AMD has started shipping its sixth-generation EPYC server processors, giving businesses new chips for running AI agents, cloud services, and everyday business applications. Built on the company’s new Zen 6 design, the processors are available in servers from leading vendors, with cloud services from Microsoft Azure and Oracle Cloud Infrastructure (OCI) coming soon.
The launch addresses the computing work behind AI tools. An AI agent may need to find information, run code, and connect to company databases to complete a task. AMD designed different processors in the lineup for these jobs, allowing customers to choose chips based on the work their servers handle.
“AI is significantly expanding the role of the CPU across every layer of the data center,” said Dan McNamara, senior vice president and general manager, Compute and Enterprise AI, AMD. “With 6th Gen AMD EPYC, customers can standardize on the proven x86 foundation while choosing a workload-optimized solution to run more agents, keep accelerated infrastructure fed to power the business-critical applications they depend on.”
The sixth-generation EPYC SP7 processors became generally available on Nov. 10. AMD, which “drives innovation in high-performance and AI computing to solve the world’s most important challenges,” said its technology serves cloud and AI infrastructure, embedded systems, AI PCs, and gaming.
For businesses, AMD said the new processors could help each server support more applications, transactions, and users. This would give companies more computing capacity within existing limits on electricity and data center space.
AMD said a system using its EPYC 9996 processor delivers up to 2.2 times the performance of an Intel Xeon 6980P system and up to 3.7 times that of an Nvidia Vera CPU-based system. These are AMD’s performance claims.
The lineup also includes processors running at speeds of up to 5.3 GHz for servers that support graphics processing units, or GPUs. AMD said these processors can help systems produce up to 1.8 times as much large language model output as systems using the previous EPYC generation.
Other processors are designed to host many AI agents in separate computing spaces, called sandboxes, where they can run code and perform tasks. AMD said these chips can support more agents for each watt of CPU power than the competing processors it cited.
AMD is also offering Helion server racks in three configurations. Helion Liquid uses liquid cooling for large AI deployments. Helion Air adds AI agent capacity to existing air-cooled data centers, while Helion Air Inference combines CPUs and GPUs in one rack to run agents and AI models.
Microsoft Azure’s HDv2 service is already in preview for AI data processing, AI agents, and high-performance computing. Azure also plans services for engineering work and large-scale AI use.
OCI’s upcoming Compute E7 cloud services will use the new EPYC processors for business applications and computing tasks that support AI models. Meta is working with AMD as a lead customer and has been testing platforms ahead of wider deployment. AMD said customers can choose different combinations of processors, cooling, and connectivity as their computing needs grow.
Promotional image provided by AMD