NVIDIA Confidential Computing expands for secure model builders

NVIDIA has just expanded its tools that enable organizations to securely deploy agents. The NVIDIA OpenShell Reference Architecture, announced today, is a crucial part of NVIDIA's growing suite of technologies that help organizations securely run agents at scale. NVIDIA Confidential Computing serves as the infrastructure layer that protects the models and data used by these agents during their operation.

With the introduction of Confidential Bare Metal, NVIDIA further expands its infrastructure layer. This new implementation option allows frontier model builders to run sensitive workloads on specific third-party infrastructures while maintaining control over their models and data. New platform security measures ensure that data is protected during transport, the integrity of the platform is measured, and access is controlled at the hardware level.

NVIDIA Confidential Computing now offers a converged security architecture with three implementation options: Confidential Bare Metal for bare-metal infrastructures, Confidential Containers for Kubernetes environments, and Confidential Virtual Machines for virtualized environments. This variety allows organizations to choose the right model option based on their trust boundaries and operational requirements.

The expansion of Confidential Computing gives goods and services with high yield the ability to leverage global capacity. Confidential Bare Metal helps providers serve valuable workloads that require strong privacy protection. Among others, Cohere and Google Cloud have announced plans to integrate this technology into their products, enhancing the security of customer data during execution and data processing.

Read the full article from Nvidia.