Ondas Holdings and Palantir Technologies have formed a joint venture called Regna AI to develop sovereign artificial intelligence models for defense applications. The partnership aims to give nations control over their own AI systems, reducing dependence on foreign technology and enhancing national security.
Why sovereign AI matters for defense
Defense agencies have grown wary of relying on AI models built and hosted by outside companies, especially those based in other countries. Sovereign AI means the models are developed, trained, and operated within a nation's own borders, under its own laws and oversight. For military and intelligence work, that control is critical. It keeps sensitive data from leaving the country and ensures that the AI's behavior can be audited and adjusted by the government that uses it.
Regna AI is betting that this demand will only grow. The joint venture will focus on building models that can run on a country's own infrastructure, using its own data, without needing to send information to external cloud servers. That's a departure from how most advanced AI is built today, where a handful of large tech firms dominate the underlying models and the compute they run on.
What each partner brings
Ondas Holdings, known for its private wireless and drone operations, brings expertise in secure communications and edge computing. Palantir, a longtime contractor for defense and intelligence agencies, offers its data integration and AI deployment platforms. Together, they plan to deliver AI systems that can operate in disconnected or contested environments — where a reliable link to a commercial cloud can't be guaranteed.
The joint venture hasn't disclosed which countries it's in talks with, nor the financial terms of the deal. But the launch itself signals a shift: defense customers are no longer satisfied with off-the-shelf AI. They want models they can own, modify, and trust.
The broader shift toward autonomous systems
Regna AI's arrival fits a pattern. Militaries and spy agencies around the world are pushing for AI that can act on its own — identifying targets, sorting intelligence, even making recommendations — without waiting for a human to click approve. That raises hard questions about accountability and rules of engagement, but the trend is clear. The U.S. Department of Defense, for one, has been vocal about wanting AI that can keep up with the speed of modern warfare.
Sovereign models are a piece of that puzzle. If a country can't control the AI it uses, it can't fully trust it in a crisis. Regna AI is positioning itself as the answer for nations that want the benefits of advanced AI without the strings attached.
What happens next
The joint venture is now operational, though Ondas and Palantir haven't said when the first sovereign model will be ready or which government will be the first customer. For now, the two companies are building out the team and the technology. Watch for announcements about pilot programs or partnerships with specific defense agencies — that's when the real test begins.
