AMD officially launched the Helion Agent AI rack on October 12, an open blueprint based on Open Compute Project (OCP) standards, featuring AMD's own computing chips designed specifically for agent AI workloads.
The platform offers two primary configurations, designed for hyperscale data centers and enterprise customers. AMD states that partner-specific configurations are currently in development and are scheduled for release in November 2026.
Eight companies—Cisco, Dell, Foxconn’s parent company Hon Hai Technology Group, Hewlett Packard Enterprise (HPE), Lenovo, QCT, Supermicro, and Wiwynn—have announced support for the platform, covering the entire ecosystem from system integration to liquid cooling solutions.
Two configurations, designed for hyperscale and enterprise scenarios respectively.
The first configuration of the Helion platform is a pure CPU rack powered by sixth-generation AMD EPYC processors, scalable up to 32,768 CPU cores. AMD states that this configuration is designed for hyperscale customers, offering dedicated agent execution capacity independent of model inference infrastructure.
The second type is an integrated rack that combines AMD EPYC CPU nodes, AMD Instinct MI350P GPU nodes, and general-purpose computing on a single platform, designed for enterprise-grade AI agent deployment.
Both configurations support liquid-cooled and air-cooled versions. Additionally, Helion features an integrated token router that distributes model calls from AI agents to leading and open-source models.
AMD executives: Agent AI is not a single workload
Dan McNamara, Senior Vice President and General Manager of AMD’s Computing and Enterprise AI, stated: “Agent AI is not a single workload—it encompasses personal agents, enterprise applications, code execution, and complex workflows, each with distinct computational requirements.”
He further stated, "AMD Helion integrates EPYC performance, open software, and flexible rack designs, enabling partners to build systems around customer needs and make efficient use of available power."
AMD also stated that Helion is built upon the design philosophy of the existing AMD Helios rack-level solution, aiming to enable partners and customers to match CPU and GPU configurations with specific workload requirements, including performance, power consumption, and cost targets.
Backed by eight leading manufacturers, with a fully integrated ecosystem.
All eight partners have issued statements expressing support for the Helion platform.
Arun Narayanan, Senior Vice President of Computing and Networking at Dell Technologies, stated that the combination of AMD Helion and Dell PowerRack infrastructure provides customers with “a complete, validated path for large-scale agent computing deployment.”
Trish Damkroger, Senior Vice President of High-Performance Computing and AI Infrastructure Solutions at HPE, said the two companies are “opening the next era of AI with open, large-scale systems,” noting that the solution supports both air-cooled and direct liquid-cooled network options.
Jeremy Foster, Senior Vice President of Cisco's Computing Business, said: "Agent AI is transforming customer requirements for infrastructure—not just in terms of computing performance, but also in the connectivity, management, and security of the entire system. Architectures like AMD Helion, combined with Cisco’s full-stack infrastructure capabilities, provide customers with greater flexibility."
Supermicro’s Chief Business Officer, Vik Malyala, said that Supermicro will bring Helion to market through its Data Center Building Block Solutions (DCBBS), “helping customers deploy faster, scale efficiently, and accelerate the deployment of agent AI.”
AMD stated that it plans to submit the Helion specification to the Open Compute Project (OCP) for further development and adoption by the broader industry community.
This initiative continues AMD’s open platform strategy on the Helios platform, aiming to expand ecosystem influence through open standards and attract more partners to build products around this specification.