Hybrid-Confidential-Cloud - Democratize Secure AI With GPUs and Confidential Containers

Explore a 30-minute conference talk that demonstrates how to implement secure AI workloads using GPUs and confidential containers across hybrid cloud environments. Learn how CNCF Confidential Containers enable infrastructure-agnostic, lift-and-shift deployments of GPU workloads while maintaining confidentiality without requiring code modifications. Discover the architecture that layers trust-oriented elements over Kubernetes for compute, networking, storage, and control plane operations, utilizing hardware-backed confidential VMs to ensure runtime integrity for GPU workloads. Understand how identity-based overlay networks and confidential storage layers protect valuable data against replay attacks, while a confidential control plane provides multi-tenancy capabilities, key lifecycle management, and trust boundary maintenance. Gain insights into building trusted, portable AI infrastructure that scales across any IaaS environment—whether on-premises, private cloud, or cloud service provider—enabling true hybrid secure AI deployment at enterprise scale.