Who Controls the Infrastructure of AI in the Arab World?
Data centres, compute access, and model training: the geopolitics of artificial intelligence is playing out on Arab soil, and the stakes could not be higher.
When the UAE announced its national AI strategy in 2017, most observers treated it as an aspirational document — the kind of ambitious policy framework that governments produce to signal modernity without committing to specifics.
Nine years later, the UAE operates one of the world's most significant AI compute clusters, has produced a frontier large language model, and is negotiating data-sharing agreements with both American and Chinese technology firms simultaneously. The aspirational document turned out to be a blueprint.
The Infrastructure Question
The fundamental question of AI sovereignty — who controls the infrastructure on which artificial intelligence runs — is one that most nations have not yet answered clearly. The Arab world is no exception. But several states in the region are moving faster than their critics acknowledge.
Compute infrastructure, broadly defined, encompasses three layers: the physical data centres and their energy supply; the semiconductor hardware on which models are trained and run; and the software frameworks, datasets, and model weights that constitute the intellectual capital of the AI economy.
On each of these layers, the picture in the Arab world is complicated.
Nour Al-Amin covers technology and policy for Imprint.
Weekly intelligence for serious readers
Original analysis, featured essays, and the ideas shaping the region — delivered every Thursday.