Ondas and Palantir establish Regna AI for customer-controlled defense AI

Ondas and Palantir announced Regna AI on October 12, creating a company to develop specialized AI for defense and other sensitive operations. The goal is to put models inside infrastructure customers control, including equipment in the field. Ondas will initially own the business outright and become its first customer; Palantir will supply training infrastructure and engineering support.

From SkyWeaver to a separate company

The venture follows the companies’ work on SkyWeaver C4I, an open-architecture command, control, communications, computers and intelligence system. That collaboration is designed to connect surveillance capabilities across platforms, from the stratosphere to equipment operating closer to the ground. Regna AI gives the partners a new commercial structure for developing models around those operational needs.

The companies describe the intended approach as sovereign AI: customers would retain ownership and control of sensitive mission data and the models developed from it. Government and commercial users would train, customize and deploy models within infrastructure they control.

No government should have to build its AI strategy on models or data infrastructure it does not control.

Eric Brock, chairman and chief executive officer of Ondas

Models beside the equipment

Ondas plans to use the models for autonomous mission orchestration at the edge—coordinating missions through computing near equipment and sensors, rather than relying exclusively on centralized cloud infrastructure. Regna AI’s proposed operating environment includes communications that are unreliable, intermittent, deliberately disrupted or otherwise restricted. Local execution is intended to support decisions without continuous access to outside networks.

The intended hardware spans drones, stratospheric balloons, ground vehicles and surface vessels, alongside command-and-control systems.

According to the announcement described by CTech, the software is designed to assist operators in three linked tasks:

• Evaluate evidence relevant to an operation.
• Compare possible courses of action.
• Suggest updates to mission plans for human review and authorization.

The stated role is to help operators assess options and recommend adjustments; people would retain authority over those proposals. Human authorization is part of the announced design, while the models’ field capabilities remain intended functions rather than demonstrated results in this announcement.

Surveillance first, wider markets later

Regna AI plans to start with intelligence, surveillance and reconnaissance, often shortened to ISR. Its proposed expansion reaches other defense applications, then security, critical infrastructure and industrial markets.

The next planned step is to scale the technology platform and support initial customer deployments over the coming months. The companies did not disclose financial terms, including investment commitments or potential revenue contributions.