Australian AI Expansion
Nvidia plans a massive Australian AI expansion to more than double data center capacity by 2027.
Anonymous
Key Points
1
NVIDIA’s Australian DSX AI factories are expected to provide access to its Nemotron open models, enabling organizations to develop regional AI models, applications and agents tailored to local industries.
2
Sharon AI’s planned infrastructure will use NVIDIA HGX B300 and GB300 NVL72 systems, with NVIDIA InfiniBand serving as the primary interconnect across most deployments; one AI factory will instead use RoCE networking.
3
Sharon AI CEO James Manning said Australia’s energy, connectivity and ambition could support an AI leadership role, but stressed that organizations need access to advanced computing without compromising security or sovereignty.
4
NVIDIA said the Australian expansion could spur additional power-generation projects serving both regional and global demand for AI computing, linking the data-center buildout to broader energy infrastructure investment.
Nvidia is planning a massive expansion of artificial intelligence computing power in Australia, aiming to more than double the country's data-center capacity by 2027. Data Center Dynamics reports the company will partner with Australian firms Firmus, CDC, NEXTDC, and AirTrunk to build up to 2 gigawatts of AI-focused computing — compared with Australia's current 1.6 gigawatts. The expansion uses Nvidia's DSX platform to give startups, researchers, and enterprises access to the computing power needed to build AI models and applications.
Sharon AI, an Australian AI cloud provider, is separately scaling its infrastructure to as many as 68,000 Nvidia GPUs as part of the broader national buildout. Yahoo Finance notes the expansion comes as Australia seeks to attract data-center development and strengthen its AI sector, even as regulators and communities raise concerns about energy and water use at such large facilities.
Why Australia Needs More AI Computing Power
AI companies worldwide are racing to build massive data centers because demand for computing power keeps growing. Australia currently has only 1.6 gigawatts of data-center capacity. Adding 2 gigawatts would more than double that — enough to power thousands of AI projects at once. Data Center Dynamics explains this expansion will let local researchers, startups, and large companies develop AI tools without sending their work overseas.
Sharon AI's Massive GPU Investment Strategy
Sharon AI plans to deploy up to 68,000 Nvidia GPUs — the specialized chips that power AI training and processing. The infrastructure will use Nvidia's high-performance HGX B300 and GB300 NVL72 systems, with Nvidia InfiniBand networks connecting most sites. One facility will use an alternative networking technology called RoCE. Sharon AI CEO James Manning said Australia's strong energy supply, connectivity, and local ambition could make the country a leader in AI — but only if companies get access to advanced computing without risking data security or national sovereignty.
Energy and Infrastructure Demands Drive National Debate
Data centers use enormous amounts of electricity and water to cool their equipment. Nvidia said the Australian expansion could drive new power-generation projects to serve both regional and global AI demand. But this massive energy appetite has sparked concerns from regulators and local communities worried about environmental impact. Data Center Dynamics reports that Australia's government is weighing the economic benefits of becoming an AI hub against these infrastructure and sustainability challenges.
Nvidia's DSX Platform Opens AI Development to Everyone
Nvidia's DSX AI factories will provide access to Nemotron open models — foundational AI systems that organizations can customize. This allows companies to build AI applications tailored to Australian industries like mining, agriculture, and finance. Instead of relying on foreign AI providers, local teams can develop and train their own AI systems inside Australia, keeping intellectual property and data within the country.
Sentiment
Mostly PositiveNEGATIVENEUTRALPOSITIVE
Publishers
13
Articles
77
Reach
90
A
AnonymousAuthor
