Under the architecture, agencies can train Nemotron models on internal datasets while keeping both the underlying data and resulting model weights within their own environments. Palantir said its Sovereign AI Operating System provides the authorization, isolation and audit capabilities required for deployments in secure settings.
As agencies continue using the models in production, they can retrain and improve them using additional operational data and feedback collected within their own environments. NVIDIA describes this process as a continuous data flywheel that improves model performance while maintaining customer control over data, model weights and auditability.
The companies said the approach is intended for government organizations operating across areas such as commerce, energy, healthcare, agriculture, education and transportation. NVIDIA noted that the U.S. government employs roughly 3 million civilian workers, making it comparable in scale to one of the world’s largest enterprises.
The announcement also reinforces NVIDIA’s broader strategy around open AI models. The company argues that open models give organizations greater transparency, allowing independent researchers to examine models for vulnerabilities, biases and unintended behavior. That visibility, NVIDIA says, can help improve model safety through ongoing review and refinement.
NVIDIA also said open models provide greater flexibility for organizations that need to tailor AI systems for specialized workloads or deploy them in regulated environments. Because customers can modify and fine-tune the models themselves, the company said they can better address security, privacy and operational requirements while maintaining control over deployment.
The company added that cost is another factor driving adoption, citing that about two-thirds of companies already use open models and report cost efficiency as an important consideration when scaling AI deployments.
“Today, open models are making frontier-level AI broadly accessible, with control over customization and trust through transparency,” NVIDIA said in the announcement.
The company also framed the launch within a longer history of open source software in the United States, pointing to milestones including DARPA’s early networking work, UNIX, the C programming language, the Linux kernel, GitHub and Docker as examples of open technologies that helped shape modern computing.
By combining Nemotron open models with Palantir’s infrastructure platform, NVIDIA said organizations can deploy frontier-quality AI while keeping data, customized models and operational knowledge inside their own secure environments.
This analysis is based on reporting from NVIDIA.
Image courtesy of NVIDIA.
This article was generated with AI assistance and reviewed for accuracy and quality.
