AI infrastructure is expanding beyond the chips that generate answers. More of the computing around AI is being built into the same systems.
AMD's 6th Gen EPYC server processors are now shipping in systems from server makers. AMD is also outlining Helion rack setups for AI work. Microsoft Azure and Oracle Cloud Infrastructure plan to offer cloud instances using the new processors.
Why does that matter? AI agents, software that can carry out multi-step tasks, need computing capacity beyond the chips that generate each response. New server processors give data-center operators another way to provide it.
Picture one rack as a small workshop with two kinds of tools. General-purpose processors handle some of the agent's work, while nearby AI chips generate responses. AMD's Helion Air Inference option brings those roles together in one rack.
One Helion option is designed for existing air-cooled data centers, giving operators a setup that fits infrastructure they already use. The processors are shipping now, while cloud access is planned.
Scaling AI isn't only about adding faster AI chips. The computing around them matters too. The test now is whether these systems make AI agents easier to run at scale, and when cloud customers can actually access them.
