The news
On October 1, 2026, Palantir Technologies (PLTR) and Armada announced a partnership to deliver sovereign AI, meaning AI systems a customer owns and controls end to end, on infrastructure manufactured in the United States and allied nations. Palantir named Armada its Inaugural Certified Modular Data Center Partner, according to the companies' announcement.
The joint offering pairs Palantir's Sovereign AI Operating System, which the companies say is built on Palantir's AIP, Ontology, Foundry and Apollo software, with Armada's infrastructure layer. That layer includes Galleon modular data centers, which are prefabricated units that can be shipped and set up on site, the Armada Platform that keeps each deployment under the customer's control, and a Sovereign AI Grid that links deployments into one distributed system.
The companies said customers will be able to run open-weight models, AI models whose trained parameters are published for anyone to run, on compute, data and physical infrastructure they own outright. They said deployments can be ready in months rather than years, and can operate on-grid, off-grid or fully air-gapped, meaning cut off from outside networks.
Palantir co-founder and CEO Alex Karp said in the announcement that sovereignty is not something you rent, and described it as owning the weights, the data and the hardware. Armada co-founder and CEO Dan Wright said the partnership makes sovereign AI faster to deploy and more resilient.
The announcement named no customers. Financial publication Parameter reported that neither company disclosed financial terms, contract values, projected revenue or deployment schedules. The companies said demand is coming from enterprise and government customers who want model weights, data and physical systems inside their own security boundary.
The numbers
- Partner designation
- Inaugural Certified Modular Data Center Partner
- Claimed deployment time
- Months rather than years (company claim)
- Disclosed deal value
- None
Why CEOs should care
For CIOs and CISOs in government, defense and regulated industries, this is a packaged alternative to running AI in a public cloud: software, hardware and networking sold as one stack that can sit in your own facility or run fully disconnected. Before engaging, ask who maintains the hardware, how model updates reach an air-gapped site, and what certifications the Galleon units carry for your sector.
For CFOs, the trade-off is ownership versus flexibility. Owning compute avoids dependence on outside providers, which the companies cite as the main driver of demand, but it moves capital spending and capacity planning onto your balance sheet. No pricing was disclosed, so request full lifecycle costs, including power, cooling and refresh cycles, before comparing with cloud options.
For boards weighing AI strategy, the 'US and allied-built' framing signals that supply chain origin is becoming a buying criterion. If your organization has national security or data residency obligations, ask management whether current AI vendors can document where hardware is made and where model weights are stored.
The bigger picture
Sovereign AI has become a selling point for software and cloud vendors as governments and large companies look to reduce reliance on hosted models and foreign hardware supply chains, the companies said. Palantir calling Armada its inaugural certified partner suggests a program that may add more hardware partners, though the announcement did not say so.
What’s next
Watch for the first named customers, any pricing or contract disclosures, and whether Palantir certifies other modular data center makers. Those details will show whether the partnership becomes a meaningful revenue line or remains a reference design.
What “Fact-checked” means
Fact-checking means testing a story’s facts against the evidence before it is published. This story went through at least two separate checks before this version was published.
- What we checked
- Its names, figures, dates, job titles, quotes and who said what were checked against the story’s sources, including its main source where it could be opened. The headline was checked for accuracy and overstatement.
- How
- A first check reviewed the whole story. If it passed, a second, skeptical check went back to the sources to look for mistakes in the most important facts. If a check flagged the story, it was edited to fix the problems found, and a separate re-check then reviewed the whole story again.
- Who
- The checks are made by our newsroom, as steps kept separate from the writing, under rules set by our editor, Hussein Mukhtar. A story the checks still flag is held for the editor, who decides whether it is fixed, published or dropped.
- If something is wrong
- “Fact-checked” does not mean error-free. If a material error is found after publication, we correct the story and add a note saying what changed. Report an error
Companies in this story








