Defence technology company Ondas and American software giant Palantir launched a new company called Regna AI on 13 October 2026. The company will develop and deploy AI solutions for governments, security agencies, and organisations operating in sensitive environments. Ondas will be the sole owner, and Palantir will provide the technology infrastructure.
What happened
The new company is American but partly based on Ondas’s 1,000 employees in Israel. This model allows Ondas to retain full ownership while using Palantir’s technology. The focus is on organisations operating in sensitive environments that require full control over infrastructure and data.
The joint launch reflects growing demand for AI systems that are not connected to external cloud services. Government and security organisations are unwilling to transfer sensitive data to commercial cloud providers, even when they offer security assurances.
Why it matters beyond the headline
The Israeli market for enterprise AI systems is becoming more competitive. Companies building AI systems for clients in regulated sectors need to understand that the technical requirements are fundamentally different from consumer-facing systems.
A controlled environment means the model, data, and infrastructure remain inside the organisation. There are no API calls to an external provider, no automatic model updates, and no dependency on cloud service availability. This requires an entirely different architecture from solutions based on OpenAI, Anthropic, or Google.
This competition also affects smaller companies. If you offer custom AI systems, potential clients in finance or government will ask you hard questions about where the data sits and who can access it.
What to do this week
If you build AI systems for clients, check whether you have a clear answer to the question: “Can the system operate in a closed environment without internet connectivity?” If the answer is no, this does not disqualify you from most of the market, but it does disqualify you from clients in regulated sectors.
If you work with clients in banking, insurance, or government, ask them what their requirements are regarding data location and infrastructure control. Sometimes they do not yet know, and such a conversation can prevent problems later in the project. For more on building AI systems in controlled environments, see AI agents for business.
Sources
Frequently asked questions
Can I use the ChatGPT API in a government project?
Not in most cases. Government and security organisations require that the model and data remain inside the organisation, without transferring information to an external provider. This requires models installed on internal infrastructure.
What is the difference between a cloud AI system and one in a controlled environment?
A cloud system sends data to an external provider for processing. A system in a controlled environment runs the model on the organisation's internal infrastructure, without internet connectivity. This requires installation, maintenance, and manual updates, but provides full control.