Australian AI infrastructure for research, innovation and national capability
Artificial intelligence is rapidly becoming a critical research and innovation capability. However, many of Australia's most valuable datasets cannot be moved offshore or processed on conventional commercial cloud platforms.
The NIIN Sovereign AI Cloud (SAIC) is being developed as a shared, Australian-hosted AI environment that provides universities and industry partners with secure GPU computing, AI development tools, and built-in governance controls. Designed specifically for research and innovation workloads, the platform enables organisations to train, fine-tune and run advanced AI models while maintaining Australian control of data, intellectual property and research outcomes.
Positioned between batch-oriented supercomputing environments and commercial hyperscale cloud services, the NIIN Sovereign AI Cloud provides continuous, secure access to high-performance AI infrastructure for sensitive and mission-critical workloads.
Why a Sovereign Cloud?
Research teams across Australia are increasingly facing three challenges:
Sensitive data must remain under Australian control
Many AI projects involve health, critical infrastructure, resources, defence-related, cyber security or commercially sensitive datasets that cannot be transferred to overseas environments.
AI needs sustained GPU access
Modern AI workloads require long-running GPU resources for training, fine-tuning and inference that existing national computing facilities were not designed to provide.
Collaboration requires shared environments
Researchers and industry partners need secure, governed platforms where teams can build, test and deploy AI applications together.
First Four Use Cases
Four NIIN universities have nominated inaugural workloads that demonstrate the breadth of demand for sovereign AI infrastructure across Australia's priority sectors.
UNIVERSITY OF TECHNOLOGY SYDNEY: EMERGENCY SERVICES
Sovereign AI and cyber research testbed
Shares real-time multimedia data between emergency service organisations and detects cyber threats, such as fake wireless access points in public spaces.
Why the Sovereign AI Cloud
The team trains models on single-GPU desktops and now needs multi-GPU capacity and a shared development environment.
CURTIN UNIVERSITY: MINING
Mission-driven AI for safer living labs
Combines computer vision, environmental sensing and agentic workflows to monitor hazards on industrial and lab sites, such as PPE compliance and restricted zones.
Why the Sovereign AI Cloud
It needs sustained GPU access, controlled for images of identifiable people.
RMIT UNIVERSITY: HEALTH
Electronic health record de-identification
Trains and runs AI models that de-identify sensitive electronic health records, so clinical data can be used safely for research.
Why the Sovereign AI Cloud
Identifiable clinical data needs a secure environment with strict sovereign controls and extended GPU access.
THE UNIVERSITY OF QUEENSLAND: MINING, RESOURCES AND UTILITIES
Sovereign AI-enabled enzyme discovery
Uses protein-structure models to analyse metagenomes from extreme environments, finding enzymes for site remediation, industrial processes, and reducing lab testing.
Why the Sovereign AI Cloud
The partner’s data is commercially sensitive and can’t be hosted on open national facilities.