As a result, the system will run on Amazon Bedrock in the technology giant’s data centres, merging AWS Trainium-powered servers, Cerebras CS-3 systems and Elastic Fabric Adapter (EFA) networking.

Additionally, AWS will also offer leading open-source LLMs and Amazon Nova, utilising Cerebras hardware.

According to the company, by combining Amazon’s purpose-built AI chip, which is already used by companies like OpenAI and Anthropic, the partnership will help organisations deal with complex reasoning and agentic coding tasks.

By assigning Trainium to prefill and Cerebras CS-3 to decode, as well as connecting them with low-latency, high-bandwidth EFA networking, each stage is handled by the hardware optimised for it, the company revealed.

Built on the AWS Nitro System, the setup also ensures security and operational reliability.

AWS VP of compute and ML services, David Brown said: “Inference is where AI delivers real value to customers, but speed remains a critical bottleneck for demanding workloads like real-time coding assistance and interactive applications.

“What we’re building with Cerebras solves that: by splitting the inference workload across Trainium and CS-3, and connecting them with Amazon’s Elastic Fabric Adapter, each system does what it’s best at. The result will be an inference that’s an order of magnitude faster and higher performance than what’s available today.”

Cerebras Systems founder and CEO, Andrew Feldman, said: “Partnering with AWS to build a disaggregated inference solution will bring the fastest inference to a global customer base.

“Every enterprise around the world will be able to benefit from blisteringly fast inference within their existing AWS environment.”

RELATED STORIES 

Cerebras expands global footprint with 6 new data centres

Amazon targets robotics team in latest round of job cuts

AWS UAE data centre hit by flying debris amid regional attacks