Australia’s ambitions to become a serious artificial intelligence hub just gained considerably more weight.
NVIDIA says it is working with a group of Australian cloud and data-centre companies on infrastructure capable of supporting up to 2 gigawatts of AI-related capacity by 2027. That is a striking figure in a country where total computing capacity currently sits at about 1.6GW, according to industry research cited by Reuters.
The announcement is not simply about shipping more GPUs to Australia. It is about building the physical foundations — power, cooling, networking, data-centre space and accelerated computing — needed for increasingly demanding AI systems to be trained and operated locally.
What NVIDIA is building in Australia
According to NVIDIA’s official announcement, the company is collaborating with Firmus, Sharon AI, IREN, Megaport, ResetData, CDC, NEXTDC and AirTrunk.
The participating providers will operate the facilities, while NVIDIA supplies its DSX platform, accelerated computing technology, networking, software and ecosystem support. NVIDIA describes DSX as a full-stack platform for what it calls “AI factories” — data-centre infrastructure designed specifically around large-scale AI workloads.
Some of the individual projects are already substantial. Sharon AI plans infrastructure capable of deploying up to 68,000 NVIDIA GPUs, while IREN is applying NVIDIA’s architecture to its 800MW Bundey campus in South Australia. CDC says it already operates more than 550MW across Australia and New Zealand with another 800MW under construction.
Why 2GW is such a big number
Reuters reports that Australia’s existing computing capacity is around 1.6GW. NVIDIA’s partners are targeting a buildout of up to another 2GW by 2027.
That does not mean two gigawatts of new capacity will appear overnight, nor that every proposed project is guaranteed to reach full scale. But it illustrates how quickly the economics of data centres are changing as AI models require far more computing power than conventional cloud workloads.
For Australia, the opportunity is bigger than hosting overseas AI companies. More domestic compute could give local startups, universities, researchers and businesses access to infrastructure that previously required relying heavily on overseas capacity.
AI sovereignty is becoming part of the story
Keeping more AI computing inside Australia can also matter for data residency, latency and sovereignty. ResetData, for example, is specifically targeting enterprises, government and research customers that require Australian data residency.
NVIDIA also says expanded local infrastructure will give Australian developers greater access to its accelerated computing platform and Nemotron open models. Australian companies including Atlassian and healthcare AI company Heidi are already highlighted by NVIDIA as users of its AI ecosystem.
That matters because the next stage of the AI race may be determined as much by access to computing infrastructure as by who develops the best model.
There is a major energy question too
The scale also creates an uncomfortable but unavoidable question: where will all the electricity come from?
Reuters reported on September 10 that Australia’s rapid data-centre expansion is already attracting calls for stronger regulation because of electricity and water consumption.
NVIDIA says the expansion can support additional power-generation projects, while several partners are emphasising more efficient cooling and renewable electricity. CDC, for example, says its facilities use renewable electricity and zero-water advanced cooling technologies.
But a 2GW-scale expansion is large enough that energy infrastructure cannot be treated as a footnote. Australia will have to balance its desire to attract AI investment with grid capacity, renewable generation, water use and the needs of surrounding communities.
Australia has a genuine opportunity
Australia has several advantages in the global AI infrastructure race: land, renewable-energy potential, strong international connectivity, political stability and proximity to fast-growing Asian markets.
What it has historically lacked compared with the United States is enormous domestic computing capacity.
NVIDIA’s latest push does not solve that gap by itself. But if even a substantial portion of the planned capacity comes online, Australia could move from being primarily a consumer of frontier AI technology toward becoming one of the places where that technology is actually built and operated.
That is why this announcement deserves attention beyond the usual NVIDIA headlines. The AI boom is increasingly becoming an infrastructure race — and Australia now wants a much bigger piece of it.
Featured image: BoliviaInteligente/Unsplash.
