The power of confidential containers on Red Hat OpenShift with NVIDIA GPUs

The power of confidential containers on Red Hat OpenShift with NVIDIA GPUs Confidential containers on bare metal Unlocking enterprise AI with NVIDIA's Blackwell GPUs and confidential containers Confidential containers: Securing the cloud-native AI stack End-to-end attestation: Building a chain of trust Red Hat AI Inference Server Demo: AI inferencing with CoCo bare metal and confidential GPUs Final thoughts The adaptable enterprise: Why AI readiness is disruption readiness About the authors Ariel Adam Pradipta Banerjee Jens Freimann Emanuele Giuseppe Esposito More like this Blog post Blog post Original podcast Original podcast Keep exploring Browse by channel Automation Artificial intelligence Open hybrid cloud Security Edge computing Infrastructure Applications Virtualization Share Artificial intelligence (AI) is rapidly moving from a theoretical concept to a central engine of enterprise value, transforming sectors from healthcare to finance. The capacity of AI to analyze, predict, and automate makes it an indispensable asset for modern innovation. Yet, this widespread adoption introduces a significant security imperative: as AI workloads scale, so does the risk of unauthorized access to proprietary AI models and the sensitive data they handle. These AI models are significant assets for organizations, representing substantial investments in research, training, and inferencing. Protecting them requires a robust security strategy that goes beyond traditional measures for data at rest (in storage) and in transit (over networks). The most vulnerable state is data in use, the moment it's actively being processed in memory. This is where confidential computing emerges as a game-changer. By providing a trusted execution environment (TEE), it strengthens application security through isolation, encryption, and attestation, protecting data and code as they are being used. This approach enables a comprehensive defense-in-depth strategy, which is crucial for regulated industries handling sensitive information. By integrating these security features with a scalable, high-performance AI and machine learning (ML) ecosystem, companies can leverage AI with confidence, without sacrificing security for speed. In this blog, we will explore the usage of Red Hat OpenShift confidential containers along with NVIDIA confidential GPUs to protect AI workloads, focusing on Red Hat AI Inference Server as the workload. Confidential containers (CoCo) are a cloud-native implementation of confidential computing that brings the security of a TEE to a standard containerized workload (such as Intel TDX or AMD SEV-SNP).