Nvidia and Palantir put AI agents inside Nvidia supply-chain operations

Nvidia and Palantir announced a new supply-chain AI collaboration built around Palantir Foundry, Palantir AIP and Nvidia Nemotron open models. The first deployment is inside Nvidia’s own operations, where the companies say the system is being used to turn supply-chain data and expert judgment into faster allocation decisions.

The target is one of the hardest operational problems in AI infrastructure: getting complex rack-scale systems from components to usable compute. Nvidia says a Vera Rubin rack contains 1.3 million parts, and its supply chain spans thousands of suppliers and manufacturing partners. The new stack is meant to surface bottlenecks, run scenario planning with Nvidia cuOpt, and post-train Nemotron models on governed operational history.

For enterprise customers, the pitch is that the same reference architecture can run in cloud or on-premises environments, with companies keeping control of proprietary data while customizing open models for their own supply chains.

The useful signal here is not another generic AI partnership. Nvidia is applying AI agents and specialized open models to its own AI-factory logistics, then packaging that pattern for other manufacturers, governments and infrastructure-heavy companies.