Palantir Technologies (NASDAQ: PLTR) announced a strategic initiative with Nvidia to deploy Nvidia's Nemotron open AI models within sovereign environments, targeting U.S. government agencies and critical infrastructure operators.

According to a press release, the offering combines Nvidia's AI platform — including compute resources and open models — with Palantir's products, including AIP, Ontology, Foundry, and Apollo. The joint system is designed to allow government customers to train, customize, and deploy AI models while retaining control over their data, intellectual property, and AI systems.

Key features listed include explicit data authorization, secure perimeter enforcement, customer-specific data isolation, data portability, right to erasure, and full auditability. The offering also includes deployment in classified and air-gapped environments, context and prompt optimization, and the ability for agencies to modify model weights using proprietary data and mission outcomes.

Palantir CEO Alex Karp said the combination would allow the U.S. government to use large language models while reducing security risks associated with proprietary data entering closed model systems. Nvidia CEO Jensen Huang said the initiative demonstrates how open models can support mission-critical AI systems for national security purposes.

The initiative builds on a previously announced Sovereign AI Operating System Reference Architecture between the two companies. Enterprise-grade deployment will include Nvidia AI Enterprise software and Nvidia NIM microservices.