Nvidia targets 2 GW AI buildout in Australia

Nvidia (NVDA) is working with a group of Australian infrastructure partners to develop up to 2 gigawatts of AI factory capacity in the country, with the buildout targeted for completion by 2027. The initiative is structured around data center campuses designed to host multiple generations of Nvidia’s hardware and software stack. The program positions Australia as a significant regional hub for AI compute, with a focus on high-performance infrastructure tailored to intensive AI workloads.

The announced partner group includes Firmus, Sharon AI, IREN, Megaport, ResetData, CDC, NEXTDC and AirTrunk. These companies span data center development, connectivity, and AI infrastructure operations, forming a network intended to scale capacity across multiple sites. The collaboration is centered on standardized platforms and networking so that capacity can be expanded while maintaining consistent performance characteristics.

Within the overall 2 gigawatt target, several large campus-level projects account for a substantial share of planned capacity. IREN’s Bundey campus in South Australia is cited as providing about 800 megawatts, making it one of the largest single components of the program. The site is positioned as a key AI facility, contributing a significant block of power and floor space to the national buildout.

CDC Group is reported to be developing a further 800 megawatts of capacity in Australia on top of an existing 550 megawatt footprint. This expansion underscores the scale of investment being coordinated with Nvidia’s technology roadmap. Together, the IREN and CDC projects represent a large portion of the planned 2 gigawatts, anchoring the broader network of AI factory sites across the country.

Sharon AI is reported to plan deployment of up to 68,000 Nvidia GPUs as part of the Australian buildout. This prospective fleet highlights the intensity of compute anticipated at these sites and the central role of Nvidia’s accelerators in delivering AI performance. The GPU deployment is expected to be integrated into facilities that are optimized for large-scale, high-density AI workloads.

The Australian AI factory sites are described as running Nvidia’s DSX AI-factory platform, a software and systems stack designed to coordinate and manage AI infrastructure. They are also reported to use Nvidia networking technologies, including Quantum InfiniBand and Spectrum-X. This combination is intended to enable high-bandwidth, low-latency communication between GPUs, which is critical for training and serving large AI models.

Early launch customers for the Australian AI factory sites include Atlassian (TEAM) and healthcare startup Heidi. Their participation illustrates initial commercial demand for the new capacity and provides reference use cases in enterprise software and healthcare. These early users are positioned to tap into the advanced compute and networking capabilities as the facilities come online.

By concentrating significant GPU deployments and leveraging a unified platform and networking stack, the program deepens Nvidia’s infrastructure presence in Australia. The coordinated efforts with data center and connectivity partners create an environment where Nvidia hardware, software and networking are tightly integrated. If the targeted buildout is realized, it would markedly expand AI-ready land, power and shell capacity in the region and could support a wide range of future AI applications and customers.