What happened
NVIDIA announced a strategic expansion of its AI infrastructure footprint in Australia, partnering with eight local providers to build up to 2 gigawatts of capacity by 2027. The initiative involves deploying NVIDIA DSX AI factories, which integrate accelerated computing, networking, and software. Partners include Firmus, Sharon AI, IREN, Megaport, ResetData, CDC, NEXTDC, and AirTrunk. The buildout aims to support local AI development, reduce latency for inference, and provide sovereign data residency options for Australian enterprises and researchers.
NVIDIA announced a collaboration with a group of Australian NVIDIA Cloud Partners (NCPs) and AI infrastructure providers to expand land, power, and shell capacity for AI factories. The partners involved are Firmus, Sharon AI, IREN, Megaport, ResetData, CDC, NEXTDC, and AirTrunk. The collective goal is to deliver up to 2 gigawatts of buildout by 2027.
The infrastructure will be based on the NVIDIA DSX platform, a full-stack solution that includes facilities infrastructure, computing, networking, and software. NVIDIA will provide the DSX platform, accelerated computing, networking, and ecosystem support, while the partners will operate the AI factories. This model is designed to make AI infrastructure more productive, fungible, and durable over its lifecycle.
Specific partner initiatives include Firmus expanding its Project Southgate for hyperscale demand, Sharon AI deploying up to 68,000 NVIDIA GPUs with Quantum InfiniBand and Spectrum-X Ethernet, and IREN applying the DSX architecture to its 800-megawatt Bundey campus in South Australia. Megaport is expanding access through its subsidiary Latitude.sh to support inference needs, while ResetData focuses on sovereign data residency for enterprises and government.
CDC and NEXTDC are providing AI-ready data center infrastructure with advanced cooling technologies, including direct liquid-to-chip cooling, to support high-density workloads. AirTrunk is expanding AI-ready powered shells across the region. These facilities are designed to handle the power and cooling requirements of increasingly compute-intensive AI systems.
The expansion is intended to give Australian innovators greater access to accelerated computing and NVIDIA Nemotron open models. Organizations such as Heidi and Atlassian are already using these models to build applications in healthcare and enterprise search, respectively. The infrastructure aims to support local development of models, applications, and agents while meeting regional and global demand for AI compute.
Source details: nvidianews.nvidia.com ↗
Why it matters
This expansion significantly increases the availability of high-performance AI compute within Australia, reducing reliance on overseas data centers for latency-sensitive and data-sovereign workloads. By integrating the NVIDIA DSX platform, the infrastructure is designed to support multiple generations of AI hardware, ensuring long-term utility. The move also strengthens the local ecosystem by enabling Australian companies to train and deploy models using NVIDIA Nemotron open models, fostering domestic innovation in sectors like healthcare and enterprise software.
The 2-gigawatt buildout represents a substantial increase in local AI compute capacity, which is critical for reducing latency and ensuring data sovereignty for Australian organizations. By bringing this infrastructure onshore, the partnership addresses the growing demand for high-performance computing that cannot be met by existing local capacity.
The use of the NVIDIA DSX platform ensures that the infrastructure is compatible with the CUDA ecosystem and can be enhanced through software updates. This approach extends the useful life of the assets and supports a variety of workloads, from training large models to serving inference requests for AI-native applications.
The initiative supports the development of a local AI ecosystem by providing access to NVIDIA Nemotron open models. This allows Australian companies to customize and fine-tune models for specific industries, such as healthcare and enterprise software, without relying entirely on external providers. This fosters innovation and creates a competitive advantage for local AI-native startups.
The expansion also has broader economic implications, as it supports additional power generation projects and creates skilled jobs in the infrastructure and technology sectors. By positioning Australia as a hub for AI compute in the Asia-Pacific region, the partnership enhances the country's role in the global AI economy.
What to watch next
Monitor the timeline for the 2027 capacity target and the specific power generation projects associated with the buildout. Watch for the adoption rates of NVIDIA Nemotron models among the cited Australian partners, such as Heidi and Atlassian, and assess how this local capacity affects the pricing and availability of AI services in the Asia-Pacific region.
Track the progress of the 2-gigawatt buildout against the 2027 target, including the commissioning of specific campuses like IREN's Bundey site and Firmus's Project Southgate. Delays in power generation or construction could impact the timeline for capacity availability.
Observe the adoption of NVIDIA Nemotron models by Australian enterprises and researchers. The success of local applications, such as Heidi's healthcare AI and Atlassian's Rovo search, will indicate the practical utility of the new infrastructure for domestic innovation.
Monitor the impact on AI service pricing and availability in the region. Increased local capacity may lead to more competitive pricing for compute resources and improved service levels for latency-sensitive applications, potentially shifting workloads from overseas data centers to Australia.