NVIDIA Partners with Eight Australian Data Center Operators in 2-Gigawatt Sovereign AI Buildout by 2027

In its most aggressive international infrastructure expansion to date, NVIDIA has announced a multi-party coalition with eight prominent Australian cloud providers and data center operators to deploy up to two gigawatts of accelerated AI compute capacity by 2027. The coordinated rollout targets surging demand from enterprise foundation model developers, national security agencies, and regional startups across Australia and the broader Asia-Pacific basin.

The participating consortium spans Australia's primary physical and digital infrastructure operators, including Firmus, Sharon AI, IREN, Megaport, ResetData, CDC Data Centres, NEXTDC, and AirTrunk. Under the framework, infrastructure partners will supply physical land, electrical grid capacity, and high-density shells, while NVIDIA delivers its full-stack DSX platform, accelerated GPU nodes, high-bandwidth networking fabric, and software tooling.

Central to the expansion is the adoption of NVIDIA DSX, a standardized reference architecture engineered specifically for multi-generation gigawatt deployments. Unlike legacy cloud hosting environments designed around generic CPU workloads and air-cooled racks, AI factories demand sustained megawatt-scale power density and continuous thermal dissipation.

Several partner operators have already commenced construction on specialized campuses. IREN is deploying the DSX blueprint across its 800-megawatt Bundey campus in South Australia, combining vertically integrated renewable power with modular data halls. Concurrently, Sharon AI is deploying up to 68,000 NVIDIA GPUs interconnected via NVIDIA Quantum InfiniBand and Spectrum-X Ethernet networking. Physical facilities developed by CDC Data Centres and NEXTDC incorporate certified direct liquid-to-chip cooling systems using 100% renewable energy and closed-loop, zero-water consumption architectures to support ultra-dense compute envelopes.

Beyond raw computational horsepower, the initiative represents a structural response to global data sovereignty mandates. By hosting accelerated clusters within domestic borders, Australian government entities, healthcare researchers, and financial institutions can train and fine-tune frontier models without transmitting proprietary weights or customer data across foreign jurisdictions.

Local organizations are already deploying the infrastructure alongside NVIDIA Nemotron open-weight models. Australian enterprise software giant Atlassian is leveraging the local accelerated computing base to scale reinforcement learning and semantic indexing within its Rovo enterprise search engine. In healthcare, clinical platform Heidi is coupling local compute clusters with domain-specific medical data to train voice-driven diagnostic models under strict national privacy protocols.

By packaging land, renewable grid interconnects, liquid-cooled physical infrastructure, and silicon into a repeatable asset class, the Australian partnership establishes a benchmark blueprint for sovereign AI deployments worldwide.

• NVIDIA Expands AI Infrastructure Capacity in Partnership With Australia's Data Center Ecosystem — NVIDIA Newsroom
• Nvidia Announces 2GW AI Data Center in Australia: Next Wave of Growth Heading Globally — TradingKey
• Driving Australia's AI Future with CDC, Firmus, and NVIDIA — CDC Data Centres