Nvidia Signs 8 Australian Partners to Build 2GW of AI Compute Capacity by 2027
Nvidia has signed eight Australian data centre and cloud operators to a coordinated AI infrastructure buildout it says could deliver up to 2 gigawatts of AI compute capacity in the country by 2027.
The eight partners are Firmus, Sharon AI, IREN, Megaport, ResetData, CDC, NextDC, and AirTrunk. Each will deploy what Nvidia calls an “AI factory” — a full-stack arrangement of facilities, compute, networking, software, and reference designs built around Nvidia’s DSX platform and the CUDA ecosystem.
The Partners and Their Positions
CDC operates more than 550 megawatts across Australia and New Zealand today, with a further 800 megawatts under construction. It is the largest established data centre operator in the group and provides the build credibility the 2027 target requires.
AirTrunk has become the primary conduit for American institutional capital entering Australian data centres — a role that gives it financial depth and an existing relationship with hyperscaler tenants.
NextDC has been funding this category of capital expenditure through convertible notes, a financing structure that signals confidence in long-term AI infrastructure demand while managing balance sheet exposure.
IREN and Firmus have both moved aggressively into AI infrastructure. Firmus has crossed 900MW of total contracted capacity across APAC, with OpenAI as its anchor tenant and a deployment underway in Malaysia, making the Australian partnership an extension of a growing regional footprint.
Why 2GW Matters
Australia’s total installed generation capacity exceeds 110 gigawatts nationally. Two gigawatts for AI compute, if fully realized, would represent roughly 1.8% of total installed national capacity dedicated to a single sector, a material allocation that will require coordination with grid operators and long-term energy procurement strategies.
Australia has positioned itself as an attractive location for AI infrastructure on the basis of energy mix, legal stability, and physical proximity to Asian markets without dataflow exposure to mainland China jurisdictions. The Nvidia announcement consolidates that positioning with a specific capacity target and named counterparties.
The AI Factory Model
Nvidia’s framing of these as “AI factories” rather than conventional data centre leases is deliberate. An AI factory in Nvidia’s usage is not a building — it is a vertically integrated stack: DSX-certified compute hardware, CUDA-native networking, and software tooling delivered as a coordinated system rather than leased floorspace. The structure binds partners more tightly to Nvidia’s ecosystem than a traditional supply relationship.
The quotations Nvidia extracted from partners tracked closely with the template: Australia has the energy, connectivity, and infrastructure ambition. The substance is in the 2GW number and the eight names. Getting eight Australian operators to commit simultaneously is the signal.