Introduction

Gulf states are embarking on an unprecedented digital expansion focused on artificial intelligence (AI). Reflecting this push, the region’s data center capacity is projected to grow significantly in the coming years, with tech giants and local players alike investing in both small and hyperscale facilities.¹ Yet this infrastructure boom is unfolding in a challenging physical environment; all six Gulf Cooperation Council (GCC) states are arid and extremely hot. Given the massive volumes of water demanded by AI data centers, the Gulf’s technology ambitions may be impeded by these environmental constraints. This challenge, while surmountable, has received less attention than it deserves as media headlines focus on blockbuster AI deals.

Additional power and desalination infrastructure has been proposed as a solution to this challenge. Aside from the high costs of constructing major desalination and wastewater treatment infrastructure (for both production and distribution), these plants also require large energy inputs, and the environmental side effects are significant. In other words, water is not a freely available resource; it is produced through energy-intensive processes that carry considerable ecological footprints. The Gulf cannot scale AI sustainably unless it treats water efficiency and summer peak resilience as first-order design constraints, not downstream externalities.

To ensure the AI sector’s long-term viability, Gulf regulators must urgently pivot from a sole focus on energy to a holistic water-energy nexus approach. This policy brief outlines four priority actions to achieve this: (1) invest in strategic seasonal storage specifically earmarked for digital infrastructure; (2) implement a dual water and energy efficiency standard; (3) mandate non-potable water use for all new data centers; and (4) invest in promising technologies and solutions that reduce the burden on water resources. Together, these solutions form a summer-neutral approach to AI infrastructure that flattens seasonal impacts.

Infrastructure Mismatch: Why Summer Breaks the Model?

The extreme summer heat already presents an annual stress test for Gulf infrastructure. Since air-conditioning demand soars in the summer, electricity consumption peaks, reaching levels twice the winter peaks in some regions, as shown in Figure 1 below. Water usage also surges, though this is mitigated by the deployment of alternative sources such as treated sewage effluent (TSE). This seasonality means that the margin between supply and demand is thinnest in summer, exactly when data centers face their greatest cooling challenge, and when traditional cooling methods struggle most under extreme conditions.²

Source: Saudi Electricity Regulatory Authority, Annual Statistical Booklet for Electricity 2024, accessed January 29, 2026, https://sera.gov.sa/en/knowledge-center/data-and-statistics/data-and-statistics-categories/data-and-information/weekly-peak-load.

Into this reality enter the data centers: massive server farms that underpin cloud and AI services. Cooling thousands of servers in the hot and humid coastal Gulf climate requires significant water consumption. By 2030, the UAE’s AI sector alone may require approximately 61 billion liters of water per year³ In Saudi Arabia, data center power demand is expected to grow at a 29% compound annual growth rate, which will drive substantially higher water demand as well.⁴ In Qatar, the Qatar Investment Authority’s subsidiary, Qai, signed an agreement with Brookfield Asset Management to construct AI infrastructure, placing a significant burden on the country’s water and electricity grid.⁵

Most large data centers dissipate heat via chilled-water systems and cooling towers, or through direct-air cooling aided by evaporative coolers. Both approaches rely heavily on ambient conditions, although chilled-water systems tolerate higher temperatures better. Still, both systems are regularly strained by Gulf summer temperatures, which average around 40 °C al cities and little nighttime relief.⁶ Cooling infrastructure must work overtime in these conditions to keep server inlet temperatures within safe limits.

Compounding the challenge, many Gulf power and water systems are co-generation based; electricity and desalinated water are typically co-produced in the same plants. This linkage is efficient but creates rigidity: the ability to produce additional water in summer can be constrained by power plant capacity, and vice versa.⁷ Although the growing role of reverse osmosis (RO) is breaking this link, it will take at least a generation before most co-generation plants are phased out across the region. In the meantime, when power demand peaks during winter months, Gulf countries may be forced to curtail the co-produced water or rely on inefficient backup boilers to meet demand surges.

During summer, both water and power systems run at maximum capacity. In this context, a large new data center acts as a constant baseload consumer. Conventional growth models risk breaking down under Gulf peak conditions, forcing disproportionate investment in infrastructure that may sit underutilized for much of the year. Additionally, purple pipe networks for the distribution of TSE remain limited, inhibiting its adoption for cooling. There is currently no overarching strategy to leverage  TSE resources in support of data center growth in the Gulf.

The practical implication is clear: Global AI and cloud data center operators cannot simply transplant their standard expansion playbook into Riyadh, Dubai, or Doha without significant adaptation. Worse yet, the cooling designs currently prevalent worldwide — which assumes plentiful water and a modest diurnal temperature swing —  will falter where water is costly to desalinate and nights offer little relief from the heat. Furthermore, while the Gulf’s infrastructure managers are adept at handling seasonal peaks, they need to adapt to handle servers that run 24/7. ⁸ Besides, shifting workloads geographically to other regions during the summer will not suffice if the Gulf aims to be a net exporter of digital services. A successful Gulf AI industry therefore requires flattening data centers’ seasonal load profile (making it summer-neutral) and rethinking water sources, storage, and cooling technology for Gulf-specific conditions.

What the GCC Is Doing so Far, and Where Risks Remain

Gulf governments and industries are treating the water–energy–technology nexus as a strategic issue. The creation of a GCC water security task force in March 2025 is a clear signal that the region is framing water, like energy security, as a shared regional interest.⁹ One of the task force’s early initiatives — a solar-powered desalination pilot in Oman — links additional water capacity to lower-carbon power instead of defaulting to gas-fired plants. The emirate of Dubai will also soon host the world’s most energy-efficient desalination plant, powered entirely by solar energy.¹⁰ This signals an emerging preference for coupling new water infrastructure with cleaner energy, rather than expanding conventional desalination on a business-as-usual basis. However, there are solutions other than new desalination infrastructure that are inexpensive and can have just as important an impact.

At the urban scale, the region’s eco-cities are serving as testbeds for integrated systems with direct relevance to data centers. For instance, NEOM in Saudi Arabia, Green Mubarak Al-Kabeer City in Kuwait, and Masdar City in the UAE have launched a joint platform to exchange technologies and standards on energy and water efficiency.¹¹ The focus on district cooling, thermal storage, and potential reuse of waste heat for other industrial processes (including desalination) creates a practical basis for low-water, high-efficiency designs that can later be applied to large-scale digital infrastructure. This kind of structured, cross-border learning reduces duplication of effort and increases the likelihood that successful pilots can be replicated across the GCC.

In that same spirit, several Gulf-based operators are already piloting next-generation cooling systems suited to extreme heat. Companies such as Khazna, DataVolt, and Alfanar have tested or deployed liquid cooling and other high-efficiency solutions that can cut water use for cooling by up to 90%while also lowering electricity demand. Global cloud providers exploring the Gulf have signaled openness to options such as seawater-based cooling at coastal sites.¹² These projects confirm that technically viable alternatives to conventional evaporative cooling exist and can function in Gulf conditions, even if they are not yet the norm.

The region’s rapid expansion of renewable energy supports these efforts by shrinking the indirect water footprint of digital growth. Saudi Arabia’s proposed 1.5 GW data center in NEOM, designed to run entirely on renewables, illustrates how new AI-enabling infrastructure can be paired with low-water, low-carbon power from the outset.¹³ Across the GCC, tens of billions of dollars in announced solar and wind projects by the mid-2020s will, if delivered, likely reduce the water intensity of electricity supply, since photovoltaic and wind power consume negligible operational water compared with gas- or oil-fired generation. Over time, this benefits all major electricity users, including data centers.

Perhaps more importantly, Gulf states are beginning to reconsider the perception of water and electricity as limitless, subsidized commodities. Tariff reforms and awareness campaigns across the region aim to curb waste, particularly during summer peaks.¹⁴ While fragmented and modest, these measures create a foundation for integrating new large loads — such as AI data centers — without eroding system reliability. While politically challenging, further reforming industrial water tariffs to better reflect the true cost of production, particularly for desalinated potable water, would create a powerful economic incentive for data center operators to adopt alternative sources like TSE. However, significant risks remain that could undermine progress. For example, efforts to utilize treated wastewater are constrained by the limited availability of purple-pipe infrastructure to deliver this water, as well as concerns about reliability and quality.

Summer-Neutral Data Centers: A Gulf Framework

To reconcile AI infrastructure with the Gulf’s environmental and resource constraints, an integrated approach is needed through summer-neutral operations that can be implemented using immediately feasible solutions while medium-to-long term solutions are developed. The success of these measures will depend on whether Gulf regulators diligently update building codes and data center compliance schemes to favor low-water designs. Some of the immediately feasible solutions include non-potable water sourcing, smart scheduling, thermal storage, and infrastructure adaptation; medium-to-long term solutions focus primarily on the development of new cooling technologies. The key recommendations are as follows:

1. Invest in strategic seasonal storage that is specifically earmarked for digital infrastructure

Summer-neutral operations at data centers mean placing no disproportionate strain on water and power systems during peak summer conditions relative to milder months. Achieving this does not mean a facility stops operating in summer; rather, it involves smart design and scheduling to flatten the seasonal load profile. For instance, Gulf data centers can employ thermal energy storage to shift some cooling effort to off-peak hours. Large, insulated tanks of chilled water or ice can be produced at night – when air temperatures are relatively lower and power demand is less – and then used for cooling during the hottest daytime hours.

Seasonal water storage is equally powerful. During winter months, when demand is lower, excess desalinated water can be stored in deep aquifers or insulated surface reservoirs and drawn down in summer. The UAE has pioneered this approach with its Liwa Aquifer Storage and Recovery project, injecting surplus desalinated water into the ground for use as a strategic reserve during emergencies.¹⁵ Expanding such schemes from emergency reserves to seasonal supply management tools could enable data centers to effectively bank water in winter for summer operations.

  2. Implement a dual water and energy efficiency standard for data centers, complemented by climate-conscious demand management.

A regulatory lag exists on water-specific standards for data centers operating in the Gulf’s climate. Global green building certifications include general water-efficiency credits, but no GCC-wide framework sets mandatory limits on water-use effectiveness (WUE), nor is there a policy requiring designs adapted to extreme heat, humidity, and water scarcity. If standards arrive only after construction, governments may find themselves locked into water-intensive facilities that are costly to retrofit. Early and clear guidance from entities such as the Saudi Water Authority, Kahramaa in Qatar, or the Emirates Water and Energy Company on the maximum allowable WUE, the mandatory use of recycled or non-potable water, and minimum performance standards for cooling systems would help align private investment with public sustainability goals.

Demand-side management and smart scheduling are complementary tools. For cloud providers with global footprints, it is feasible to schedule specific flexible tasks — such as non-urgent AI model training or batch processing — during the cooler seasons, when spare capacity is available. Gulf-based data centers could be programmed to offload some workloads to overseas sites during extreme weather without undermining local service delivery. Conversely, during cooler months, they could absorb more. While latency-sensitive services must remain local year-round, seasonal load shifting for non-critical processes can meaningfully alleviate peak stress.

Coordinating with grid operators will also be essential. Data centers can enroll in demand response programs (e.g., temporarily raising thermostat setpoints to reduce cooling power during critical peaks) and in return benefit from financial incentives. Although data centers run continuously, they can be more flexible than commonly assumed, especially with AI-driven management that predicts and responds to grid conditions.

3. Implement and mandate non-potable water use for all new data centers.

Gulf regulators should institute a non-potable-first policy as soon as possible, ensuring that data centers are designed from day one to use recycled wastewater, TSE, brackish groundwater, or lightly treated seawater for all cooling needs. The region already has extensive experience in utilizing non-potable water for cooling, and this principle can be enforced through regulation. Early adoption is critical: Qatar is already reaping the benefits of a long-standing policy to incentivize the construction of district cooling facilities that are designed to use non-potable water, as shown in Figure 2 below.

Gulf cities are already investing in wastewater treatment capacity; the logical next step is building TSE distribution networks (purple pipe infrastructure) to supply industrial users. Data centers clustered in dedicated technology parks or industrial zones present a ready opportunity – those zones can be plumbed with their own non-potable water loop. Most Gulf countries have set ambitious targets to reuse the majority of TSE; these targets should be expanded to explicitly include the growing data center sector.

4. Invest in promising technologies and solutions that reduce the burden on water resources

Adopting circular water solutions is essential to optimize data center water use. Closed-loop systems that recycle wastewater or capture rainwater can cut freshwater use by an estimated 50–70%.¹⁶ Embedding these circular water principles is immediately feasible and therefore critical if data centers are to support the next wave of technological growth in an increasingly water-stressed region.

Direct seawater cooling is also an option, though living organisms and pollutants in seawater present challenges requiring specialized distribution systems. Furthermore, Gulf seawater can be quite warm in the summer, limiting its utility for cooling –  and the shallowness of the Gulf means that the usual solution to this challenge of withdrawing seawater from depth is not available. That said, Saudi Arabia and Oman could test this option at facilities located near the Red and Arabian Seas, which are considerably deeper. Testing in the Caspian Sea demonstrated that using seawater from depths of 700 meters or more to cool data centers can reduce electricity consumption (and emissions) for cooling by 45%.¹⁷ Another interesting proposition currently being tested by both China and Microsoft involves submerging AI data centers underwater.¹⁸

The broader advantage of these alternatives is eliminating freshwater use while reducing electricity demand for chillers. Similarly, district cooling systems using seawater or TSE could serve multiple data centers simultaneously, achieving economies of scale and enabling centralized environmental controls (like better brine dispersion). The non-potable-first approach also extends to humidity control: even the water used for adiabatic cooling or humidification in IT facilities should come from non-potable sources wherever possible. Potable water – produced through expensive desalination or drawn from precious aquifers – should be reserved for human and agricultural use, not industrial cooling.

Gulf nations should leapfrog to ultra-efficient cooling technologies as a matter of policy. Liquid cooling is a prime candidate: by directly circulating coolant to servers, heat is removed far more efficiently than air-based systems, effectively eliminating the need for evaporative cooling towers. Major data center operators like Amazon are already deploying this technology, and Gulf regulators can support local players in doing the same through incentives such as grants or tax relief for facilities meeting a low WUE threshold.¹⁹ Other promising technologies – including immersion cooling and absorption cooling – could benefit from targeted Gulf investments to accelerate deployment.

The concept of “regenerative data centers” is also gaining traction and may be feasible in the medium-to-long term. These facilities would use their waste heat to desalinate water, thus becoming net producers of freshwater.²⁰ In the Gulf context, this could mean pairing a data center with a thermal desalination unit powered by its excess heat, producing fresh water as a byproduct and creating a closed loop in which increased operations yield increased water production. While still experimental, such ideas underscore a mindset shift: treating data centers not as isolated loads but as integrated components of the water-energy-technology nexus

Gulf governments can also look towards long-term solutions by encouraging R&D and pilot projects in experimental technologies. Manufacturers such as TSMC, chip designers such as Nvidia, and labs such as NASA’s Glenn Research Center are working on server racks and semiconductors more tolerant of higher temperatures, including through the use of materials like silicon carbide in the manufacturing of chips.²¹ Gulf sovereign wealth funds can leverage their investments in key nodes of the AI supply chain to accelerate progress in this area. These technical tweaks all feed into the broader goal: minimizing water withdrawal, reducing the energy burden from cooling (especially in summer), and decoupling from freshwater dependence.

Conclusion

The region’s window of opportunity for building sustainable AI data centers is now. To avoid the burden of inefficient facilities and costly retrofitting, Gulf governments should implement policies that mandate stringent efficiency standards, non-potable water use for cooling, and sufficient storage capacity to handle summer peaks. This will eliminate the need to build out costly desalination infrastructure. The benefits of this approach will be felt throughout the economy, as a successful rollout of these policies for the AI sector will demonstrate that innovation and sustainability are essential to sustainable economic diversification in the long run.

Endnotes

¹ Chris Hamill-Stewart, “What’s happening with data centres in the Gulf?” Arabian Gulf Business Insight, October 10, 2025, https://www.agbi.com/tech/2025/10/whats-happening-with-data-centres-in-the-gulf/.

² Güney Yıldız, “Gulf Faces Blackout Threats Unless Grids Upgrade Fast,” Forbes, July 2, 2025, https://www.forbes.com/sites/guneyyildiz/2025/07/02/gulf-ai-growth-tests-power-grid-capacity-and-renewable-energy-plans/.

³ Divisha Bhat, “Big Tech is building AI in the desert. The water may not last,” Rest of World, August 4, 2025, https://restofworld.org/2025/gulf-ai-water-crisis/.

⁴ Mai Barakat, “The Saudi Arabia data center market: A catalyst for economic innovation,” S&P Global, December 2, 2025, https://www.spglobal.com/en/research-insights/special-reports/look-forward/data-center-frontiers/saudi-arabia-data-center-market.

⁵ “Brookfield and Qai Form $20 Billion Strategic Investment Partnership for AI Infrastructure,” Qatar Investment Authority, December 9, 2025, https://www.qia.qa/en/Newsroom/Pages/Brookfield-and-Qai-Form-$20-Billion-Strategic-Investment-Partnership-for-AI-Infrastructure.aspx.

⁶ To view average temperatures and humidity levels in Gulf cities, see: https://weatherspark.com/; for additional information, see: Colin Raymond, Tom Matthews, and Cascade Tuholske “Evening humid-heat maxima near the southern Persian/Arabian Gulf,” Communications Earth & Environment 5, (2024), https://doi.org/10.1038/s43247-024-01763-3; Aziz El Massassi, “’We Cannot Work’ — Why Gulf Summer Feels Even Hotter Than Usual,” Al Monitor, July 13, 2023, https://www.al-monitor.com/originals/2023/07/we-cannot-work-why-gulf-summer-feels-even-hotter-usual.

⁷ Laura Parmigiani, Water and Energy in the GCC: Securing Scarce Water in Oil-Rich Countries, Note de l’Ifri (Paris: Institut français des relations internationales, September 2015), 1, https://www.ifri.org/sites/default/files/migrated_files/documents/atoms/files/water_energy_gcc_parmigiani.pdf.

⁸ For instance, utilities stagger maintenance to maximize summer availability, and some have discussed seasonal pricing to curb peak usage.

⁹ Gulf Research Center, Gulf Research Center Annual Strategic Survey 2025, (Jeddah: Gulf Research Center, 2025), 47-48, https://www.grc.net/documents/6882330f58bd3AnnualStrategicFinal20252.pdf#:~:text=Eco,will%20allow%20them%20to%20pool.

¹⁰ “Veolia wins $320 million water technology contract for world’s most energy-efficient desalination plant, enhancing water security in UAE,” Veolia, May 14, 2024, https://www.veolia.com/en/our-media/press-releases/veolia-contract-most-energy-efficient-desalination-plant-uae.

¹¹ Gulf Research Center, Gulf Research Center Annual Strategic Survey 2025, 48.

¹² Khaled Al Khawaldeh, “How Gulf states can develop data centers without straining scarce water resources,” Arab News, August 21, 2025, https://arab.news/89dpy.

¹³ Al Khawaldeh, “How Gulf states can develop data centers.”

¹⁴ Saroj Kumar Jha, “Advancing water and energy security in Gulf nations,” World Bank Blogs,  October 30, 2023, https://blogs.worldbank.org/en/arabvoices/advancing-water-and-energy-security-gulf-nations; Mohamed Abdelraouf, “Water Issues in the GCC Countries: Status, Challenges, and Solutions,” Gulf Research Center, October 20, 2024, https://www.grc.net/single-commentary/196

¹⁵ Mohamed A. Dawoud,“A strategic water reserve in Abu Dhabi,” Groundwater Solutions Initiative for Policy and Practice, accessed February 17, 2026, https://gripp.iwmi.org/natural-infrastructure/water-storage/a-strategic-water-reserve-in-abu-dhabi/.

¹⁶ Wesley Spindler, Luna Atamian Hahn-Petersen, and Sadaf Hosseini, “Why circular water solutions are key to sustainable data centres,” World Economic Forum, Nov 7, 2024, https://www.weforum.org/stories/2024/11/circular-water-solutions-sustainable-data-centres/.

¹⁷ Reza Mokhtari and Ahmad Arabkoohsar, “Feasibility study and multi-objective optimization of seawater cooling systems for data centers: A case study of Caspian Sea,” Sustainable Energy Technologies and Assessments 47, no. 1 (October 2021), https://doi.org/10.1016/j.seta.2021.101528.

¹⁸ You Xiaoying, “China Is Putting Data Centers in the Ocean to Keep Them Cool,” Scientific American, July 16, 2025, https://www.scientificamerican.com/article/china-powers-ai-boom-with-undersea-data-centers/; John Roach, “Microsoft finds underwater datacenters are reliable, practical and use energy sustainably,” Microsoft Source (blog), September 14, 2020, https://news.microsoft.com/source/features/sustainability/project-natick-underwater-datacenter/.

¹⁹ Alex Davies, “Next-gen AI demands smarter cooling tech. Here’s how AWS delivered it in just 11 months,” Amazon News, June 11, 2025, https://www.aboutamazon.com/news/aws/aws-liquid-cooling-data-centers.

²⁰ Hany Ghanem, “Facing scarcity, the Gulf’s ‘smart water’ future lies in desalination,” MENASource (blog, Atlantic Council), July 17, 2025, https://www.atlanticcouncil.org/blogs/menasource/gulf-water-scarcity-deslination/.

²¹ “TSMC x Nvidia : Breaking the Thermal Wall: How Advanced Cooling Is Powering the Future of Computing,” Semivision (blog), October 5, 2025, https://tspasemiconductor.substack.com/p/tsmc-x-nvidia-breaking-the-thermal ; Marc Hamilton, “Chill Factor: NVIDIA Blackwell Platform Boosts Water Efficiency by Over 300x,” Nvidia, April 22, 2025, https://blogs.nvidia.com/blog/blackwell-platform-water-efficiency-liquid-cooling-data-centers-ai-factories/; “Deep Dive into NVIDIA GB200 Liquid Cooling Plate Design,” Kenfa, June 15, 2025, https://www.kenfatech.com/gb200-liquid-cooling-plate-design/ ; “Silicon Carbide Electronics and Sensors,” NASA, last accessed March 3, 2026, https://www.nasa.gov/glenn/research/silicon-carbide-electronics-sensors/; “U-M awarded up to $7.5M to bring heat-tolerant semiconductors from lab to fab,” University of Michigan, February 20, 2025, https://news.umich.edu/u-m-awarded-up-to-7-5m-to-bring-heat-tolerant-semiconductors-from-lab-to-fab/.