Anthropic Bolsters AI Computing Footprint by Securing First Australian Data Centre Deal

Deep News
3 hours ago

Following its recent call for AI firms to temper the pace of model advancement, Anthropic has pivoted to expanding its computing infrastructure, securing its inaugural data centre lease in Australia.

According to sources familiar with the matter, cited by Reuters on September 16, Anthropic has entered into a lease agreement with Singapore-based data centre developer Zerra DC. The project, situated roughly 250 kilometres west of Brisbane, is designed for a total capacity of 2.16 gigawatts, a scale approaching that of the world's largest existing data centres.

This transaction signals that Anthropic, following in the footsteps of OpenAI, is now initiating AI infrastructure development in Australia. The abundance of renewable energy, relatively affordable land costs, and a market with substantial room for growth are increasingly drawing AI enterprises to incorporate Australia into their computing capacity strategies. Economists project that by 2030, total investment in Australian data centres could reach A$150 billion.

Notably, Anthropic CEO Dario Amodei publicly urged AI firms last weekend to decelerate the enhancement of model capabilities, citing concerns over the potential for misuse of artificial intelligence. This perspective garnered public endorsement from xAI CEO Elon Musk and OpenAI CEO Sam Altman. The recent Australian data centre venture has nonetheless reignited scrutiny of Anthropic's expanding computing footprint.

2.16-Gigawatt Facility in Queensland, Primarily for AI Inference

According to Reuters' sources, Zerra DC will procure renewable energy power purchase agreements and bear the costs of grid connectivity, without transferring these expenses to existing users. The project is slated to commence phased operations in 2027.

The facility will employ a closed-loop air-cooling system, which, compared to conventional liquid cooling methods, diminishes the consumption of clean water resources for server heat dissipation. The report, citing insiders, indicated that the data centre will be chiefly dedicated to AI inference—providing real-time computation for user requests post-model training—rather than for model training purposes.

Australia is emerging as a fresh competitive arena for AI enterprises vying for computational resources. OpenAI has previously inked a power purchase agreement for a Sydney data centre. With major markets in the United States and Asia encountering constraints on land, electricity, and grid interconnection capacity, Australia's renewable energy conditions and market expansion potential are drawing heightened interest from AI infrastructure investors.

Nevertheless, the extensive growth of data centres also raises concerns over land use, water consumption, and electricity demand. The Australian government plans to impose stricter limits on data centre resource utilisation beginning in 2027, even as it continues to court foreign investment in this domain.

This transaction remains subject to approval from Australia's Foreign Investment Review Board.

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Most Discussed

  1. 1
     
     
     
     
  2. 2
     
     
     
     
  3. 3
     
     
     
     
  4. 4
     
     
     
     
  5. 5
     
     
     
     
  6. 6
     
     
     
     
  7. 7
     
     
     
     
  8. 8
     
     
     
     
  9. 9
     
     
     
     
  10. 10