OpenAI-Backed Australian Data Center Drops Recycled Water Cooling Plan, Raising Sustainability Conce

BY COMFORT OGBONNA

Developers behind a massive artificial intelligence data center planned for western Sydney have abandoned plans to cool the facility with recycled water, opting instead for a water-free cooling system that relies more heavily on electricity. The change has raised fresh questions about the environmental impact of rapidly expanding AI infrastructure and the ability of existing utilities to support the industry’s growing demands.

The proposed S7 data center, a 612-megawatt facility that would rank among the largest in the world, was initially designed to use treated wastewater to cool thousands of servers. The approach would have reduced reliance on drinking water while also improving overall energy efficiency.

However, the project’s developers said the plan became impractical because the infrastructure needed to transport recycled water was not available. Building a dedicated pipeline would have required planning approvals that were not secured, forcing the company to adopt a different cooling strategy.

Craig Scroggie, chief executive of Australian data center operator NEXTDC, said recycled water remains an attractive option for improving sustainability where the necessary infrastructure already exists.

He explained that recycled water had been evaluated during the project’s early planning stages, but western Sydney currently lacks the supporting infrastructure required to deliver it to the site.

Instead, the facility will use a liquid cooling system that circulates coolant around computer processors to absorb heat before it is removed through electrically powered fans. Industry experts say the approach eliminates the need for water during cooling but generally increases electricity consumption compared with recycled-water systems.

The facility is part of a major artificial intelligence expansion in Australia, where OpenAI plans to help develop, operate and purchase computing capacity from the project. Once completed, the center is expected to provide the enormous processing power needed to support advanced AI models and cloud computing services.

The decision highlights a growing challenge facing governments around the world as they compete to attract billions of dollars in AI infrastructure investment while balancing environmental concerns. Analysts expect spending on AI-related infrastructure across Asia to exceed $1 trillion by the end of the decade, fueling a surge in demand for new data centers.

Cooling systems have become one of the most significant design considerations for modern data centers because they consume substantial amounts of energy and, in many cases, large volumes of water. Operators generally choose between water-based cooling methods, which improve energy efficiency but require significant water supplies, and air- or liquid-based systems that reduce water consumption but typically increase electricity demand.

The issue is particularly sensitive in Australia, the world’s driest inhabited continent, where recurring droughts have led to periodic water restrictions for residents and businesses. NEXTDC has emphasized that the S7 facility will not rely on potable drinking water for its operations despite abandoning the recycled-water proposal.

Electricity supply is also becoming a growing concern. Sydney’s transmission network has experienced increasing pressure as numerous data center developers seek connections to support expanding AI and cloud computing operations.

Transgrid, the operator responsible for Sydney’s high-voltage electricity network, has warned that available transmission capacity for large new data centers is limited without significant upgrades to the broader power grid. The company has indicated that developers seeking new large-scale connections will likely need to help fund network expansions.

Industry analysts say the change in cooling strategy underscores a broader mismatch between the rapid pace of AI infrastructure development and the slower rollout of supporting public infrastructure, including recycled-water networks and electricity transmission projects.

Sarvaesh Mohan, an analyst specializing in data center infrastructure, said the absence of recycled-water delivery systems illustrates how utility development has struggled to keep pace with the industry’s explosive growth. He added that if the new cooling system consumes more electricity, the developer will face additional pressure to demonstrate how it plans to maintain high sustainability standards.

Neither NEXTDC nor OpenAI disclosed whether the revised cooling system will increase the facility’s overall energy consumption. OpenAI referred questions regarding the cooling design to its Australian development partner.

Australia’s federal government recently announced plans to introduce new rules governing the environmental impact of artificial intelligence infrastructure. The proposed regulations are expected to establish limits on energy and water consumption while creating broader standards for AI development and operations.

Several Australian states have already begun imposing stricter requirements on large data center proposals. Planning authorities in New South Wales now require developers to disclose projected water usage and commit to meeting specific energy-efficiency targets before receiving approval.

Planning documents show the OpenAI-supported S7 project is among the developments subject to the updated state requirements. The project must satisfy a maximum Power Usage Effectiveness (PUE) target of 1.3—a metric widely used across the industry to measure how efficiently a data center uses electricity beyond the computing equipment itself. Lower PUE values indicate greater operational efficiency.

NEXTDC said it has successfully deployed the water-free liquid cooling technology at other facilities, although on a much smaller scale. In its most recent sustainability reporting, the company said the system had been used to cool approximately 50 megawatts of computing capacity, far below the 612 megawatts planned for the S7 project.

Environmental advocates and policymakers have urged governments to strengthen regulations as data center construction accelerates. Some lawmakers argue future rules should require greater transparency on water use, renewable energy sourcing, computing capacity reserved for domestic needs and land-use planning to minimize impacts on surrounding communities.

The S7 project illustrates the increasingly difficult balance governments and technology companies must strike as artificial intelligence fuels unprecedented demand for computing infrastructure. While advances in cooling technology may reduce water consumption, they also risk placing greater strain on electricity networks, highlighting the complex trade-offs involved in building the next generation of AI-powered data centers.