NVIDIA Confidential Computing Expands, Enabling Frontier Model Builders to Securely Scale Across Global Infrastructure
nvidia ·
NVIDIA has expanded its Confidential Computing capabilities with the introduction of Confidential Bare Metal, allowing frontier model builders to run sensitive workloads on dedicated third-party infrastructure while maintaining control over their models and data. This new deployment option complements existing Confidential Containers and Virtual Machines, offering a converged security architecture that protects data in transit and enforces hardware-level access restrictions. The expansion includes support for NVIDIA Grace Blackwell and Rubin systems, as well as integration with the NVIDIA DSX OS to serve as a reference architecture for cloud providers building AI factories. Key partners such as Cohere and Google Cloud are already integrating these technologies to enhance privacy and protect intellectual property during training and inference. Cohere plans to use trusted execution environments to extend hardware-rooted security directly to the GPU within its Model Vault service. Meanwhile, Google Cloud is extending support to A4X Max instances powered by GB300 NVL72 systems to provide non-virtualized compute with cryptographic protection. A broad ecosystem of global systems partners, including Cisco, Dell Technologies, and HPE, will support these capabilities through NVIDIA-Certified Systems. These advancements aim to enable organizations to securely scale AI operations across global, sovereign, and enterprise infrastructures.
nvidia