AI at Scale: Can the Middle East Keep Its Cool?

Saudi Arabian Monetary Agency Data Centre, Source: Omrania

A recent PwC Middle East analysis argues that the region is rapidly becoming a global data-centre powerhouse, with capacity projected to triple from around 1 GW in 2025 to 3.3 GW within five years.

This growth is driven by three main forces:

• surging cloud and AI workloads,

• strong government-led digital strategies, and

• large-scale investment from hyperscalers and sovereign wealth funds.

GCC countries, especially the UAE and Saudi Arabia, sit at the centre of this transformation.

AI mega-projects across the region

The report highlights several high profile initiatives reshaping the regional landscape:

• In the UAE, Khazna Data Centres has announced a 100 MW AI facility in Ajman, with more capacity in the pipeline.

• In Saudi Arabia, the US$100 billion Transcendence AI Initiative, backed by the Public Investment Fund (PIF), aims to build AI infrastructure and an AI hub in partnership with global players such as Google.

• AWS has committed US$5.3 billion to develop new data centres in the Kingdom.

• Additional large projects are planned in Qatar, Kuwait, Oman and Bahrain, often through joint ventures between regional telecoms and global operators.

Collectively, these projects signal that the Middle East intends not just to consume AI services, but to host and export AI compute at scale.

Why the Middle East is attractive for AI data centres

PwC points to several structural advantages that make the region appealing for AI-focused data centres:

• Lower land costs: industrial land in Saudi Arabia can be an order of magnitude cheaper than in traditional hubs like Northern Virginia.

• Relatively low electricity tariffs: in Saudi Arabia and the UAE, typical industrial tariffs of US$0.05–0.06/kWh are well below many mature markets, which is critical for power-hungry AI clusters.

• Strategic connectivity: the region sits on major submarine cable routes (2Africa, SMW6, Blue/Raman, etc.), linking Asia, Europe and Africa and offering competitive connectivity pricing.

• Access to capital: sovereign wealth funds (PIF, Mubadala, QIA) and private equity firms are investing billions into digital infrastructure, including AI-optimised facilities.

For AI workloads, which can demand up to ten times more power than traditional data-centre activity, these cost and connectivity advantages are especially attractive.

Power, cooling and sustainability pressures

The same PwC report is clear that this boom comes with challenges:

• Power availability is becoming a constraint, particularly for AI-centric facilities drawing hundreds of megawatts.

• Governments are being pushed towards integrated planning, aligning new data centre projects with new generation assets, including solar, wind and nuclear.

• Cooling systems need to evolve: AI clusters require advanced thermal management, from high-efficiency air and liquid systems to immersion and hybrid approaches suitable for hot climates.

• Supply chains for GPUs, AI chips and high-end cooling gear are under pressure, with long lead times and rising costs.

• There’s a growing emphasis on renewable integration, with projects like Masdar’s 1 GW continuous renewable-energy facility and ACWA Power’s large solar portfolio seen as enablers for cleaner AI power.

In parallel, other regional analyses stress that operators are increasingly exploring liquid cooling, district cooling integration, and on-site solar plus storage to manage both cost and carbon in a harsh climate.

A system-level engineering challenge

For engineering and simulation teams, the Middle East’s AI data-centre boom is not just a story of more racks and bigger campuses. It is a system-level problem:

• At rack and hall level, designers must handle high power densities, airflow organisation, liquid or immersion cooling, and failure scenarios.

• At site level, there are trade-offs between air, liquid and district cooling; water availability; and heat-reuse options.

• At grid and generation level, planners need to understand how large, relatively inflexible AI loads interact with solar, wind, nuclear and storage assets across very hot climates.

High fidelity CFD and thermo-fluid simulation, coupled with techno-economic modelling, is central to exploring these trade-offs before construction, for example:

• testing different cooling architectures and retrofits in silico,

• quantifying the impact of layout and airflow on hotspots and PUE,

• assessing scenarios where data centres are paired with solar, nuclear or green hydrogen infrastructure.

  
  

Mansim's Perspective

PwC’s conclusion is that the Middle East’s data-centre market offers “unmatched upside” for investors and operators, but only if they can navigate constraints around power, cooling, supply chains and talent.

From Mansim’s perspective, the region is also strategically important as a testbed for research-led, innovative solutions to exactly these problems.

As Bob Baghaei, Head of Business Development at Mansim, puts it:

“The Middle East is one of the fastest-growing data-centre markets in the world, and AI is only accelerating that trajectory. It’s a region where energy intensity, cooling challenges and ambitious digital plans all come together. For a company like Mansim, which focuses on research-based and innovative simulation solutions, this is a natural place to demonstrate how advanced CFD, thermal modelling and system-level thinking can help tackle the next generation of data-centre energy and cooling problems.”

As AI campuses and hyperscale facilities multiply across the GCC, the question will not just be how quickly they can be built, but how efficiently, resiliently and sustainably they can operate over decades. The answer will depend heavily on how power, cooling and infrastructure are modelled, designed and optimised from the very beginning.