AMD Bets on System-Level AI Over Faster Chips

Agentic AI workloads are forcing AMD to rethink infrastructure from the ground up, moving beyond chip performance toward system-level optimization across heterogeneous computing architectures. CTO Mark Papermaster says complex end-to-end AI processes require different computing engines working together at massive scale, spanning data centers, edge deployments and AI-enabled PCs. AMD has expanded its portfolio through acquisitions of Xilinx, Pensando and ZT Systems, evolving from chip designer to rack-level system optimizer. Its unified ROCm software stack runs identically across all tiers, allowing enterprises to route workloads to the most cost-efficient compute without replacing existing x86 infrastructure.