Built on the Zen 6 architecture, the new server CPU portfolio targets AI agent workloads, cloud services and high-performance computing, with systems now shipping from leading server manufacturers.
AMD has announced the general availability of its 6th Gen EPYC server processors, based on the new Zen 6 architecture, as it looks to extend its data centre CPU position amid rising demand for agentic AI and enterprise computing.
The SP7 processors are now shipping in systems from leading server original equipment manufacturers and original design manufacturers. Cloud instances powered by the new CPUs are also expected across major providers, with Microsoft Azure and Oracle Cloud Infrastructure (OCI) among the first to outline deployments.
AMD is positioning the portfolio for a data centre environment in which CPUs increasingly support more than conventional application processing. Agentic AI workflows require compute resources to coordinate agents, retrieve information, execute code, interact with databases and manage services around AI models.
The new EPYC range is designed to address these varied requirements through workload-optimised processors balancing core density, frequency, memory bandwidth and power efficiency.
“AI is significantly expanding the role of the CPU across every layer of the data center.” — Dan McNamara, Senior Vice President and General Manager, Compute and Enterprise AI, AMD
AMD claims its EPYC 9996 is the world’s best general-purpose server processor. According to the company, an EPYC 9996-based system delivers up to 3.7 times the performance of an Nvidia Vera CPU-based system, and up to 2.2 times the performance and twice the performance per dollar of an Intel Xeon 6980P system. These are AMD’s stated comparisons and are based on its specified test configurations.
For agentic AI infrastructure, AMD is targeting three CPU roles: general-purpose services, GPU host nodes and agent sandboxes. The company says its high-frequency processors, operating at up to 5.3 GHz, can deliver up to 1.8 times the large language model output throughput of the previous EPYC generation. Its high-density processors are claimed to offer an agents-per-CPU-watt advantage of approximately 3.3 times over Nvidia Vera and 2.5 times over Intel Xeon 6980P in agent sandbox workloads.
The launch also extends to rack-scale infrastructure through AMD Helion configurations, including liquid-cooled and air-cooled options. Some rack configurations can reach up to 73,728 threads, according to AMD, providing customers with different approaches to scaling agent workloads.
Cloud adoption is expected to broaden the platform’s reach. Microsoft Azure is bringing the processors to services including HDv2 for AI data processing, agentic AI and HPC, and HXv2 for electronic design automation and technical computing. OCI Compute E7 instances will also use 6th Gen EPYC CPUs for enterprise applications and CPU services surrounding AI models. Meta has been validating EPYC-based platforms ahead of planned deployments at scale.
The launch marks another step in AMD’s effort to position the CPU as a critical component of AI infrastructure, alongside accelerators and GPUs. As enterprises move from AI experimentation towards production deployments, the ability to support orchestration, data movement and concurrent workloads efficiently will become an increasingly important part of data centre economics.