We are excited to launch SAFIR (Secure AI for Intelligent Resilience and Confidentiality in the Cloud), a new project focused on building the next generation of secure and resilient cloud-edge systems.

AI is becoming a core part of cloud and edge systems, but ensuring that these systems are secure, resilient, and trustworthy remains a major challenge.

SAFIR (Secure AI for Intelligent Resilience and Confidentiality in the Cloud) addresses this by developing next-generation cloud-edge infrastructures that combine AI, confidential computing, and secure high-performance communication. The goal is to enable AI systems that can operate on sensitive data while maintaining strong guarantees of confidentiality and reliability.

The project focuses on four key areas: protecting AI workloads through confidential execution, securing high-speed data flows across distributed systems, ensuring that AI-driven orchestration is verifiable and safe, and making infrastructures resilient to real-world events by anticipating demand surges from societal signals such as news and crises .

SAFIR brings together a strong collaboration between academia and industry, including KTH, CanaryBit, Ericsson, Nvidia, Red Hat, RISE, and Saab. Running from 2025 to 2028, the project aims to build secure-by-design AI infrastructure for critical domains such as telecom, defense, and cloud services