CONFIRMED: AMD announced Helion in a press release on 12 October 2026. AMD says Helion rack configurations are planned to be available in November 2026.
AMD has introduced the AMD Helion agentic AI rack, which it calls “an open blueprint for agentic AI infrastructure based on an Open Compute Project spec-design and powered by AMD compute” [1]. Agentic AI means AI agents: software that carries out tasks step by step, such as running code or calling other tools, rather than just answering a single question.
Helion is a different product from AMD Helios, AMD’s rack-scale platform. AMD says Helion builds “on the open-platform approach” of Helios [1].
What AMD announced
AMD gave two example configurations [1]:
- A CPU-only rack “powered by 6th Gen AMD EPYC processors that can support up to 32,768 CPU cores”. AMD says it is designed for hyperscale customers (the biggest cloud operators), “providing dedicated capacity for agent execution and connections to separate model inference infrastructure”.
- An all-in-one rack that “combines AMD EPYC CPU nodes, AMD Instinct MI350P GPU nodes and general-purpose compute on one platform”, aimed at enterprise agentic AI.
AMD says Helion can come in “liquid- and air-cooled versions” and includes “an integrated token router that directs agents’ model calls across frontier and open-source models” [1]. In plain terms, the router decides which AI model each request from an agent is sent to.
“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, senior vice president and general manager, Compute and Enterprise AI, AMD [1].
Open specification and timing
“AMD plans to contribute the AMD Helion specification to OCP, making it available to the community for further development and adoption,” the company said [1]. The Open Compute Project (OCP) is an industry group that shares open hardware designs for data centres.
AMD said Helion configurations “are currently under development with ecosystem partners and are planned to be available in November 2026” [1]. AMD did not give pricing or name customers.
Partners
The release carries supporting statements from Cisco, Dell Technologies, Hon Hai Technology Group, HPE, Lenovo, QCT, Supermicro and Wiwynn [1]. Supermicro called Helion “the open, ORv3 compliant rack-scale architecture” and said it will bring it to market through its Data Center Building Block Solutions [1].
Sources
- AMD, “Introducing AMD Helion, Agentic AI Compute at Rack Scale”, press release (GlobeNewswire), 12 October 2026. https://ir.amd.com/news-events/press-releases/detail/1301/introducing-amd-helion-agentic-ai-compute-at-rack-scale
