AMD to partner with high-speed AI inference systems company Cerebras to ship ultra-low-latency AI systems in the second half of 2026

AMD and Cerebras announced a split architecture for AI inference, assigning low-latency and high-throughput roles to different systems.

Cerebras will deploy AMD’s rack-scale Helios platform in its own data centers and pair it with Wafer-Scale Engine hardware.

The joint setup aims to deliver faster response times and better efficiency for real-time and agentic AI applications.

The companies plan to bring the combined inference solution to market in the second half of 2026.