Palantir Technologies Showcases Sovereign AI Wins With NVIDIA, Cisco and FAA

Palantir highlighted sovereign AI deployments at its AIPCon event, emphasizing enterprise-controlled data, infrastructure and fine-tuned models that preserve security and operational control. NVIDIA is using Palantir’s Ontology, AIP and cuOpt tools to optimize its complex supply chain, while Cisco plans to offer Palantir’s cybersecurity model through its AI Factory with NVIDIA. Government and commercial customers reported measurable benefits, including the FAA’s national aviation-safety risk system, L3Harris’s rapid supply-chain analysis and Hexion’s stated $300 million EBITDA improvement from its broader data transformation.

Palantir Technologies NASDAQ: PLTR used its AIPCon event to highlight customer deployments of its Foundry, Ontology and Artificial Intelligence Platform products across supply chains, aviation safety, defense manufacturing, insurance, media, pharmaceuticals and industrial operations. Speakers repeatedly emphasized “sovereign AI,” or the use of enterprise-controlled data, infrastructure and fine-tuned models for specific operational tasks. Co-founder and CEO Alex Karp said large language models require an application layer and operational context to deliver precise, useful results. He argued that organizations can retain control of their proprietary data and business advantages by fine-tuning open-weight or frontier models within their own environments. Karp said Palantir is using such approaches in classified settings, where organizations can operate models at lower cost while maintaining control over sensitive information.

Alex Neefus, senior director of solution architecture at NVIDIA, described a joint supply-chain application called Command Center. The platform is intended to help NVIDIA manage component allocation across a global manufacturing network supporting systems such as the Grace Blackwell NVL72. Neefus said NVIDIA’s supply chain involves millions of parts from thousands of suppliers and that manufacturers cannot begin production until all critical components are available. The company tracks “Time of Ownership,” a metric designed to reduce the time components sit idle while awaiting other materials, while increasing throughput toward the production of AI tokens.

According to Neefus, Palantir’s Ontology connects data on materials, manufacturers, sites, capacity, allocations and production commitments. NVIDIA is combining that data layer with its cuOpt decision-optimization library and Palantir’s AIP tools to evaluate supply-chain scenarios, such as reduced memory availability or the addition of a factory. NVIDIA said its supply-chain planners remain central to allocation decisions, with the system designed to capture their expertise and create evolving benchmarks for model evaluation. Neefus said NVIDIA fine-tuned its Nemotron 3.5 Lightning model for the workflow and found that it outperformed the company’s larger Nemotron 3 Ultra model for the specialized task. He said the smaller model can be retrained on a single GPU in a few hours. The deployment runs on premises at NVIDIA and was jointly published as the “Palantir Sovereign AI Operating System,” Neefus said.

Cisco President and Chief Product Officer Jeetu Patel announced an expanded collaboration with Palantir. Under the arrangement, Palantir’s cybersecurity Ontology model will be offered through Cisco’s AI Factory with NVIDIA.