Nvidia expands Confidential Computing with Confidential Bare Metal for model builders
Nvidia announced the Confidential Bare Metal option for its Confidential Computing platform, letting frontier model builders run sensitive workloads on dedicated third‑party bare‑metal servers while keeping full control of models and data. The new deployment adds hardware‑rooted isolation, cryptographic attestation, and platform‑integrity checks across Nvidia’s Grace Blackwell, Rubin and other…
Key points
- Nvidia introduced Confidential Bare Metal, a new deployment option for its Confidential Computing platform.
- Confidential Bare Metal enables model builders to run workloads on dedicated third‑party bare‑metal servers with hardware‑rooted security.
- Cohere and Google Cloud announced integration of Confidential Bare Metal into Cohere Model Vault and Google Cloud A4X Max instances.
Cohere said it will embed Confidential Bare Metal into its Model Vault, extending hardware‑rooted security to the GPU. Google Cloud also announced support for Confidential Bare Metal on its A4X Max instances powered by Nvidia GB300 NVL72 systems, promising end‑to‑end cryptographic protection for training data and model IP. A broad ecosystem of cloud and hardware partners, including CoreWeave, Firmus, GMI Cloud, Lambda, Microsoft Azure, and others, now supports the expanded Confidential Computing modes.
NVIDIA Confidential Computing Expands, Enabling Frontier Model Builders to Securely Scale Across Global Infrastructure
NVIDIA Blog · 21 October 2026
Loading the full article…
This text was published by NVIDIA Blog and written by John Fanelli. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Hardware & Compute
All →- Intel and AMD collaborate on oneDNN optimizations for PyTorch · 2 src
- Nvidia says 1GW AI factory costs $50 billion, unveils DSX POD · 4 src
- Nokia and Nvidia target 50% spectrum efficiency by 2027 · 3 src
- NVIDIA pays $1.1B to KYEC for US test capacity · 1 src
- HPE plans first Moon supercomputer benchmarks with Spaceborne Computer-4 in late 2026 · 1 src
Comments
via GitHub Discussions