Amazon Web Services (AWS) partnered with AI chip startup Cerebras Systems to enhance its cloud computing infrastructure.
AWS will deploy Cerebras CS-3 Wafer-Scale Engine chips within its data centers. Customers can access this hardware through the Amazon Bedrock platform for high-speed AI inference tasks.
The collaboration utilizes a disaggregated inference architecture to combine Cerebras hardware with AWS Trainium chips. This system assigns initial processing to Trainium and output generation to Cerebras to increase performance speeds.