Palantir named L3Harris, McCarthy Building Companies, Tampa General Hospital, Ondas and Regna AI, and Nvidia as "a new wave of flagship customers" for its sovereign AI work. The list came in an October 21 release.
The update followed a September 10 announcement. On that day, Palantir and Nvidia announced a collaboration to bring sovereign AI to supply chains, "starting with NVIDIA's own operations."
What the companies announced
The companies said the stack brings Nvidia Nemotron open models into Foundry and AIP, "grounded in the Palantir Ontology." Customers can post-train Nemotron on their own data. Nvidia cuOpt software runs inside AIP for optimizations and scenario planning.
The deployment runs on Nvidia reference architectures. It also uses the jointly developed Palantir Sovereign AI Operating System Reference Architecture (SAIOS), which Dell Technologies and Cisco support.
Customers can deploy the stack on premises with Cisco and Dell. They can also deploy it in co-location and cloud with Rackspace and Nebius.
Customer results
Palantir's October release reported these results:
- L3Harris: Heidi Wood, chief transformation officer, said that in a test environment the company was able to "outperform the frontier labs" and "cut costs by 95%."
- McCarthy: The release said McCarthy used post-trained Nemotron for subcontractor drafting. It said this delivered "better performance than the frontier" while "saving 30k hours annually."
- Tampa General: The release said the hospital fine-tuned and post-trained Nemotron. It said this improved model performance by 7% and reduced latency by 22x.
- Ondas: The company is launching Regna AI, built on AIP and open-weight models including Nemotron.
Jeff Whitmer, Nvidia's vice president of system supply chain, said Nvidia's supply chain teams "now move faster and with greater confidence." The release gave no figure for Nvidia's own results.
Neither release disclosed financial terms or contract values.
Results context
Palantir has since reported results for the third quarter of 2026, July to September. Its results release reported revenue growth of 96% year on year and US commercial revenue growth of 148%.
The same release said Palantir raised its full-year 2026 revenue guidance to 91% growth. Investing.com's summary of the earnings call said the company described demand for its Sovereign AI products as strong.
Why it matters (Foundry News analysis)
In this stack, Palantir supplies the software and data layer: Foundry, AIP and the Ontology. Nvidia supplies the models, and partners supply hardware and hosting. That casts Palantir as an integrator of open models rather than a model builder.
For buyers, the new element is a supported path to run and post-train open models on their own infrastructure inside AIP. That path rests on named hardware and cloud partners.
The evidence is still thin, though. The October release covers a broad set of sovereignty efforts. It does not show that every named customer uses the supply-chain deployment announced in September.
All the customer figures come from Palantir's own release. The L3Harris figures come from a test environment.
Three things are worth watching:
- whether Palantir puts a value on this work in its filings
- whether Nvidia publishes figures for its own deployment
- whether the named customers document outcomes beyond Palantir's release
