Every goal, every pass, every yellow card at the 2026 FIFA World Cup will be logged by a human. Not a referee. Not a statistician in a Zurich office. A data annotator in Brazil, Cambodia, or the Philippines, paid roughly 60 euros per match to tag match actions for a foreign sports data company, according to Rina Chandran and Michael Beltran reporting for Rest of World. That annotated data trains the computer vision algorithms that power the AI-enhanced broadcast experience. The same pipeline—humans labeling raw footage so machines can learn—is the invisible infrastructure behind every chatbot, every recommendation engine, every generative AI tool that references a YouTube video. The World Cup is the most visible deployment of this model, but it is not the only one. The same logic that sends annotation work to Rio de Janeiro and Phnom Penh is already looking at Cairo, Casablanca, and Riyadh.
The geography of data annotation is not random. Rafael Grohmann, assistant professor of media studies at the University of Toronto, told Rest of World that annotation workers cluster in cities like Manila, Cairo, Chennai, and Ternopil. The data value chain has a clear geography: high-value analytic work stays in wealthy centers, while the repetitive, low-margin work of labeling data flows to Eastern Europe, Africa, South Asia, and Southeast Asia. Cairo is already on that map. The question is not whether the annotation pipeline will reach MENA. It is already here. The question is how it grows: as a low-wage, low-agency labor pool, or as something more strategic.
MENA creators are already feeding the machine, whether they know it or not. Sam Gutelle, reporting for Tubefilter, notes that Jellyfish found more than 25% of chatbot prompts are answered with responses that reference content from YouTube creators, and that more than one million unique YouTube videos are cited by AI chatbots in the U.S. each day in the CPG category alone. Marques Brownlee was one of the first YouTube stars to call out generative AI services for mimicking his style without his consent, and is now leading the charge for stronger protections against unlicensed scraping. But Brownlee has influence. A creator in Kuwait or Morocco, with a smaller audience and no legal team, does not. Their content is being scraped, their dialect and cultural context are being absorbed into training data, and they receive nothing in return. The same forces that turn a World Cup match into structured data are turning creator content into AI training material. The creator is the raw material supplier. The annotator is the factory worker. Neither owns the output.
There is a structural lesson here, and it comes from an unlikely place: the EV battery supply chain. Indranil Ghosh, reporting for Rest of World, states that China has built over 85% of global EV battery recycling capacity and mandates immediate recycling of spent batteries for metal recovery, while the U.S. prefers giving depleted batteries a second life as grid storage first. The Council on Foreign Relations wrote in February 2026 that the U.S. cannot out-mine and out-process China, and should instead leapfrog by scaling disruptive innovation, recovery, and recycling. The lesson for MENA creators and operators is not about batteries. It is about infrastructure ownership. China does not just produce batteries; it owns the recycling infrastructure that recovers their value. The annotation pipeline is the recycling infrastructure of the AI economy. The region that owns the labeling hubs, the training datasets, and the quality-control workflows for Arabic dialects will capture far more value than the region that simply supplies raw content or raw labor.
The region that owns the labeling hubs, the training datasets, and the quality-control workflows for Arabic dialects will capture far more value than the region that simply supplies raw content or raw labor.
This is the fork. One path: MENA becomes the next annotation pool. Cairo, already named by Grohmann as a concentration point for annotation workers, scales up as a low-cost labeling hub for global AI companies. Creators continue to have their content scraped without compensation, and the region supplies both the labor and the raw material without capturing the value. The other path: MENA operators—MCNs, agencies, platform teams—build regional annotation infrastructure that specializes in Arabic dialects, local cultural context, and the specific data needs of regional AI products. This is not a charity project. Arabic is a high-value language with dozens of dialects. The AI companies that want to serve the MENA market need training data that reflects how people actually speak in Riyadh, in Cairo, in Casablanca. That data does not exist in the global annotation pools. It has to be built locally.
The World Cup model shows the template. The battery recycling model shows the strategic imperative. The scraping of creator content shows the stakes. The region that builds the annotation infrastructure for Arabic will own the next layer of the AI economy. The region that waits will supply the labor and the content, and watch the value flow elsewhere.