Nvidia Backs 2 Gigawatts of Australian AI Data Centers by 2027

Nvidia is anchoring 2 gigawatts of Australian data center capacity by 2027, partnering with eight operators across cloud, fiber and colocation infrastructure.

Key Takeaways

• Nvidia is partnering with eight Australian operators to anchor 2 gigawatts of data center capacity by 2027
• Sharon AI plans to deploy as many as 68,000 Nvidia GPUs across its share of the buildout
• IREN will run its portion of the infrastructure from an 800-megawatt campus in South Australia
• Australia joins Malaysia and Finland in absorbing hyperscale AI capacity as US sites face grid interconnection queues

The eight partners, Firmus, Sharon AI, IREN, Megaport, ResetData, CDC, NEXTDC and AirTrunk, span the full regional supply chain, suggesting Nvidia is constructing a vertically integrated ecosystem rather than backing a single operator. Sharon AI alone plans to deploy as many as 68,000 Nvidia GPUs, the graphics processors that power most large-scale AI model training and inference.

IREN will run its share out of an 800-megawatt campus in South Australia.

The scale places Australia among the handful of countries hosting gigawatt-class AI infrastructure clusters this year.

Australian AI Buildout Signals Power, Not Chips, Is Now The Binding Constraint

Nvidia, the dominant supplier of AI training chips, framed the deal as meeting “surging demand” from AI labs and enterprises, according to the company’s announcement. A gigawatt of data center capacity is roughly enough electricity to power 750,000 US homes, giving a sense of the physical footprint before a single chip processes a query.

The capital committed across eight partners says more about where the bottleneck sits than any press release does.

Power availability, not chip supply, has become the binding constraint on AI expansion through 2026, with data centers moving from an obscure infrastructure debate into a fight over electricity bills.

Also Read: OpenAI Rogue Agents Uncovered Using 10 Unauthorized Sites

Why Gigawatt Clusters Keep Multiplying

Australia joins a growing list of countries, including Malaysia and Finland, absorbing hyperscale AI capacity as US sites hit grid interconnection queues. Grid access and land that would take years to permit in the US or Europe can move faster here.

IREN’s pivot from Bitcoin (BTC) mining toward AI hosting, mirrored by peers like Hut 8, shows how compute-hungry AI labs are reshaping which infrastructure operators get built out first.

The spending pattern confirms the strategic shift regardless of how either company describes it publicly.

Watch whether Sharon AI’s 68,000-GPU order ships on the stated 2027 timeline given global GPU allocation constraints. Delivery schedules, not announcements, will determine whether this buildout hits its gigawatt target on time.

Read Next: PPI Is the Cheaper Risk Before the Sept. 16 Fed Rate Decision, say Experts