AMD-Google Chip Collaboration Could Reshape AI Landscape

Google and AMD are reportedly working on a 10th-generation Tensor Processing Unit (TPU) chip, according to market chatter from SemiAnalysis. The collaboration would mark AMD's first major involvement in a custom AI accelerator project.

Wedbush analyst Matt Bryson called the potential partnership significant, citing its implications for Google's long-term silicon strategy and AI infrastructure margins.

The 10th-gen TPU is expected to target reinforcement learning and agentic AI workloads, which demand tighter CPU-accelerator integration than current chip designs provide. AMD's MI300A already integrates x86 cores and AI accelerator chiplets together, providing a working prototype of the hybrid design needed for this project.

The partnership may indicate Google's growing reliance on custom silicon development to reduce its dependence on third-party GPU suppliers and widen its AI infrastructure margins.