AMD and Cerebras partner to deliver disaggregated AI inference solution for data centers

Type: Partnership · Technology: Data Center · Country: United States · Announced: 2026-07-23

AMD and Cerebras Systems announced a partnership on 2026-07-23 to jointly develop an AI inference solution that combines AMD Helios rack‑scale hardware with the Cerebras Wafer‑Scale Engine. The joint solution will integrate AMD Helios into Cerebras’ data centers, leveraging the Helios high‑throughput engine and the Wafer‑Scale Engine’s ultra‑low‑latency token generation. Cerebras plans to make the combined offering available through its Cerebras Cloud platform in the second half of 2026, targeting latency‑sensitive AI workloads such as real‑time agents and autonomous coding assistants. By pairing AMD’s scalable GPU‑based throughput with Cerebras’ memory‑bandwidth‑intensive engine, the partnership aims to deliver up to five times higher tokens per second per watt compared with existing solutions. The collaboration expands AMD’s Helios presence beyond traditional server markets and positions Cerebras as a provider of disaggregated AI inference infrastructure for enterprise customers. No financial terms were disclosed, and the announcement was made jointly from San Francisco and Sunnyvale, California.

Counterparties

Source article

📈 The Sherpa Signal — one notable infrastructure deal a day.