack-scale AI architecture that brings together leadership AMD compute, networking and software to help customers accelerate time to deployment."

Supermicro is among the first partners to collaborate closely with AMD to bring the Helios solution to market, reinforcing its leadership in delivering comprehensive AI infrastructure solutions.

The Helios platform is a robust 72-GPU double-width rack-scale system powered by AMD Instinct MI455X GPUs, 6th Gen AMD EPYC™ CPUs, and AMD Pensando™ networking technologies, all unified through the open AMD ROCm™ software stack. This architecture is optimized for large-scale AI deployments, offering exceptional compute density and performance for frontier model training and high-throughput inference.

Notable capabilities of the Helios platform include:

These features collectively enable clients to accelerate their time-to-market for AI services, optimize resource utilization, and maintain reliable, high-performance AI capabilities at hyperscale.

Supermicro's A+A+A approach—Architecture, Accelerators, and Advancements—encapsulates its strategy of combining rack-scale system design with leading AMD AI compute solutions and integrated software innovations. This comprehensive approach allows customers to deploy AI infrastructure more rapidly, operate with greater efficiency, and scale seamlessly as demand rises.

The DCBBS framework provides a complete, modular AI infrastructure built from validated components and subsystems, making it versatile for deployment ranging from individual servers and networking solutions to full rack-scale and data center-level systems.

Attendees at Computex will have the opportunity to see the AMD Helios rack-scale solution in action at Supermicro's booth located in the Taipei Nangang Exhibition Center Hall 1, 4F, N0602. The display will include an interactive demo through Supermicro's A+ Superverse, allowing visitors to explore the platform’s design and capabilities firsthand.

Supermicro’s announcement regarding the AMD Helios platform signifies a pivotal step in the advancement of AI infrastructure solutions. With a focus on efficiency, scalability, and robust performance, Supermicro and AMD are poised to meet the growing demands of AI workloads in an increasingly digital world. As organizations look to harness the power of AI, the Helios platform stands ready to deliver the next generation of computing capabilities, ensuring that clients can navigate the complexities of AI deployment with agility and confidence.