Sovereign AI Infrastructure for Local Iwi

Empowering Māori communities with private, culturally-aligned AI capabilities

A transformative initiative to build sovereign AI infrastructure that respects data privacy, cultural values, and iwi self-determination

Vision

Empowering local iwi with private, culturally-aligned AI capabilities

To empower local iwi with private, culturally-aligned AI capabilities by building a sovereign, locally hosted AI cluster. This infrastructure will prioritize data privacy, Māori data sovereignty, and tikanga Māori, while enabling iwi-led innovation across education, governance, and cultural preservation.

“This initiative presents a practical, respectful pathway to bring generative AI into the hands of Māori communities—controlled by them, for them. By combining cost-effective GPU clusters, open-source tools, decentralized storage, and deep cultural alignment, we can build a resilient digital infrastructure that serves the future of te ao Māori.”

Objectives

Building a foundation for digital sovereignty

Our objectives center on empowering iwi with full control over their data and the AI systems that process it, ensuring alignment with Māori values and needs.

Open-Source AI Cluster

Establish an open-source, GPU-powered AI cluster controlled and governed by iwi.

Full Data Sovereignty

Ensure full data sovereignty: no reliance on cloud providers or foreign infrastructure.

Culturally-Aligned Tools

Provide tools and models aligned with te ao Māori values, developed in consultation with iwi.

Sustainable Access

Offer sustainable, secure, and scalable access to generative AI for whānau, hapū, and iwi use cases.

Key Capabilities

Transformative AI tools designed for and with iwi

Our infrastructure enables a wide range of culturally-grounded AI applications that support language preservation, governance, and education while maintaining data sovereignty.

Language & Cultural Tools

  • Chatbots trained on iwi-specific history, documents, and dialects
  • Voice assistants understanding and speaking te reo Māori
  • Generative tools to assist with mihi, karakia, and reo learning

Knowledge & Governance Support

  • Secure search and summarisation of iwi archives
  • Meeting minute summarisation, grant writing aids, proposal templates
  • Decision-support tools powered by locally tuned models

Education & Youth Engagement

  • AI tools for kura kaupapa Māori students to interact with language and culture
  • Workshops on building and using AI with iwi values

Technical Architecture

How we're building the infrastructure

Hardware

  • Central coordination server (EPYC 7551P-based node)
  • Distributed worker nodes using repurposed GPU mining rigs
  • Encrypted local storage (NAS or Ceph cluster)
  • Optional: Integration of decentralized storage (e.g., IPFS, Filecoin)

Software Stack

  • LLM inference: vLLM / TGI / Ollama
  • Fine-tuning: LoRA, QLoRA, PEFT methods
  • Vector DB for document Q&A: FAISS or Qdrant
  • APIs & UI: FastAPI + lightweight web dashboard
  • Container orchestration: Docker Swarm / Ray / Ansible

Decentralization & Data Sovereignty

Our architecture emphasizes local control and ownership. By using distributed compute nodes and encrypted local storage, we ensure that sensitive iwi data remains completely under iwi control. The optional integration with decentralized storage technologies provides additional resilience for public cultural data that iwi choose to share.

Governance & Ethics

Ensuring cultural integrity and community control

Iwi Leadership

Led by a steering group of iwi representatives and Māori data experts

Consent-Based

Opt-in participation for data inclusion

Transparency

Transparency of model training data and outputs

Cultural Alignment

Alignment with kaupapa Māori principles and tikanga

The governance structure of our sovereign AI infrastructure is designed to ensure that iwi maintain complete control over their data and how it's used. This approach respects the principles of tino rangatiratanga (self-determination) and kaitiakitanga (guardianship).

All participating iwi will have representatives on the steering committee, with decisions made through consensus and guided by cultural protocols. This ensures that the technological implementation always remains aligned with Māori values and priorities.

Deployment Roadmap

Our implementation timeline and sustainability approach

Implementation Timeline

Phase 1: Pilot

  • Collaborate with one iwi partner
  • Deploy local cluster (1 server + 4 GPU rigs)
  • Integrate small language model (e.g., Mistral or DeepSeek) trained on iwi-approved documents
  • Provide private dashboard + chatbot interface

Phase 2: Community Feedback

  • Gather usage insights and feedback
  • Run training workshops with rangatahi and kaumatua
  • Adjust governance, access controls, and features

Phase 3: Expansion

  • Add more compute nodes as needed
  • Extend to additional iwi on invitation
  • Support multilingual and multimodal models (audio, visual)

Funding & Sustainability

Note: We are actively seeking funding partners for this initiative. If you're interested in supporting sovereign AI infrastructure for iwi, please contact us.

Seeking Initial Funding From

  • Te Puni Kōkiri
  • MBIE
  • Callaghan Innovation

Proposed Long-term Model

  • Tiered access (free for iwi, grant-funded dev work)
  • Paid services to NGOs or researchers under iwi terms
  • Public-good AI hosting with usage limits