Anthropic is planting its biggest Australian flag yet, securing capacity at a planned Queensland data center campus that could rival the world’s largest computing sites.
Anthropic has signed its first Australian data center lease for the proposed Western Downs Digital Park near Dalby, Queensland, according to Reuters, which cited two people familiar with the agreement. The campus is planned to reach 2.16 gigawatts of capacity and could begin coming online in 2027.
The project is being developed by Singapore-based Zerra DC on about 725 hectares of farmland. Planning documents call for four data center buildings, each with roughly 36,000 square meters of floor space, according to The Straits Times.
The financial terms and length of Anthropic’s lease have not been disclosed. The agreement also requires approval from Australia’s Foreign Investment Review Board and local council approval.
Built for Claude inference
Anthropic plans to use the facility for inference, the computing required to respond to requests from users of its Claude AI models, rather than training new models, Reuters reported. That distinction matters because inference capacity needs to remain close to users and available continuously as AI services grow.
The Queensland site could therefore give Anthropic a substantial pool of computing capacity in the region without making it a primary training hub. The project is also designed to reduce pressure on local water supplies. Zerra’s plans call for primarily air-cooled technology, while Reuters reported that the site would use a closed-loop system.
The project arrives as Australia debates how much electricity and land its growing data center industry should consume. Zerra plans to connect the facility to the existing grid and use renewable power purchase agreements, while Anthropic is expected to cover the cost of its grid connection and any required upgrades. The site is also near gas-fired power stations and renewable energy projects.
That has created a tension between Queensland’s push to attract AI infrastructure and concerns about how the additional electricity demand will be supplied.
Queensland Premier David Crisafulli called the agreement a “major win” that would create jobs and strengthen the state’s electricity supply, according to ABC. Environmental groups have raised a different concern. The Climate Council said the project could produce millions of tonnes of climate pollution if powered by emissions-intensive electricity, while Greenpeace criticized the possibility of gas being used during the project’s early stages.
More must-read AI coverage
- SS&C Intralinks DealCentre AI vs. Datasite: Which platform is built for the future of dealmaking?
- SS&C Intralinks FundCentre AI vs. Juniper Square: Which platform better supports modern private markets fund managers?
- Why Data, Not Models, Determines AI Success
- The Rise of the AI-Native Factory: How Physical AI Is Transforming Manufacturing
What the deal signals for AI infrastructure
The Queensland project shows that the AI infrastructure race is increasingly becoming a competition for land, electricity and grid access, not simply computing hardware.
For Anthropic, reserving capacity years before the facility is operational provides a path to expand Claude’s regional inference footprint as demand grows. For Australia, the project tests whether large AI investments can deliver jobs and infrastructure without placing additional pressure on electricity supplies, rural communities and the country’s renewable-energy goals.
For businesses using Claude and other generative AI services, where that computing capacity is built matters too. More regional inference infrastructure can help providers reduce latency, add capacity and build greater geographic resilience as demand grows.
Regional infrastructure can also become important for organizations weighing data residency, regulatory, and business continuity requirements, although Anthropic has not announced which customer-facing regional deployment options the Queensland facility will support.
The project also highlights a less visible constraint behind the rapid expansion of enterprise AI: computing capacity ultimately depends on available power. Companies may be able to buy more GPUs and deploy more AI systems, but those investments still require data centers, grid connections and enough electricity to keep the infrastructure running.
That means enterprise technology leaders evaluating AI providers may increasingly need to look beyond model performance and pricing. Infrastructure capacity, service availability, resilience and the geographic footprint of AI providers could become more important considerations as organizations move AI from experiments into business-critical systems.
The timing adds another wrinkle. Australia introduced national expectations for data center developers in March 2026, covering areas including energy and water use, and the federal government is pursuing stronger requirements around power supply, grid connections and water efficiency.
The Anthropic agreement also still faces regulatory and local approvals. If the Western Downs project moves forward, it could become an early test of whether the infrastructure supporting the generative AI boom can expand as quickly as demand without outrunning the power systems underneath it.
Read next: Anthropic previously explored up to 5 GW of AI data center capacity in Australia, raising questions about whether the country’s power grid could keep pace.