Amazon has expanded the availability of EC2 G6e instances powered by NVIDIA L40S Tensor Core GPUs to additional regions in Asia Pacific and Europe.

• Now available in Asia Pacific (Tokyo) and Europe (Frankfurt, Spain)
• Delivers up to 2.5x better performance compared to G5 instances
• Up to 20% lower inference costs than P4d instances
• Supports large language models up to 13B parameters
• Features up to 8 NVIDIA L40S GPUs with 384 GB total GPU memory
• Supports up to 192 vCPUs and 400 Gbps network bandwidth

The G6e instances are ideal for machine learning, AI inference, and spatial computing workloads, with support for various AWS services and deployment options.