www.industry-asia-pacific.com
22
'26
Written on Modified on
NVIDIA & Australian Cloud Ecosystem Partner for 2-GW AI Buildout
NVIDIA collaborates with eight Australian cloud and infrastructure partners to expand AI factory capacity up to 2 gigawatts by 2027.
www.nvidia.com
NVIDIA has announced a nationwide collaboration with an ecosystem of Australian NVIDIA Cloud Partners (NCPs) and data center operators to deploy multiple generations of NVIDIA DSX™ AI factories. The multi-partner initiative targets an infrastructure expansion of up to 2 gigawatts (GW) across Australia by 2027, delivering accelerated computing, networking, and software stacks to support domestic and regional AI demand.
Context of the Cooperation
Accelerating artificial intelligence adoption across enterprise, academic, and industrial sectors is driving unprecedented demand for sovereign, high-density computing infrastructure. Developing local AI agents, large language models, and multimodal applications requires access to gigawatt-scale power envelopes, advanced liquid cooling, and low-latency network interconnects without relying entirely on offshore cloud instances.
To address these national compute requirements, cooperation was established between NVIDIA and eight Australian cloud and infrastructure operators: Firmus, Sharon AI, IREN, Megaport, ResetData, CDC Data Centres, NEXTDC, and AirTrunk. The partnership leverages Australia's established sustainable energy reserves, digital infrastructure, and technology talent to build sovereign AI factories capable of serving both domestic innovators and broader Asia-Pacific computing workloads. Raj Mirpuri, Vice President of Global AI Clouds and Infrastructure Ecosystem at NVIDIA, stated that the full-stack DSX platform is compatible with the CUDA ecosystem and enhanced through software over its lifecycle, transforming AI factories into durable, investable infrastructure assets that enable Australian innovators to build AI locally.
Technical Solution and Responsibilities
The initiative centers on the NVIDIA DSX platform, an end-to-end AI factory reference architecture unifying data center facilities, accelerated hardware, networking topologies, and software layers. Responsibilities are distributed between the platform architect and the infrastructure operators:
- NVIDIA delivers the DSX architecture, accelerated compute platforms, networking fabrics, software libraries, and access to NVIDIA Nemotron™ open models.
- Firmus scales its Project Southgate initiative to deliver NVIDIA-powered AI factories for hyperscale and AI-native demand across Australia and the Asia-Pacific region.
- Sharon AI deploys up to 68,000 NVIDIA GPUs interconnected via NVIDIA Quantum InfiniBand and NVIDIA Spectrum-X™ Ethernet networking fabrics.
- IREN applies the DSX blueprint across its vertically integrated platform, contributing power, land, and data center operations, including its 800-megawatt (MW) campus in Bundey, South Australia.
- Megaport provides low-latency edge AI inference access via its subsidiary, Latitude.sh, integrated with its global software-defined network (SDN).
- ResetData provides sovereign cloud infrastructure and localized high-density computing for government, enterprise, and medical research clients requiring Australian data residency.
- CDC Data Centres provides AI-ready facilities featuring 100% renewable electricity and certified zero-water, direct liquid-to-chip cooling architectures.
- NEXTDC develops liquid-cooled, high-density AI data center facilities engineered for DSX clusters interconnected with global cloud networks.
- AirTrunk provides hyperscale powered shells equipped with direct-to-chip liquid cooling to host high-density AI workloads.
At a system level, the NVIDIA DSX architecture maximizes useful compute per megawatt through codesigned hardware-software integration. The platform supports multiple chip generations over its lifespan, maintaining backward and forward compatibility with the NVIDIA CUDA® programming model while allowing dynamic workload distribution across multi-tenant GPU clusters.
Deployment or Implementation
The implementation roadmap outlines a phased buildout across Australian metropolitan and regional hubs, scaling toward 2 gigawatts of cumulative AI factory capacity by 2027.
Early application-layer deployment is already active across Australian enterprises and startups utilizing NVIDIA accelerated infrastructure and Nemotron open models. Healthcare platform Heidi integrates Nemotron 3 Ultra models with clinical data to deploy voice-driven AI for medical record synthesis, while enterprise software leader Atlassian incorporates NVIDIA accelerated compute and fine-tuning pipelines to power its Rovo semantic search and workplace AI agent platform. By pairing local data center facilities with full-stack software and open foundation models, the deployment provides Australian developers, enterprises, and research institutions with sovereign access to production-scale AI compute.
Edited by Romila DSilva, Induportals Editor, with AI assistance.
Deployment or Implementation
The implementation roadmap outlines a phased buildout across Australian metropolitan and regional hubs, scaling toward 2 gigawatts of cumulative AI factory capacity by 2027.
Early application-layer deployment is already active across Australian enterprises and startups utilizing NVIDIA accelerated infrastructure and Nemotron open models. Healthcare platform Heidi integrates Nemotron 3 Ultra models with clinical data to deploy voice-driven AI for medical record synthesis, while enterprise software leader Atlassian incorporates NVIDIA accelerated compute and fine-tuning pipelines to power its Rovo semantic search and workplace AI agent platform. By pairing local data center facilities with full-stack software and open foundation models, the deployment provides Australian developers, enterprises, and research institutions with sovereign access to production-scale AI compute.
Edited by Romila DSilva, Induportals Editor, with AI assistance.
