What changed
NVIDIA announced the expansion of its Confidential Computing platform with the introduction of Confidential Bare Metal, a new deployment option that allows model builders to run sensitive workloads on dedicated third-party infrastructure while maintaining control over their models and data. This addition joins existing options like Confidential Containers and Confidential Virtual Machines, offering organizations more flexibility in securing their AI operations.
What this means in practice
The new deployment option enables frontier model builders to use bare-metal instances to operate their bespoke software stacks, including host and platform infrastructure software, on NVIDIA accelerated computing infrastructure across third-party cloud, neocloud, and enterprise environments. This allows for greater control and customization, particularly for high-value workloads requiring strong privacy protections.
Limitations and availability
NVIDIA Confidential Computing now supports NVIDIA Grace Blackwell and Rubin systems, with integration into NVIDIA DSX OS as a reference architecture for cloud providers. Partners such as Google Cloud, CoreWeave, and Microsoft Azure Infrastructure are supporting these capabilities, with global systems partners like Cisco and Dell Technologies also offering NVIDIA-Certified Systems. The supplied announcement does not specify exact availability timelines for all deployment options.
