key insights:
- Australia’s proposed 2GW AI buildout would exceed its current 1.6GW computing capacity.
- Power availability, grid connections and approvals remain major execution factors.
- Eight infrastructure partners will operate facilities while NVIDIA supplies computing technology and software.
NVIDIA plans to expand Australia’s AI computing capacity through partnerships targeting up to 2 gigawatts by 2027. NVIDIA announced the initiative on Sept. 9, bringing together eight infrastructure providers to support rising local demand.
The participating companies include Firmus, Sharon AI, IREN, Megaport, ResetData, CDC, NEXTDC and AirTrunk. They will develop and operate AI factories, while NVIDIA will provide accelerated computing, networking, software and technical support.
The planned capacity exceeds Australia’s current computing capacity of about 1.6 gigawatts. However, the 2GW figure represents potential infrastructure capacity, rather than completed facilities or actual electricity consumption.
NEWS: NVIDIA and Australia’s AI infrastructure ecosystem are expanding land, power and shell capacity to host AI factories, supporting the nation’s growing demand for AI compute.
Built on NVIDIA DSX, these full-stack AI factories provide high-performance AI compute for…
— NVIDIA Newsroom (@nvidianewsroom) September 10, 2026
Eight Partners Divide Australia’s Infrastructure Push
The partnership assigns different infrastructure roles across Australia’s expanding AI ecosystem. Firmus is developing Project Southgate, which aims to serve hyperscale and AI-native demand across Australia and the Asia-Pacific region.
Sharon AI plans to deploy as many as 68,000 NVIDIA GPUs through the DSX platform. Its systems will use Quantum InfiniBand and Spectrum-X Ethernet networking. The company expects to serve startups, enterprises, government agencies and research organizations.
Meanwhile, IREN plans to apply the DSX architecture at its planned Bundey campus in South Australia. The site has a targeted capacity of 800 megawatts. IREN has historically operated Bitcoin mining infrastructure but has expanded into AI cloud services and high-performance computing.
Megaport will expand accelerated computing through its Latitude.sh subsidiary. Its focus includes AI inference services and reducing processing latency for enterprise customers. ResetData will provide high-performance computing through its AI cloud platform. The company targets enterprise, government and research customers requiring Australian data residency.
CDC operates than 550 megawatts across Australia and New Zealand. CDC also has another 800 megawatts, under construction according to company figures. NEXTDC and AirTrunk are also developing AI‑ready infrastructure.
Their projects include high-density facilities, liquid cooling and connectivity designed for increasingly demanding AI workloads.
DSX links computing with AI software
NVIDIA’s DSX platform combines facilities infrastructure, computing, networking, software and reference designs. The architecture supports the CUDA ecosystem and aims to accommodate future generations of computing hardware.
The structure could allow operators to extend infrastructure without redesigning every layer for each hardware generation. Consequently, the partners can target broader AI workloads while maintaining infrastructure flexibility.
The expanded capacity should also increase Australian access to NVIDIA Nemotron open models. Local companies already use the technology across healthcare, enterprise search and other applications.
Heidi combines Nemotron models with specialized clinical knowledge for healthcare applications. Atlassian uses NVIDIA technology to improve semantic search within its Rovo AI platform
However, NVIDIA has not disclosed the project’s total cost or individual investment commitments. It also has not confirmed how much planned capacity has secured financing, approvals or grid connections.
Power access could shape the 2027 target
The proposed expansion places electricity supply, grid access and cooling capacity at the center of execution risks. Large AI facilities require continuous power, while dense GPU clusters can require advanced cooling systems.
The company said expanded AI capacity could support additional power-generation projects. However, it did not identify specific generation assets or electricity contracts linked to the initiative.
Australian authorities and infrastructure operators will still need to address grid connections, construction approvals and environmental requirements. These factors could influence how quickly individual projects progress.
NEXTDC separately announced plans to raise A$1.1 billion through convertible notes. Reuters reported that the financing supports its broader AI infrastructure expansion, rather than the NVIDIA collaboration alone.
The next indicators will include site approvals, power agreements, construction milestones and GPU deliveries. Partners have not published a common timetable showing how much capacity will become operational at each stage.
Australia’s AI ambitions face execution test
The planned 2GW expansion could substantially increase Australia’s access to high-performance computing. It also reflects a broader shift as AI demand pushes infrastructure providers toward larger power and data center projects.
The initiative creates several important implications for the market
- Capacity growth could strengthen Australia’s position as a regional AI infrastructure hub.
- Power availability could become a major constraint as operators pursue larger GPU deployments.
- Local computing could help Australian businesses and researchers develop applications within the country.
NVIDIA also warned that infrastructure projections depend on demand, partner execution, regulation and technology availability. Therefore, the 2GW target remains a forward-looking plan rather than completed computing capacity.
Conclusion
The plan, from Australia brings together data centers, power systems, internet connections and artificial intelligence programs under one growth plan. How well it works will depend on whether the people involved get the approvals, power and money before the year 2027.
For Australia, the projects could expand domestic AI capabilities while supporting regional demand. For infrastructure operators, the buildout highlights the growing value of facilities capable of supporting successive generations of high-density AI computing.
