Amazon Trainium 2 and Inferentia 2 — custom AI chips for AWS.
Amazon's 3rd-generation AI training chip for next-gen EC2 Trn3 instances with 3x Trainium 2 performance.
AWS's 4th-generation custom Arm processor for EC2 with 30% better compute performance over Graviton3.
Amazon's 2nd-generation AI training chip for EC2 Trn2 instances with ultra-high performance.
AWS's 2nd-generation AI inference chip for cost-effective, low-latency model deployment.
Purpose-built cloud AI inference accelerator delivering up to 400 TOPS for data center inference workloads.
