The Algorithem #2: From the Gulf to India, the fight for who owns the AI future.
AI across the Global South, tracked weekly
AI is being built, regulated, and resisted across the Global South at a pace that rarely makes it into mainstream tech coverage. Each week, The Algorithm tracks the most important developments: new models, policy moves, funding gaps, and the harder questions about who AI is actually for.
1. The Gulf’s AI infrastructure just became a military target
Source: Rest of World · March 2026
On March 1, military drones struck Amazon Web Services data centers in the UAE and Bahrain, the first confirmed attack on hyperscale cloud infrastructure in history. Billions in committed AI investment are now under threat: Microsoft’s $15B UAE commitment, Amazon’s $5B Riyadh hub, OpenAI’s five-gigawatt “Project Stargate” campus in Abu Dhabi.
Seventeen submarine cables pass through the Red Sea, the data arteries connecting Europe, Asia, and Africa, and in an active war zone, repair ships cannot safely operate. Standard commercial insurance policies exclude acts of war. CSIS had warned explicitly, weeks before the conflict began, that adversaries would target data centers the same way they target pipelines and refineries.
Tech giants are already rerouting critical data workloads to India.
The Gulf had positioned itself as the AI capital of the non-Western world. It became its most visible liability overnight.
2. From Chile to Kenya, communities are pushing back against AI’s physical costs
Source: Rest of World · March 24, 2026
AI has always had a physical footprint. The Global South is where most of that footprint lands.
The environmental and labour costs of AI are landing hardest in the Global South, and communities are starting to organise. In Chile, environmentalists are protesting the seven billion litres of water consumed annually by Google’s data centers in drought-stricken regions. In Kenya, previously invisible data labelers have formed the continent’s first annotation workers’ union. In the Philippines, Code AI is responding to mass white-collar tech displacement. In Mexico, new legislation aims to criminalise AI deepfake political propaganda.
In this piece in Rest of World, read about the people taking charge of this conversation.
3. Egypt quietly launched Africa’s first sovereign AI model
Source: Cape Times · March 20, 2026
Egypt’s Ministry of Communications unveiled Karnak, a 30–80 billion parameter Arabic-language sovereign AI model. Applications unveiled alongside the model include SIA, a personalised AI tutor for Arabic language and Egyptian history, a legal assistant to help citizens and small businesses navigate regulatory frameworks, and healthcare AI engines for early detection of diabetic retinopathy and breast cancer, developed in collaboration with the UNDP.
Egypt has been building toward this since 2019, when it launched a national AI strategy prioritising skills development, data governance, and sector-specific deployment. Karnak is the most visible output of that strategy's second phase.
4. Burkina Faso is building AI tools for four languages that the Global North has never heard of
Source: Ecofin Agency · March 26, 2026
Burkina Faso, one of West Africa’s least-resourced nations, launched a national workshop to build AI voice tools for four local languages: Mooré, Dioula, Fulfuldé, and Gulmancema. The project, part of the country’s 2030 digital roadmap, focuses on voice recognition and machine translation to bring the non-French-speaking majority online. This is happening as Tunisia and Algeria launch a joint AI research platform and Ghana begins training its civil service in AI literacy. The infrastructure for African AI is being built simultaneously at every level of government, from the smallest to the largest.
5. India is moving from data labeling to training frontier AI models
Source: TechCrunch · March 25, 2026
India-based Deccan AI raised a $25M Series A to scale its network of over 1 million Indian domain experts, doctors, lawyers, and PhDs, who provide high-level feedback for training frontier AI models at clients like Google DeepMind. Workers earn between $10 and $700 per hour, with top contributors reportedly earning up to $7,000 per month. India wants to move away from the image of the world’s back office for cheap annotation. It is moving towards becoming the essential expert layer that makes the most powerful AI systems reliable.
That’s it for this edition of The Algorithm. If you spotted a story we missed, a startup, a policy move, a community using AI in ways nobody’s writing about yet, drop a line in the comments. The best leads come from readers on the ground.