AMD has unveiled Helion, a customizable rackscale offering designed to support agentic AI workloads.

Announced by AMD at the OCP Global Summit in San Jose today (October 12), the company said Helion builds on the “open-platform approach” of its Helios rackscale solution, which was launched earlier this year.

Designed for hyperscale customers, AMD said users can customize Helion’s configurations by integrating CPU and AMD Instinct GPU nodes in order to match agent execution and model inference, and tailor compute to each deployment’s performance, power, and cost needs.

Customers will also be able to deploy a CPU-only version of the rack, powered by 6th Gen AMD Epyc 'Venice' server CPUs, supporting up to 32,768 CPU cores.

Available in both liquid and air-cooled versions, AMD says this approach allows users to “create virtually any configuration needed” to support their requirements.

“Agentic AI is not a single workload. It spans personal agents, enterprise applications, code execution and complex workflows, each with different compute requirements,” said Dan McNamara, SVP and GM, compute and enterprise AI, AMD. “To do that takes optimized compute and infrastructure that can adapt to different workloads. With AMD Helion, we are bringing AMD Epyc performance, open software and flexible rack designs together so our partners can build systems around customer needs and make efficient use of available power.”

AMD said it plans to contribute its Helion specification to OCP, making it available to the community for further development and adoption. At present, the Helion agentic rack configurations are still under development with ecosystem partners, but AMD expects them to be available in November 2026

Ecosystem partners working with AMD for Helion’s launch include Cisco, Dell, Hon Hai Technology (aka Foxconn), HPE, Lenovo, Supermicro, and Wiwynn.