AMD launches Helion intelligent agent AI rack design, supporting up to 32,768 cores, jointly endorsed by eight manufacturers including Cisco and Dell.

On October 12, AMD officially launched the Helion Agent AI Rack, an open blueprint based on Open Compute Project (OCP) standards. It is equipped with AMD’s own computing chips and is designed specifically for agent AI workloads.

The platform offers two main configurations, targeting hyperscale data centers and enterprise-level customers respectively. AMD stated that partner-configured products are currently under development and are expected to be launched in November 2026.

Eight companies—Cisco, Dell, Hon Hai Technology Group (parent company of Foxconn), HPE, Lenovo, QCT, Supermicro, and Wiwynn—have announced their support for the platform, building a complete ecosystem ranging from system integration to liquid cooling solutions.

Two configurations for hyperscale and enterprise scenarios

The first configuration of the Helion platform is a pure CPU rack, featuring the 6th generation AMD EPYC processor and scalable up to 32,768 CPU cores. According to AMD, this configuration is designed for hyperscale customers, offering dedicated agent execution capacity independent from model inference infrastructure.

The second is an integrated rack that combines AMD EPYC CPU nodes, AMD Instinct MI350P GPU nodes, and general-purpose computing into a single platform, positioned for enterprise-level agent AI deployments.

Both configurations support liquid cooling and air cooling versions. In addition, Helion includes a built-in integrated token router, which can distribute AI agent model calls to both frontier and open source models.

AMD Executive: Agent AI is not a single workload

Dan McNamara, Senior Vice President and General Manager of AMD Computing and Enterprise AI, stated in a release: “Agent AI is not a single workload; it encompasses personal agents, enterprise applications, code execution, and complex workflows, each scenario having different computing demands.”

He further said, “AMD Helion integrates EPYC performance, open software, and flexible rack design, enabling partners to build systems around customer needs and efficiently utilize available power.”

AMD also stated that Helion builds on the design philosophy of the existing AMD Helios rack-level solutions, aiming to allow partners and customers to match CPU and GPU configurations with specific workload requirements, including performance, power consumption, and cost objectives.

Eight major manufacturers endorse for a complete ecosystem

All eight partners issued statements expressing their support for the Helion platform.

Arun Narayanan, Senior Vice President of Dell Technologies Compute & Networking, stated that the combination of AMD Helion with Dell PowerRack infrastructure provides customers with “a complete, validated agent compute deployment path at scale.”

Trish Damkroger, Senior Vice President of HPC and AI Infrastructure Solutions at HPE, commented that the two companies are “opening the next era of AI with open, large-scale systems,” and pointed out that the solution offers both air cooling and direct liquid cooling network options.

Jeremy Foster, Senior Vice President of Cisco Compute Business, stated: “Agent AI is changing customers’ infrastructure needs, not just in terms of computing performance but also in system-wide connectivity, management, and security. Architectures like AMD Helion, combined with Cisco’s full-stack infrastructure capabilities, provide customers with greater flexibility.”

Vik Malyala, Chief Commercial Officer of Supermicro, said Supermicro will bring Helion to market through its Data Center Building Block Solutions (DCBBS), “helping customers deploy faster, scale efficiently, and accelerate go-live with agent AI.”

AMD announced 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 move continues AMD’s open platform strategy from the Helios platform, aiming to expand ecological influence through open standards and attract more partners to build products around this specification.