How BMW Group accelerated connected vehicle workloads on AWS Graviton

This article describes how BMW Group successfully migrated its Connected Vehicle platform to AWS Graviton processors, achieving significant cost and performance gains while maintaining service continuity across 24.5 million connected vehicles.

• Adopted Karpenter for dynamic, policy-driven node provisioning to support multi-architecture workloads
• Implemented dual-architecture CI/CD pipelines using native Graviton runners and Docker Buildx with QEMU for flexible builds
• Created dedicated Graviton node pools with taints and tolerations for controlled, opt-in workload migration
• Validated migration on platform services first before enabling team adoption to minimize risk
• Achieved $600,000 annual infrastructure savings and 40% price-performance improvement
• Reduced CPU consumption by 57% on NGINX ingress controller while maintaining performance
• Over 65% of production platform now runs on Graviton, supporting sustainability goals

The migration succeeded through incremental, deliberate choices that reduced risk while keeping adoption optional for individual teams, establishing Graviton as the default compute path.