How Public AI delivers sovereign LLM inference on AWS and Intel (opens in new tab)

Open-weight large language models are being released by research institutions worldwide, but turning published weights into production inference services remains a challenge—especially under strict data residency requirements. This post shows how Public AI built a scalable inference platform on Amazon EKS and Intel-powered Amazon EC2 instances to serve Switzerland's Apertus model family, and why this architecture provides a repeatable blueprint for sovereign LLM initiatives.