‘Build it and they will come’: How the UAE’s age-old strategy will expand to AI exports

For decades, the UAE built its reputation on making and moving things. Oil from the ground. Aluminium from smelters. Passengers through airports. Cargo through ports. Each time, the formula was the same — invest early before demand was obvious, build infrastructure at the centre of global flows, and wait for the world to arrive.

Al Kaissi says the country is running that formula again. The product this time has no weight, no volume, and plays into years of building and foresight. 

“We went from exporting barrels of oil, diversifying into financial services, aluminium, structural composites, ports and aviation, to now manufacturing intelligence, for both domestic consumption and export,” he told media.

“In a world that will consume not just billions, but ultimately trillions of tokens a day, demand for machine intelligence is structurally insatiable. Every new model, every new agent, every new use case asks more questions than the last. At G42, we have designed for that reality with two tightly linked engines: a token factory that industrialises the production of intelligence, and an agent factory that industrialises the deployment of that intelligence into useful, autonomous systems.”

What a token factory actually is

The concept requires a simple reframe. An AI data centre is a factory. Energy and compute go in. Tokens, the unit of output every AI system produces when it reasons, responds, or acts, come out. Manufacture enough tokens efficiently and they become a commodity. Own the factory and you control the supply.

Core42 is the digital infrastructure company that runs this system as part of G42’s token factory, alongside Khazna, which builds the physical data centre infrastructure. It provides the sovereign AI and cloud backbone that government agencies, hospitals, energy companies, and regulated industries depend on, without routing sensitive data through foreign platforms.

“Capacity is strategy,” Al Kaissi said. “You can have the best ideas in the world, but you either rent compute and become dependent on someone else, or rent intelligence at the model layer, unless you build indigenous capability.”

The five-gigawatt AI campus under construction in Abu Dhabi is the factory at the centre of this ambition. Domestic demand will absorb a portion of its output. A government committed to becoming AI-native and a population with the highest per capita AI adoption globally, according to Microsoft’s latest AI diffusion report, will consume steadily. The rest is exportable.

Geography and the distribution model

“Within a 3,200-kilometre radius, you can serve 3.9 billion people, roughly half the world’s population, with very low latency and high-bandwidth connectivity from the UAE,” Al Kaissi said. “The same reason UAE ports and airline carriers became such powerhouses. Part of it was geography, part of it was leaning in early and building a port in Jebel Ali when nobody thought there was any value. The build-it-and-they-will-come mentality paid off. The same logic applies here, only the demand is growing at an exponential pace globally for compute.”

The distribution mechanism is what Core42 calls the intelligence grid, a network that can include sovereign AI nodes deployed across multiple countries, each governed by government-to-government agreements in the form of Digital Embassies that preserve data sovereignty regardless of where the infrastructure sits. Core42 already operates clusters across the UAE, the United States, and Europe, with further expansions in capacity and geography planned for 2027.

The data no one else has

What compounds the advantage is not just compute. It is the datasets the UAE has assembled by circumstance and policy that the rest of the world cannot easily replicate.

In a rough example, every expatriate entering the UAE submits a chest X-ray for tuberculosis screening. Most countries only take those scans when something appears wrong. The UAE’s routine has created something rare.

“We are one of the only countries with a dataset of healthy chest X-rays,” Al Kaissi said. “Elsewhere, scans are taken only when there is suspicion of illness. That changes what AI can do in anomaly detection, augmenting physicians in both speed and accuracy. That dataset sits alongside an incredible dataset of whole genome sequenced nationals from the Emirati Genome Programme, which lends itself to population health management, precision medicine, and drug discovery. Satellite imagery from Space42 enables geospatial analytics. Decades of geological data from oil exploration support reservoir simulations and seismic analysis. Together they represent ways to create new innovation across multiple industries, all requiring sovereign cloud and AI compute at scale.”

“In other countries, you oftentimes have less permissive regulations around the use of those datasets or even the ability to access them,” he said. “Here, the government is constantly looking for ways to create enablement through technology, balancing risk, privacy, and innovation to improve outcomes.”

The first 200 megawatts of the campus is nearing completion, and plans to scale to five gigawatts are already in motion. The export model is still building. But the logic is identical to the one that turned a sandbank into the world’s busiest international airport and a creek into a global shipping hub.