The Gulf’s AI strategy is beginning to look less like a technology policy than a new form of industrial policy. Saudi Arabia and the United Arab Emirates are trying to turn abundant capital, relatively low-cost energy and sovereign balance sheets into a new export: computing power. The opportunity is real. But the decisive constraint may not be money. It may be whether Gulf power systems can deliver enough reliable electricity, cooling and network capacity quickly enough to keep pace with the AI investment cycle.
A race measured in megawatts
The scale-up is already visible. Saudi Arabia says its operational data-centre capacity rose from 68 megawatts in 2021 to more than 440 MW in 2025, with more than 60 data centres and investment exceeding US $4.3 billion, according to the Saudi Press Agency. Humain, the Public Investment Fund-backed AI company, has secured financing terms for up to 250 MW of new capacity and is targeting roughly 6 GW by 2034, Reuters reported in January.
In the UAE, Stargate UAE is being built as a 1 GW AI cluster inside a planned 5 GW AI campus in Abu Dhabi. G42 said construction of the first 200 MW was well underway in late 2025, with delivery planned for 2026. Those numbers are large enough to shift the region from a market that mainly consumes cloud services toward one that could host globally traded AI workloads.
Yet announced capacity is not the same as energised capacity. The International Energy Agency said in April 2026 that electricity consumption by AI-focused data centres jumped 50% in 2025, while total data-centre demand reached about 485 TWh. It expects total demand to roughly double to 950 TWh by 2030, but warns that bottlenecks across power infrastructure, equipment and chips are already constraining more aggressive expansion scenarios.
The energy advantage becomes a grid problem
The Gulf’s apparent comparative advantage is straightforward: abundant gas, strong solar resources, available land and governments capable of coordinating large infrastructure projects. For AI developers, the cost and speed of securing electricity can matter as much as proximity to customers. Where grid queues stretch for years, the GCC can plausibly compete by offering land, financing and generation together.
But cheap energy is not the same as unlimited power. Gigawatt-scale AI campuses resemble major industrial facilities in their electricity requirements, while Gulf grids must simultaneously accommodate air conditioning, desalination, industrial expansion and population growth. The IEA estimates that Middle Eastern electricity demand grew by nearly 4% in 2025, even before many of the region’s largest AI projects reached full scale.
Cooling adds another constraint. An April 2026 policy note from the Middle East Council on Global Affairs argues that the GCC’s extreme heat and water scarcity create a difficult water-energy trade-off for data centres. More water-efficient cooling can reduce direct consumption, but may increase electricity requirements or require costlier liquid-cooling technologies.
A comparative advantage – or an expensive subsidy?
The deeper economic question is whether the Gulf can turn this infrastructure into a durable export industry rather than a sovereign-funded capacity race. If global technology companies rent Gulf compute because it is genuinely cheaper, faster and more reliable, the region could convert its energy base into a new digital comparative advantage. The logic resembles earlier Gulf strategies in aluminium, petrochemicals and aviation: use a structural input advantage to build globally traded products and services.
But AI carries different risks. Hardware depreciates rapidly, improvements in model and chip efficiency could change capacity requirements, and access to frontier processors remains geopolitical. In July 2026, Washington eased export restrictions for approved UAE entities, allowing easier access to advanced US computing equipment; that preferential treatment itself illustrates how infrastructure economics can depend on strategic alignment as much as electricity prices.
The real test, therefore, is not how many gigawatts are announced. It is how many can be connected, cooled, equipped with leading-edge chips and sold to paying customers without excessive fiscal or environmental costs. The Gulf has the capital for the AI boom. Whether it has the power will depend on how successfully it turns its energy system into digital infrastructure — and whether global demand arrives fast enough to justify the build-out.
