UAE’s AI energy advantage tested by security and grid constraints

During the artificial intelligence boom, the primary constraint has shifted from chip availability to electricity supply. In the United States, the world’s largest data centre market, the grid interconnection queue holds roughly 2,300GW of projects, and new AI campuses can face years-long waits for the power they need.

TrendForce projects that by 2030, global data centre demand could outstrip available grid capacity by more than 250GW. This shift has made the UAE’s natural advantages — abundant energy and rapid construction capability — increasingly valuable. The key question is whether these advantages can be maintained.

Planned projects and power requirements

Two initiatives are often confused. The UAE-US AI Campus is the master plan, a site spanning about 26 square kilometres near Abu Dhabi designed to accommodate up to 5GW of data centres. Stargate UAE is its first phase, a 1GW cluster being built by G42 in partnership with OpenAI, Oracle, Nvidia, Cisco and SoftBank. The first 200MW of this phase are expected to come online this year. Everything beyond that initial stage remains on paper.

According to Wood Mackenzie, the UAE consumed approximately 173TWh of electricity in 2025, with data centres accounting for about 3TWh, or 2 per cent of total consumption.

The arithmetic is striking. Running at full capacity, Stargate’s first gigawatt alone would use nearly three times as much electricity as all data centres in the country currently consume. The full campus would add demand equivalent to a quarter of the UAE’s present national electricity consumption. No country has that much spare capacity sitting idle; it must be built.

The UAE is building it, and that is where the advantage becomes concrete. This year, Masdar and the Emirates Water and Electricity Company (Ewec) reached financial close on a $6.1 billion project that combines 5.2GW of solar capacity with 19GWh of battery storage to deliver 1GW of round-the-clock power. Scheduled for completion in 2027, it is the first project of its kind at this scale. Additionally, Taqa and Ewec are constructing a 1GW gas-fired plant at Al Dhafra, financed at $980 million, specifically to supply data centres. When combined with Barakah’s nuclear baseload and existing gas-fired capacity, few markets can assemble firm, low-cost power as quickly.

Speed is the real competitive edge. In the UAE, state-backed entities plan generation, transmission and much of the demand together. In the US, a developer can wait years in a queue behind other projects. That co-ordination, more than the energy itself, is the advantage competitors find hardest to replicate.

Delivery challenges and security risks

Generating power is one task. Delivering it reliably and cleanly to the right sites is another. Wood Mackenzie notes that the UAE does not yet allow corporate or virtual power purchase agreements, and renewables are procured only through utility auctions. For hyperscalers with clean-energy commitments, this limits how they can buy green power — a gap that rival hubs may close first.

The regional conflict has added a tougher test. In March, Iranian strikes damaged two Amazon Web Services data centres in the UAE, and in April an Iranian military video named Stargate as a target.

In September, Reuters reported that the UAE was reworking the 5GW master plan into a network of smaller sites, with underground facilities, air defences and duplicated power and cooling. The first phase is going ahead, though it is unclear how far it has progressed.

This month, Mohsen Rezaei, a former commander of Iran’s Islamic Revolutionary Guard Corps, warned that data centres and AI facilities across the region could be targeted if the conflict resumed.

Dispersal changes the power equation. Every additional site needs its own grid connection, backup generation and cooling. Large AI training runs also work best when chips sit close together, so splitting clusters can reduce performance. The UAE’s power advantage now has to be delivered to more places, at higher cost, and defended by military force.

Sustaining the edge

To make the advantage last, supply must stay ahead of demand. The 2027 round-the-clock solar project and new gas capacity are the right signals. Further phases of the campus should be matched, gigawatt for gigawatt, with committed new generation.

Opening the market for clean power by allowing corporate power purchase agreements would let global tech companies meet their own climate targets in the UAE, removing a reason to build elsewhere.

Pricing resilience into the offer is also important. Hardened, redundant sites cost more, but that protection is also a selling point. Few markets can promise capacity that is protected by design.

Securing the chips that use the power is another factor. The UAE’s access to advanced US chips without individual licences runs until about April 2027. Power without processors is idle capacity.

The verdict: The UAE does not control the world’s energy for AI. But it holds a natural advantage in abundant, low-cost power, and a structural one in its ability to plan and build at speed.

On current evidence, that combination is real and is already attracting the world’s leading AI companies. It becomes a lasting strategic advantage only if the UAE proves it can deliver that power securely, to dispersed sites, through a regional war. That is the test of the next two years.

Andrea Zanon is senior adviser to entrepreneurs and global leaders, with more than 20 years of experience in economic strategy, market access and geopolitical risk.