The Eclipse Foundation has launched the Sovereign AI Foundation, a neutral forum comprising 17 organizations aimed at mapping AI dependencies and evaluating open-source alternatives.
On October 1, the Eclipse Foundation, an international non-profit organization based in Brussels that hosts over 400 open-source projects, announced the formation of the Sovereign AI Foundation. This neutral initiative, independent of vendors, brings together 17 founding organizations, including CEA-List, Ericsson, Infosys, Red Hat, Robert Bosch, and Thales, with the goal of helping them understand their AI dependencies, assess open-source alternatives, and make informed decisions regarding the technologies they adopt. Additional organizations are expected to join, and all members of the Eclipse Foundation are encouraged to participate.
This announcement follows a series of initiatives that includes a free readiness toolkit for the CRA and a partnership with OWASP to share open-source recommendations ahead of the September 11 deadline. The foundation is thus structuring its ecosystem around compliance requirements, this time focusing on AI dependencies.
A Neutral Forum for Mapping Dependencies
According to Eclipse, as AI evolves from experimental to critical operations and decision-making, many organizations rely on models, platforms, and services that they cannot fully inspect, adapt, or control. This dependence can limit their ability to manage data processing locations, choose technologies, and pivot when costs, requirements, or vendor terms change.
Mike Milinkovich, Executive Director of the Eclipse Foundation, emphasizes that sovereignty in AI involves maintaining significant freedom of choice and control over the systems organizations depend on. Eclipse clarifies that this sovereignty does not imply isolation, complete self-sufficiency, or a rejection of proprietary technologies. Instead, it suggests retaining the ability to understand, choose, adapt, and regulate critical AI systems, with viable alternatives and the flexibility to change course.
The Sovereign AI Foundation’s efforts are concentrated in six areas. Four focus on analysis: identifying needs and use cases, assessing technologies and their dependencies, evaluating relevant open-source projects, and sharing experiences. The other two pertain to production, involving the creation of practical resources, such as white papers, action plans, and technology landscape analyses, as well as translating research results—including EU-funded projects—into actionable insights for its members.
Open Source as a Viable Alternative
Open source plays a pivotal role in this approach, according to Eclipse. It provides better visibility into dependencies, greater flexibility to adapt systems, and more options regarding AI deployment methods. The foundation highlights several projects from its portfolio. Eclipse Theia and Theia AI focus on AI-integrated development tools, while Eclipse Enclave offers controlled execution of coding agents in isolated environments regulated by policies. Eclipse PanEval evaluates model capabilities, security, and cybersecurity, and Eclipse LMOS supports multi-agent systems.
The Open VSX registry complements this ecosystem. The foundation describes it as an independent extension infrastructure for open-source and AI-based development tools, boasting nearly one billion downloads per month, a figure it claims. Additionally, it is exploring a broader AI registry for emerging components, including MCP servers, agent skills, and plugins.
The role of the Sovereign AI Foundation is to act as a liaison. It brings together organizations deploying AI, identifies their common needs, and communicates these to the communities developing the technologies, with each project maintaining its governance and roadmap. The foundation will not develop software or specifications; instead, technical development will remain within the Eclipse Foundation projects, creating a feedback loop between deployers and developers.
Industry, Academic, and Open Source Members
The 17 founding members hail from technology, industry, research, and the open-source community. In addition to the organizations already mentioned, the list includes EclipseSource, Engineering Ingegneria Informatica, Eurotech, Kentyou, KU Leuven, Open Elements, Renesas Electronics, The IO Foundation, TypeFox, the University of York, and Vector Informatik. The initiative operates as a forum exclusive to members of the Eclipse Foundation, and other organizations can inquire about membership to join.
Participants have shared testimonials alongside the announcement. CEA-List sees this as a contribution to Europe’s technological independence, leveraging its open-source ecosystem Aidge. Infosys emphasizes the importance of strategic autonomy concerning models, data, and operational execution. Dimitri Van Landuyt from KU Leuven summarizes the challenge with an equation: trust equals transparency multiplied by control, which he believes open-source development principles can improve.
A Dependency Framework Ahead of Tool Deployment
For IT departments, the initiative primarily offers a vocabulary and a framework. The three capabilities associated with sovereignty by the foundation are managing data processing locations, choosing technologies, and being able to pivot when costs and conditions evolve. These represent key questions to address for each AI tool before it reaches users.
These questions await practical resources that the foundation has yet to produce, inviting members to contribute to their development. An IT department affiliated with Eclipse can thus assert its requirements, while others can follow white papers and action plans as they are released. The issue of open-source alternatives within the AI stack will arise, and the framework allows for preparation ahead of tool deployment.



