AMD Introduces Helion Agentic AI Rack With Support From Major Technology Partners

Advanced Micro Devices (NASDAQ:AMD) has unveiled AMD Helion, a new rack-scale infrastructure platform designed to support agentic artificial intelligence workloads across hyperscale data centres and enterprise computing environments.

Announced on 12 October 2026, the platform provides an open infrastructure blueprint based on an Open Compute Project (OCP) specification design, combining AMD processors, graphics processing units and integrated software capabilities.

The company said Helion is intended to address the varied computing requirements of agentic AI, which involves artificial intelligence systems capable of carrying out tasks, coordinating workflows and interacting with different models and applications.

Unlike infrastructure designed primarily for individual AI model operations, the new platform aims to support multiple types of workloads within a common computing architecture.

AMD Helion will be offered in flexible configurations, including both liquid-cooled and air-cooled systems, allowing customers to select infrastructure suited to their computing requirements and data centre environments.

Two Rack Configurations Target Different AI Workloads

AMD outlined two primary configurations for the Helion platform, targeting different segments of the AI infrastructure market.

The first is a CPU-only rack powered by sixth-generation AMD EPYC processors, designed for large-scale computing deployments.

This configuration supports up to 32,768 CPU cores and is aimed primarily at hyperscale operators requiring substantial processing capacity for distributed AI workloads.

The second configuration combines AMD EPYC CPU nodes, AMD Instinct MI350P GPU nodes and general-purpose computing resources within a single rack.

This all-in-one architecture is intended for enterprise agentic AI deployments, where applications may require different combinations of CPU and GPU processing capabilities.

By integrating multiple computing resources, AMD aims to provide a platform capable of supporting a broad range of AI applications without requiring separate infrastructure designs for every workload.

The system also includes an integrated token router that directs model calls across frontier AI models and open-source alternatives.

This capability is designed to support the coordination of different models within agentic AI applications, where individual tasks may have varying computational 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, senior vice president and general manager, Compute and Enterprise AI, AMD.

The company said the platform’s open hardware and software architecture is intended to provide flexibility for organisations developing and deploying agentic AI systems.

AMD Plans Open Compute Project Contribution

AMD intends to contribute the Helion specification to the Open Compute Project, an industry initiative focused on developing open designs for data centre infrastructure.

The planned contribution would allow the wider technology community to participate in the development and evolution of the platform’s specifications.

The approach reflects AMD’s emphasis on open infrastructure standards as demand grows for AI computing systems that can accommodate different hardware configurations and software environments.

Development of the Helion rack configurations is continuing in collaboration with technology and manufacturing partners.

Companies supporting the platform include Cisco, Dell Technologies, Hon Hai Technology Group, Hewlett Packard Enterprise (HPE), Lenovo, QCT, Supermicro and Wiwynn.

The participation of these companies provides AMD with an established ecosystem of server manufacturers, infrastructure providers and technology partners to support the platform’s development.

AMD said Helion rack configurations are planned to become available in November 2026.

The announcement expands AMD’s infrastructure offering for AI applications, combining its EPYC processor family and Instinct GPU technology within a rack-scale architecture.

As enterprises and hyperscale operators explore increasingly complex AI workloads, Helion represents AMD’s effort to provide an open, configurable infrastructure option for agentic AI deployments.

However, the company has not disclosed pricing, expected sales volumes or specific customer deployment commitments for the new platform.

Advanced Micro Devices stock price