Gulf AI data centers re-engineer cooling to avoid water depletion amid compute thirst

The six Gulf Arab states, Saudi Arabia, the UAE, Qatar, Bahrain, Kuwait, and Oman, are racing to build massive computing infrastructure to establish the region as a major hub in the global AI market. This drive, underpinned by deploying financial surpluses and cheap energy into advanced compute capacity, is led by Saudi Arabia's Public Investment Fund, the kingdom's sovereign wealth fund, which has allocated more than $40 billion to AI-related projects including the Humain platform. Alongside it, the UAE has allocated $355 million to AI hardware in 2025, with projections reaching $892 million by 2029. Yet this compute expansion faces a geographic and climatic reality that places mounting pressure on freshwater resources.

Thousands of graphics processing units used to train and run advanced models generate intense heat inside servers, demanding continuous cooling in an environment where summer temperatures exceed 45 degrees Celsius in Saudi Arabia and the UAE with little relief at night. Estimates from the Middle East Institute indicate that the UAE's AI sector alone could consume around 61 billion liters of water annually by 2030, while demand for data center power in Saudi Arabia grows at 29% per year. These requirements come as the World Bank estimates that worsening water insecurity driven by climate change could cost the Middle East up to 14% of its gross domestic product by 2050.

Cooling compute chips in extreme heat can no longer rely on traditional evaporative towers and freshwaterShaolei Ren, an associate professor at the University of California, Riverside, explains that a technical distinction between water withdrawal and water consumption is essential, as consumed water is permanently lost through evaporation. He projects that AI globally will withdraw between 4.2 and 6.6 billion cubic meters of water in 2027, of which 380 to 600 million cubic meters will be consumed through evaporation. With generative AI workloads rising, Ren argues that governments must plan power, water, and data center systems holistically rather than treating each facility as an isolated project.

To address these constraints, infrastructure operators are turning to alternative cooling technologies, such as direct-to-chip liquid cooling, immersion systems, and closed loops. Elisabetta Baronio, head of sustainability at UAE-based Khazna Data Centers, explains that 90% of the company's portfolio relies on air cooling, while the remaining three facilities use treated sewage effluent rather than potable water, alongside closed-loop liquid cooling for dense workloads to circulate coolant without continuous water consumption. Adnan Masood, chief AI architect at digital transformation firm UST, emphasizes that resource scarcity is driving innovation, calling for direct-to-chip liquid cooling and requiring developers to be transparent about water volumes, power sources, and heat discharge methods.

Building responsible compute centers extends beyond cooling engineering to securing supply chain sustainability and critical infrastructure stabilityMohammed Soliman, a researcher at the Middle East Institute, notes that shifting to liquid cooling and closed-loop systems could reduce freshwater consumption by 90% or more. He stresses that the defining challenge of the next decade is whether Gulf states can move beyond building infrastructure to becoming native AI economies, measured by model quality, technical talent, and the resilience of their operational and policy frameworks.