More than a decade after AMD created the “Zen” high-performance x86 architecture, the platform that scales from PCs to the data center is entering AI, particularly agentic AI. As agents orchestrate increasingly complex workflows, demand is growing for both high core density and strong per-core performance.

In this September 2026 episode of the AMD “Advanced Insights” series, Mark Papermaster, chief technology officer and executive vice president at AMD, and Mike Clark, senior vice president, Corporate Fellow and chief architect of AMD CPUs, discuss the decisions leading to the start of “Zen” architecture, its evolution and where CPUs are going next.

Their conversation traces how AMD brought CPU teams together around a clean-sheet architecture, set an audacious goal and built the “Zen” platform to evolve across generations. Papermaster and Clark explore the “wheel of performance” behind CPU design – balancing performance, frequency, area and power – as well as the decisions behind technologies such as simultaneous multithreading, wider vector processing, chiplets and dense cores.

They connect the origins of “Zen” architecture with the emerging demands of agentic AI, showing how a long-term architectural foundation can adapt as computing changes. The conversation also explores the advantages of the established x86 software ecosystem, the importance of listening to customers and the engineering practices that allow AMD to refine each generation while using AI to evaluate design possibilities, accelerate verification and help engineers reach better decisions faster.

More: View all episodes of “Advanced Insights” 

Follow: Mark Papermaster on LinkedIn

TopicsData CenterArtificial IntelligenceSoftware #zen-architecture #agentic-ai
Share / cite