Anthropic has taken a major step in its global infrastructure expansion by signing its first data-centre agreement in Australia. The deal gives the AI company access to a planned campus with a stated capacity of 2.16 gigawatts — a scale that immediately shows how intense the race for AI compute has become.
Reuters reported that the site is located around 250 kilometres from Brisbane and is expected to begin operations in 2027, subject to approvals. The project is being developed by Singapore-based Zerra DC and is designed to use renewable energy and a closed-loop air-cooled system intended to reduce water use.
The location is strategically interesting. Australia combines large land availability, renewable-energy potential and a political push to attract more digital infrastructure. For AI companies that need massive amounts of electricity and physical capacity, those factors can be as important as access to customers or software talent.
Anthropic plans to use the Australian capacity for inference rather than model training. That distinction matters because inference is the computing work required when users actually interact with AI systems at scale. As usage grows, serving those models can become an enormous infrastructure challenge of its own.
The deal also fits a broader investment boom. Australia has been trying to position itself as a major destination for next-generation data centres, while governments are simultaneously debating power demand, water use and how AI infrastructure should be regulated.
For Anthropic, the move is therefore about more than one lease. It is a signal that the company expects demand for AI services to keep expanding and that securing physical computing capacity has become one of the most important competitive advantages in the industry.
