Cerebras launches disaggregated inference with AMD, delivering 5x more tokens per megawatt
AccBalanced · x · 2026-11-22
Cerebras announced disaggregated inference with AMD, billed as the first production-scale deployment combining GPU and non-GPU systems: AMD Instinct GPUs handle prefill while Cerebras CS-3 systems handle decode. Production data shows 5x more tokens per megawatt than CS-3 alone, at Cerebras speeds.
- AWS, Fireworks AI, Gimlet Labs and General Compute already have access
- Broad rollout planned across hundreds of megawatts by end of 2027
- Pairs with the newly announced CS-4, scaled manufacturing, and planned 700MW of data center capacity
More from Infra
- Fugaku supercomputer breach: 430 SSH keys stolen, 233 had no passphrase — kingslyj · 2026-11-22
- OMEN: free open-source tool tests 714 GPU setups to pick the best card for local AI models — Srijith_Legend · 2026-11-22
- GTA VI is a megakernel: devs marvel at its GPU-driven software rendering pipeline — charles_irl · 2026-11-22
- Nvidia's circular GPU financing questioned: funded purchases, guaranteed debt, insurer backstop — MatthewChang · 2026-11-22
- Project Maya v1.1.73 brings Inkling 2 Mini to 8GB GPUs, 144 tok/s on a 4090 — teortaxesTex · 2026-11-22
- Ex-Google TPU engineer: cutting frontier research will ultimately sacrifice the hardware business — jasondeanlee · 2026-11-22