AMD has announced the general availability of its 6th Generation AMD EPYC SP7 server processors, built on the new ‘Zen 6’ architecture. These processors are designed to address the escalating demands of data centers driven by the rapid advancements in artificial intelligence (AI) and enterprise computing.
Systems powered by the new processors are already being shipped by leading server hardware manufacturers and ODM partners. Cloud services from Microsoft Azure and Oracle Cloud Infrastructure (OCI) will also soon offer capabilities based on these processors.
“AI is significantly expanding the role of CPUs at every level of the data center,” said Dan McNamara, Senior Vice President and General Manager of the Compute and Enterprise AI division at AMD. “With 6th Gen AMD EPYC, customers can standardize on a proven x86 architecture while selecting task-optimized solutions to run more agents, accelerate infrastructure, and power critical business applications.”
World’s Leading Server Processor Portfolio for the Modern Data Center
The 6th Generation AMD EPYC aims to deliver an optimized suite of solutions that provide leading performance across diverse applications. It enhances infrastructure cost-efficiency while maintaining the flexibility of the broad x86 ecosystem. The new processors are engineered to support agent-based AI, a paradigm where a single workflow can involve multiple agents, model calls, data retrieval, code execution, and interactions with databases and applications.
Supporting Enterprise and Cloud Services
The 6th Generation AMD EPYC 9996 processor is positioned as the world’s leading general-purpose processor. Systems based on the AMD EPYC 9996 reportedly achieve up to 3.7x higher performance compared to systems using Nvidia Vera and up to 2.2x higher performance and 2x better performance-per-dollar compared to Intel Xeon 6980P. This enables greater support for applications, transactions, and users on each server.
Leadership Across All Segments of Agent AI
Agent-based AI demands diverse CPU roles, including those for general services, GPU host nodes, and agent sandboxes. The 6th Generation AMD EPYC offers processors optimized for each of these functions. High-frequency processors for host nodes, reaching up to 5.3 GHz, deliver up to 1.8x higher LLM throughput compared to previous EPYC generations. High-density AMD EPYC processors can support approximately 3.3x more agents per watt compared to Nvidia Vera and around 2.5x more agents per watt compared to Intel Xeon 6980P for agent sandbox deployments. Furthermore, servers powered by AMD EPYC 9996 can accommodate approximately 18% more agents per rack within the same power budget (35kW) when compared to systems using standard power management configurations.
AMD is offering three deployment options for agent AI infrastructure: AMD Helion Liquid, AMD Helion Air, and AMD Helion Air Inference. Developed in collaboration with the server solutions ecosystem, these solutions can achieve up to 73,728 threads per rack, maximizing agent density.
Key partners already leveraging the 6th Generation AMD EPYC include Meta, Microsoft Azure (offering Azure HDv2, Azure HXv2, and Azure ND MI455X v7), and Oracle Cloud Infrastructure (OCI) (with new OCI Compute E7 instances).
AMD continues to extend its leadership in server processors by adapting to the new era of agent AI, offering increased compute power, enhanced economics, and broader choice from the server to the rack level.
