Urgent.News

What's breaking now, across thousands of outlets.

AI

AMD unveils Helion racks for AI agents at OCP summit

CPU-only and mixed CPU/GPU configurations due from partners in November

AMD unveils Helion racks for AI agents at OCP summit

AMD has introduced its Helion rack design for artificial intelligence agents at the Open Compute Project (OCP) Summit 2026 in San Jose, California. This new rack reference design complements AMD's existing Helios system, which focuses on frontier-scale AI training and large-scale inference. Helion is engineered to support various configurations, including CPU-only, CPU/GPU, liquid-cooled, and air-cooled options.

The enterprise configuration of Helion integrates agent execution and inference within a single rack, offering flexibility to customers. They can choose between a CPU-only hyperscale configuration or an enterprise configuration with a blend of CPU and GPU blades. Depending on the configuration, Helion can deliver up to 32,768 Zen 6 CPU cores through compute nodes equipped with AMD's Venice Epyc processor, up to 16 Instinct MI350P GPU accelerators, and up to 102 Tbps of networking bandwidth.

Both configurations include dense, balanced, and high-performance CPU blades. For dense deployments, the EPYC 9006 chips rated at 400 W are used, while performance blades feature a 5 GHz version of the chip with 96 cores. Balanced blades utilize 256-core processors rated at 600 W. Additionally, Helion features GPU inference blades with eight Instinct MI350P GPUs in a 5U server, as well as general-purpose blades designed for database and storage servers or technical compute workloads.

AMD's Senior Vice President for Compute and Enterprise AI, Dan McNamara, emphasized that agentic AI encompasses a diverse range of applications and workflows, each with unique compute requirements. To accommodate these varying needs, AMD Helion combines Epyc performance, open software, and adaptable rack designs to enable partners to build systems tailored to customer needs and optimize power usage.

AMD plans to contribute the Helion specifications to the OCP community for further development and adoption, with partner-built Helion configurations slated for release in November 2026. Major technology companies such as Cisco, Dell, HPE, Foxconn, Lenovo, Supermicro, and others have joined AMD in this initiative.

Written by urgent.news from The Register's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

This story

This is one outlet's version. Read the fullest account.

Read the original at theregister.com →

More in AI

Andrej Karpathy says this AI gap is widening

Renowned AI researcher Andrej Karpathy says the gap is widening among the small subset of the population that uses AI.

  • 75% of world's population, 6 billion, minimal LLM contact
  • 1-2 billion casual/infrequent ChatGPT-like users
  • 5,000 people with inside access to frontier AI systems

Run multi-node inference workloads using Ray LLM Deep Learning Containers

Learn how to deploy one large language model across two GPU nodes on Amazon EKS using the multi-node Ray LLM Deep Learning Containers, with separate prefill and decode nodes behind one…

  • Run LLM inference across multiple nodes using Amazon EKS and Ray DLCs
  • Separate prefill and decode nodes to improve performance in complex workloads
  • Ray LLM DLC with vLLM, Ray Serve, NIXL, and EFA simplifies deployment

More from Monday 12 October →