Palantir Technologies has unveiled a new collaboration with NVIDIA to build an artificial intelligence engine that allows NVIDIA’s Nemotron open AI models to be deployed in secure, sovereign environments. The initiative is designed for US government agencies and organisations responsible for critical infrastructure, enabling them to use advanced AI while keeping complete control over their data and systems.
The joint solution combines NVIDIA’s AI infrastructure, ecosystem and Nemotron open models with Palantir’s Artificial Intelligence Platform (AIP), Foundry, Ontology and Apollo. Together, the platform enables organisations to develop, deploy and refine AI models without compromising ownership of their data, intellectual property or AI assets.
According to the companies, the platform is built with enterprise-grade security features, including strict data access controls, customer-specific isolation, secure perimeter protection, data portability, audit logging and support for data deletion requests. These capabilities are intended to help organisations meet security and compliance requirements while operating sensitive AI workloads.
The platform also allows customers to customise and continuously improve AI models using operational telemetry, trace data and post-training feedback generated within their own environments. This enables organisations to create AI systems tailored to their specific operational needs instead of relying solely on pre-trained models.
For government users, the offering supports the deployment of both base and customised Nemotron models in classified, air-gapped and other highly secure environments. It also provides tools for context engineering to optimise prompts and workflows, model engineering to fine-tune model weights with proprietary data, and enterprise deployment through NVIDIA AI Enterprise software and NVIDIA NIM microservices.
Palantir Co-founder and CEO Alex Karp said the partnership will help US government agencies take advantage of large language models while reducing security risks associated with closed AI models. He added that several government organisations and operators supporting critical US infrastructure are already using these technologies, and the new platform will enable wider adoption.
NVIDIA founder and CEO Jensen Huang highlighted the growing importance of open-source AI for national security, public safety and US technological leadership. He said the collaboration provides government agencies with a secure, customisable and fully controlled AI foundation for building mission-critical applications.
The announcement expands on the companies’ earlier Sovereign AI Operating System Reference Architecture, a framework designed to help governments deploy, operate and manage AI systems within infrastructure that remains under their own control.
Also Read: NVIDIA Invests $150 Million in AI Inference Firm Baseten
