Global Infrastructure SherpaAWS partners with Cerebras Systems for AI inference solution deployment on Amazon Bedrock in AWS data centers
Amazon Web Services (AWS) has partnered with Cerebras Systems to deliver an AI inference solution supporting generative AI applications and LLM workloads. The solution will combine AWS Trainium-powered servers with Cerebras' wafer-scale CS-3 systems and Elastic Fabric Adapter (EFA) networking, deployed on Amazon Bedrock in AWS data centers. This partnership aims to improve AI inference speed and performance by disaggregating workloads across Trainium and CS-3 hardware.
Counterparts (2)
Subscribe to access full counterpart details, deal analysis, and timeline.
Subscribe from €20/mo →Deal Analysis
- AWS Trainium servers paired with Cerebras wafer-scale CS-3 systems.
- Deployment on Amazon Bedrock within AWS data centers.
- Partnership announced for March 13, 2026.
- Aims to disaggregate generative AI workloads for performance gains.
Source Intelligence
The solution will combine AWS Trainium-powered servers with Cerebras' wafer-scale CS-3 systems and Elastic Fabric Adapter (EFA) networking. AWS will also offer open-source LLMs and Amazon Nova, its own foundation models, using Cerebras hardware “later this year.” The combined Trainium/CS-3 solution will enable “inference disaggregation,” splitting AI inference into a compute intensive prompt processing (‘prefill’) stage and a memory bandwidth-intensive output generation (‘decode’) stage. Trainium chips will be optimized for prefill, while Cerebras CS-3 hardware will be optimized for decode.
"Amazon Web Services (AWS) has partnered with Cerebras Systems"
"Amazon Web Services (AWS) has partnered with Cerebras Systems"
Timeline
Get the full picture — timeline, source intelligence, and counterpart analysis.
Subscribe from €20/mo →