AMD CTO Mark Papermaster discusses AMD's silicon strategy for AI workloads, drawing on the company's history in heterogeneous CPU/GPU computing.

Ryan interviewed AMD CTO Mark Papermaster on the floor of HumanX (recorded May 8, 2026) to discuss AMD's approach to designing chips for artificial intelligence applications.

The conversation covered how chipmakers are addressing the wide range of AI workloads spanning both training (the phase where an AI learns from data) and inference (the step where an AI produces answers), as well as AMD's long history combining CPUs and GPUs on single chips.

Mark and Ryan explored the paradox that AI agents (software that makes decisions and takes actions autonomously) consume large amounts of computing power while simultaneously helping AMD speed up chip innovation.