Palantir brings Nemotron models to secure government deployments

Palantir has introduced a new intelligent engine that uses NVIDIA Nemotron open models to support U.S. government agencies in secure, isolated environments. The system is designed to run on NVIDIA accelerated computing in air-gapped setups, giving agencies the ability to customize models while retaining control over proprietary data, model weights and deployment infrastructure.

The company plans to use Nemotron models to build frontier-quality custom models for government operations spanning commerce, energy, healthcare, agriculture, education and transportation. With about 3 million civilian employees, the U.S. government is positioned as one of the world’s largest enterprise-scale operators, with AI expected to help improve productivity and support services such as food safety and highway infrastructure management.

Palantir’s Sovereign AI Operating System, built on AIP, Ontology, Foundry and Apollo, provides the operational and data authorization layer for deployments in sensitive environments. Agencies can train models on their own data, keep ownership of the resulting models and use production feedback to improve performance while maintaining auditability.

NVIDIA and Palantir frame open models as a way to improve transparency, customization and cost efficiency. Open models can be reviewed for vulnerabilities and biases, adapted for regulated use cases and deployed where closed systems may raise data security or privacy concerns.