Nvidia announced on September 9, 2026 that it is working with a group of Australian NVIDIA Cloud Partners and AI infrastructure providers to expand land, power and shell capacity for NVIDIA DSX AI factories, “with up to a 2-gigawatt buildout by 2027.” According to Nvidia’s announcement, the participating providers will operate the AI factories, while Nvidia supplies the DSX platform, accelerated computing, networking, software and ecosystem support.

On this page

Who is doing what

The 2 GW figure covers a group of operators rather than a single campus. Nvidia lists the partners and their roles as follows:

Partner Role as stated by Nvidia
Firmus Expanding Project Southgate, NVIDIA-powered AI factories for Australia and Asia-Pacific
Sharon AI Deploying up to 68,000 NVIDIA GPUs with Quantum InfiniBand and Spectrum-X Ethernet
IREN Applying NVIDIA’s DSX architecture across its portfolio, including the 800 MW Bundey campus in South Australia
CDC Operates more than 550 MW across Australia and New Zealand, with a further 800 MW under construction
NEXTDC Liquid-cooled, high-density infrastructure, including AI factories built on NVIDIA DSX
AirTrunk AI-ready powered shells with direct-to-chip liquid cooling
Megaport Expanding access to NVIDIA accelerated computing through its subsidiary Latitude.sh, for inference
ResetData NVIDIA accelerated computing through its AI cloud platform

The largest numbers

Sharon AI gives the release’s GPU count: up to 68,000 NVIDIA GPUs connected with NVIDIA Quantum InfiniBand and NVIDIA Spectrum-X Ethernet networking, using the DSX platform to serve startups, enterprises and research organizations.

On power, two operators give site-level figures. IREN says it is bringing its delivery and operating experience “to our 800-megawatt Bundey campus in South Australia.” CDC “operates more than 550 megawatts of capacity across Australia and New Zealand, with a further 800 megawatts under construction.”

Users named

Nvidia also names Heidi and Atlassian among organizations using NVIDIA Nemotron open models, datasets and tools. Heidi’s chief technology officer, Yu Liu, said Nemotron 3 Ultra lets the company “build and post-train models for the real complexity of healthcare.” At Atlassian, Guihong Cao, head of engineering for Rovo + AI, said Nvidia’s infrastructure and open models are “foundational to how Atlassian builds and delivers AI.”

What changes

The announcement sets a combined ceiling for the program: up to 2 GW of AI-factory capacity in Australia by 2027, operated by the partners on Nvidia’s DSX platform. The division of labor is explicit in Nvidia’s release: the providers expand land, power and shell capacity and run the AI factories, and Nvidia supplies the platform, computing, networking and software. Progress against the 2 GW figure will show up project by project, at sites such as IREN’s Bundey campus and in CDC’s construction pipeline.

The Model Press tracks the individual campuses in its AI data-center project tracker, and Nvidia’s data-center hardware in its AI chip catalog.

Sources