A cooling question - AI infrastructure and the Gulf's water-energy challenge
Technology companies and governments are committing over $30 billion to AI data centre infrastructure across the GCC between now and 2030, and the pace of that commitment is accelerating. The UAE's Stargate mega-project, announced in 2025, aims to deploy one gigawatt of AI data centre capacity. Microsoft's expansion through a major UAE data centre operator, announced in November 2025, involves $7.9 billion of investment over 2026 to 2029. Saudi Arabia, through its national AI champion HUMAIN, is pursuing comparable scale. This is infrastructure investment of a kind the region has not seen before. And like all infrastructure, it sits inside physical systems: utility grids, water supply networks, and cooling chains that share their capacity with the cities and industries they serve. The sustainability question behind these announcements is not theoretical. It is already active.
The scale of the demand
Global electricity consumption from data centres is projected to double by 2030, reaching around 945 TWh according to the IEA. AI is the primary driver. Electricity consumption in accelerated servers, the hardware running AI workloads, is projected to grow by 30% annually. The character of the demand is shifting, not just the volume. AI workloads require rack densities of 40 to 100 kW or more, compared with 5 to 10 kW for conventional compute. Higher density generates higher heat and demands more intensive cooling decisions.
The GCC is forecast to absorb $5 to $7 billion of AI data centre investment in 2026 alone. The UAE data centre market, valued at $1.26 billion in 2024, is projected to approach $3.33 billion by 2030. By the end of the decade, the UAE alone is expected to use 61 billion litres of water annually for data centre cooling, according to research firm Mordor Intelligence.
A different operating context
In most markets, data centre sustainability is primarily a carbon and energy question. In the Gulf, it is inseparable from water.
Cooling already defines a disproportionate share of the region's electricity demand. In Dubai, the Regulatory and Supervisory Bureau estimated that electrical demand for cooling reached 24,520 GWh in 2023: 48% of the total load on the DEWA grid. AI data centres are arriving on a grid where summer heat already defines the peak.
Water is not a freely available local input. In the UAE, it is an industrial product with its own embedded energy cost. Legacy desalination methods can require 13 to 25 kWh per cubic metre. More efficient reverse osmosis processes operate in the low single digits, but even those figures embed a significant upstream energy burden. A cooling strategy that performs well at the facility level can therefore be shifting significant pressure onto water production systems that conventional site metrics, like power usage effectiveness (PUE), do not always capture.
A facility that achieves a strong PUE through water-intensive cooling may be increasing desalination demand. One that reduces water use through dry or air-based approaches may raise direct energy consumption at the site. Understanding how those choices interact, rather than optimising each in isolation, is what a system-level assessment of AI infrastructure actually requires.
Why timing matters
The decisions that will shape AI data centre sustainability across the Gulf are being made now, before infrastructure is built and procurement choices are fixed. Cooling architecture cannot easily be revisited once a facility design is committed. The assumptions embedded at design stage shape performance across the asset's entire operating life.
This is the case for earlier and more rigorous analysis of water, energy, and cooling trade-offs: not as a retrospective sustainability review, but as front-end decision support applied before capital is locked in. The frameworks for making these assessments systematically under Gulf conditions are still developing. Operator practice is advancing quickly. The analytical basis for comparing options in this specific operating context is less established.
Getting this analysis done at the scoping stage reduces avoidable pressure on shared utility systems and strengthens the long-term resilience of what the region is building. The scale of investment being committed makes the cost of not doing it proportionate.
TCC's work in this area
TCC works with organisations to develop sustainability frameworks for AI and data infrastructure decisions in the GCC, translating technical and sustainability complexity into decision-ready outputs for planners, operators, and policymakers. If this work is relevant to programmes you are developing, we would welcome a conversation.
