"State-owned AI won't solve inequality": Rest of World's bold thesis on AI in the Global South

Rest of World investigates the national AIs launched by several countries in the Global South. The finding is stark: nothing competitive has emerged, and the sovereignty argument masks a repetition of the telecom errors of the 1990s.

The fact

An investigation by Rest of World (July 24, 2026) documents the waves of "state AI" launched in recent years by several countries in the Global South - in Southeast Asia, Africa, and Latin America - in response to the fear of digital colonialism. The article's conclusion is straightforward: none of these initiatives, to date, has produced competitive AI nor reduced a measurable inequality. Under-trained models, compute rented from foreign actors, budgets spent on licensing fees.

Our analysis

The sovereignty argument is legitimate. But the investigation reminds us of an unpleasant truth: a model that does not find domestic use does not get funded. The trajectories of autonomous AI that have succeeded are based on a private sector, corporate demand, and public use, not on an isolated state order. Conversely, the "state builder of national AI" stance reproduces a familiar pattern: subsidizing infrastructure without building demand. The political narrative changes; the economic mechanism does not - AI is no less a market than the telecoms of the 1990s.

To watch

The hybrid state-private models that are emerging post-European AI Act, and initiatives backed by a pre-existing massive public use (payment rails, digital identity). This is where the sustainable version of AI sovereignty is at stake - not in the announcements of a "national ChatGPT".