AI Hyper-Scaling Digital Inequality
IEEE Spectrum Danica Radovanović
AI is going global, but shaping it is staying local — to the US, China, and a handful of firms. South Africa and Indonesia show what happens when countries try to get a seat at the table instead of just using the tech.
Based on reporting by IEEE Spectrum, Danica Radovanović — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
There's a comfortable story the AI industry likes to tell: the technology is spreading everywhere, so eventually everyone benefits. The reality, according to a decade of digital-inclusion work summarized in a recent IEEE Spectrum piece, looks more like the old digital divide wearing a new outfit — except this time the gap isn't just about who has a laptop, it's about who gets to decide what the technology is for.
The numbers back that up starkly. The US hosts more than 5,000 data centers, over ten times any other single country, and according to World Bank figures it controlled roughly 87 percent of global cloud and data-storage exports in 2023. That's not just a technical imbalance. It means most nations are renting their AI future from someone else's servers, subject to someone else's commercial and geopolitical priorities. Add in skills: across OECD countries only about 40 percent of adults have more than basic digital problem-solving ability, and tertiary-educated workers are twice as likely to get AI training as those with just upper-secondary schooling. The people least equipped to shape these systems are the ones most likely to be shaped by them — think the Dutch childcare-benefits algorithm scandal, or Amazon's scrapped AI hiring tool.
What's more interesting than the gap itself is what a few countries are trying to do about it. South Africa's Department of Communications and Digital Technologies put out a draft national AI policy in April 2026, then yanked it within days after a journalist found at least six fabricated academic citations — AI hallucinations baked into a government policy document. The minister called it an unacceptable lapse, which is generous phrasing for a genuinely embarrassing moment, though a newly formed AI panel now has a chance to salvage some credibility.
Indonesia's approach is quieter and, frankly, more instructive. Rather than chasing frontier models, its National Research and Innovation Agency has built an app that uses satellite data to help artisanal fishermen find fish, multilingual models trained on Javanese and Sundanese alongside Indonesian, and government chatbots for public services. The country's digital ministry wants to train 100,000 AI-skilled workers a year. It's a bet that usefulness and local relevance beat scale for scale's sake.
Africa's Continental AI Strategy and the 2025 Kigali AI summit point toward a similar idea: regional coordination, open-source tooling, and public universities as a hedge against permanent dependency on Silicon Valley and Beijing. None of this requires rejecting cloud infrastructure or venture capital, which the source is careful to note. It just reframes the central question of the AI moment away from who moves fastest, and toward who actually gets to decide where all this is headed.
My take — AI-written commentary, not fact-checked reporting
I've spent enough time around EU tech policy to recognize this pattern instantly: everyone drafts a strategy, nobody builds the underlying capacity, and then a fabricated citation or two exposes the whole thing as theater — which is exactly what happened in Pretoria. Indonesia's boring, unglamorous fish-finder app is more meaningful than another frontier-model press release, and I'd bet on quiet public-sector competence over sovereign-AI grandstanding every time.
Read more about this at: IEEE Spectrum
