Ondas and Palantir launch Regna AI: sovereign AI models for defense, fine-tuned on open weights
Ondas and Palantir announced Regna AI today: a new company fine-tuning open-weight AI models that governments and defense firms deploy inside their own environments — sovereignty first, benchmarks later.
- Ondas and Palantir launched Regna AI, a standalone private company, powered by Palantir and bootstrapped by Ondas, developing mission-specialized fine-tuned models for sovereign environments.
- Target deployments include ISR systems — drones, stratospheric balloons, ground vehicles, surface vessels — plus command-and-control nodes, built for denied, degraded, intermittent or limited connectivity.
- Models assess evidence, compare options and propose mission-plan updates for human review and authorization.
- No model is named and no benchmarks are shown; first customer deployments are expected "over the coming months."
- Ondas is the first customer, deploying fine-tuned models for autonomous mission orchestration on the edge; the platform is built on Palantir's AIP.
- Ondas chairman/CEO Eric Brock: "No government should have to build its AI strategy on models or data infrastructure it does not control."
Ondas and Palantir are spinning up a company to do one thing: fine-tune AI models that governments and defense firms fully own. Announced today, Regna AI is a standalone private company, powered by Palantir and bootstrapped by Ondas, that will develop and deploy mission-specialized, fine-tuned AI models inside customers' own sovereign environments. (Newswire)
The pitch is sovereignty, not just performance. Regna AI's models are designed to be deployed in customer environments and on edge hardware — intelligence, surveillance, and reconnaissance systems including drones, stratospheric balloons, ground vehicles, and surface vessels, plus command-and-control nodes — built to operate under denied, degraded, intermittent, or limited connectivity conditions. The capabilities are meant to help operators assess evidence, compare options, and propose mission-plan updates for human review and authorization.

Why sovereign matters
“No government should have to build its AI strategy on models or data infrastructure it does not control,” said Eric Brock, Ondas' chairman and chief executive. Regna AI combines open-weight base models, customer-controlled mission data, and Palantir's operational platform, he said — and Ondas believes it can become “a foundational AI platform for the U.S. Department of War, allied governments and the defense and industrial companies that serve them.”
Ondas will be Regna AI's first customer, deploying fine-tuned models for autonomous mission orchestration on the edge. Beyond that, the company is set up to partner with governments, businesses, and institutions seeking sovereign AI capabilities for mission-critical operations. Palantir contributes model-training infrastructure and support delivery, plus a forward-deployed engineering collaboration to tailor the capabilities to customer workflows. (Newswire)

What we actually know — and don't
The announcement is a strategy and product move, not a model release. No model is named, no benchmark is shown, and the first customer deployments are expected “over the coming months.” Per ExplainX's analysis, Regna AI is initially wholly owned by Ondas (per Seeking Alpha), and Brock confirmed the base models will be open-weight — none named so far. The platform is built on Palantir's AIP, and the human-in-the-loop framing is explicit: models propose, humans authorize.
Why it matters
Sovereign AI is becoming a procurement category of its own. Governments from the EU to the Gulf are writing tenders that require models to run inside national infrastructure, and the frontier labs sell primarily through cloud APIs. Ondas and Palantir are betting that gap is a business — and that open-weight models, fine-tuned on customer data and run at the edge, are how you fill it. Regna AI is the vehicle; whether defense buyers pay for it is the open question.
Frequently asked questions
What is Regna AI?
A standalone private company from Ondas and Palantir that fine-tunes open-weight AI models for governments and defense firms to deploy inside their own sovereign environments — pitched at buyers who need to own, not rent, their AI.
Which models does Regna AI use?
Open-weight base models — none named so far — combined with customer-controlled mission data and built on Palantir's AIP platform. Palantir contributes model-training infrastructure and forward-deployed engineering.
Is there a human in the loop?
Yes, explicitly: the models help operators assess evidence, compare options, and propose mission-plan updates for human review and authorization.
Has Regna AI shipped anything yet?
No. The announcement is a strategy and product move, not a model release: no model is named, no benchmark is shown, and first customer deployments are expected "over the coming months." Ondas will be the first customer, deploying fine-tuned models for autonomous mission orchestration on the edge.