/A%20close-up%20of%20an%20AMD%20office%20by%20gehapromo%20via%20Adobe%20Stock.jpeg)
Cerebras Systems (CBRS) and Advanced Micro Devices (AMD) announced a technical partnership on Thursday at AMD's Advancing AI 2026 event in San Francisco, unveiling a disaggregated AI inference architecture that pairs AMD’s Helios Rackscale solutions with CBRS’s Wafer-Scale Engine (WSE).
The joint solution divides the artificial intelligence inference workflow into two optimized stages: AMD Helios processes initial prompts and large context windows at high throughput, while CBRS’s specialized chip architecture handles the memory-bandwidth-intensive token generation with ultra-low latency.
