Nexus Foundation launches local AI to dodge cloud costs

What happened

Nexus Foundation has developed Lumen, an air-gapped local AI system that runs open-source models on private hardware within an institution's own building. The foundation operates a two-layer model in which the foundation handles research and maintains open-source standards, while a separate commercial entity manages deployment and support. Institutional deployment costs €25,000–60,000 one-time, with annual service and R&D contribution of €5,000–15,000 per year.

Why it matters

Commercial platforms like ChatGPT, Gemini, and Copilot create single points of failure (if the server goes down, the business goes down), change model behaviour without user consent, and send sensitive data to external servers in ways that may conflict with GDPR and professional confidentiality rules. A commercial AI assistant confirmed to the author that in public cloud systems, verification of data handling is in most cases simply 'trust' because clients cannot audit what happens inside the provider's data centre. Local AI systems address these constraints by keeping all data and control within the institution.

What to watch

The cost comparison shows enterprise cloud deployments cost €270,000–590,000 in year one plus €100,000–160,000 annually, whereas the Nexus model's upfront investment is permanent and carries no recurring licence fees. The foundation's stated philosophy is that AI should amplify human responsibility—not replace it—and should belong to the institution using it rather than to a vendor.