Defense contractor Ondas Holdings and data-analytics firm Palantir Technologies are forming a new company, Regna AI, built around a problem that has grown more urgent as artificial intelligence spreads through military and government operations: how to use advanced AI without handing sensitive data or decision-making systems to outside providers. The venture will develop mission-specific AI models designed to run entirely within a customer's own environment, including on hardware deployed in remote or contested locations.
Ondas, a U.S.-listed defense technology group, will own Regna AI outright at the outset and serve as its first customer, while Palantir contributes the infrastructure needed to train and adapt the models to specific operational needs. The arrangement reflects a broader trend among governments and regulated industries weighing how much control to cede when adopting commercial AI tools, a tension that has also shaped consumer debates over data sovereignty and cross-border processing - for readers tracking how providers handle jurisdiction and data exposure in a different but related context, you can see the comparison of how privacy-focused services approach similar trade-offs. For organizations handling classified or operationally critical information, the stakes are considerably higher than for typical enterprise software adoption.
Why Data Control Has Become a Defense Priority
Commercial large language models typically require sending data to external servers for processing, an arrangement that is unworkable for agencies operating under strict security classifications. Even when providers offer assurances about encryption or data handling, the underlying architecture often still depends on infrastructure the customer does not fully control. Regna AI's premise is that military and government users should not have to choose between powerful AI and operational security. Instead of training a general-purpose model from scratch, an enormously costly undertaking, the company intends to build specialized models around a customer's existing operational knowledge, then deploy them inside environments the customer already secures.
From Surveillance to Autonomous Systems
Regna AI's initial focus is intelligence, surveillance and reconnaissance work, with plans to expand into broader defense applications, critical infrastructure and industrial security. A central use case involves autonomous systems - drones, ground vehicles, surface vessels and stratospheric balloons - operating in places where communications are unreliable or actively jammed. In those conditions, a system cannot depend on a constant link to a remote server; it needs to reason locally. The models are intended to help operators weigh evidence, compare possible courses of action and suggest adjustments to mission plans, with final decisions remaining subject to human review rather than automated execution.
Building on an Existing Partnership
The new venture extends an earlier collaboration between the two companies on SkyWeaver C4I, an open-architecture system meant to link surveillance capabilities from high-altitude platforms down to units operating at the tactical edge. That groundwork gives Regna AI a tested foundation for coordinating multiple platform types under limited connectivity, rather than starting from zero. The move also fits a pattern of rapid consolidation at Ondas, which has acquired more than fifteen defense technology businesses over the past two years, including three Israeli firms - Insignito, Ottopia Defense and Caribou Labs - picked up last month for a combined $56 million, spanning drones, robotics, counter-drone systems and communications.