AMD CTO Mark Papermaster discussed heterogeneous computing, chiplets, domain-specific computing, AI-related efficiency and open standards at World Summit AI in Amsterdam on October 7, 2026. The conversation focused on design themes rather than announcing a specific AMD chip. The summit was scheduled for October 7–8.
The themes in Papermaster’s interview
The reported topics ranged from holistic design to the way computing hardware can be tailored for different workloads. Heterogeneous computing combines different kinds of computing resources, while domain-specific computing focuses a design on particular tasks. The discussion also included chiplets, efficiency related to AI, and open standards.
What heterogeneous and domain-specific computing mean
A heterogeneous system brings together different types of computing resources so they can handle different kinds of work. Domain-specific computing takes a more targeted approach: its design is shaped around a particular workload or class of tasks. These concepts help frame the interview’s focus on designing computing systems for varied AI needs.
Chiplets and tailored computing
Chiplets are smaller silicon dies combined within a larger package. A chiplet-based design can use multiple components rather than relying on one large die. Papermaster’s interview covered chiplets and tailored computing as design topics; it did not announce a particular configuration or product.
AMD’s product context
Separately, AMD’s product portfolio includes Instinct data-center accelerators and Ryzen AI client processors. Those product families provide context for AMD’s work in AI computing, while the interview’s reported themes concern design approaches rather than detailed specifications for either family.
