Analysis: the concentration of computing infrastructure and skills deepens global inequality in AI
An analysis by IEEE Spectrum describes deepening global inequality in AI: the concentration of data centers and cloud services in the USA, differences in digital skills, the withdrawal of the South African AI policy over nonexistent citations, and the Indonesian BRIN project for local communities.
An analysis published on IEEE Spectrum describes how the so-called digital divide between countries is deepening with the rise of AI. According to Stanford University data from AI Index 2026, the United States hosts over 5000 data centers, more than ten times as many as any other individual country, and according to World Bank, the USA accounted for roughly 87 percent of global exports of cloud and data services in 2023. The author states that this concentration of computing infrastructure and commercial cloud services in a small group of countries puts AI development beyond the technological, commercial and geopolitical control of most other countries.
According to the article, inequality concerns not only infrastructure but also skills. According to OECD data, across member countries, only around 40 percent of adults have digital skills beyond the basic level, and participation in AI-related training is strongly linked to education – 36 percent of people with a university education took part in AI training in the past year, compared with 18 percent of people with a secondary education. According to the author, people on the disadvantaged side of this divide are more likely to experience AI as an opaque system that affects them from outside – the article mentions, for example, the Dutch scandal involving an algorithmic childcare benefits system or the discontinued recruitment tool developed by Amazon.
The article cites South Africa and Indonesia as specific examples. In April 2026, the South African Ministry of Communications and Digital Technologies published a draft national AI policy proposing new oversight institutions, but withdrew the draft a few days later after a journalist discovered that at least six academic citations in the document did not exist and were apparently AI-generated hallucinations. The minister called it an “unacceptable failure”. In Indonesia, by contrast, BRIN (National Research and Innovation Agency), which now leads the implementation of the national AI strategy, is, according to the article, developing practical tools for disadvantaged communities instead of pursuing cutting-edge models – for example, an application that uses satellite data and machine learning to locate schools of fish for small-scale fishers, and multilingual language models trained on Indonesian and local languages such as Javanese and Sundanese.
The full text of the source article is not available; further details on both cases and other examples can be found in the source article.
Why it matters
The text shows that access to computing power, cloud services and digital skills remains concentrated in a small number of countries and companies, affecting whose languages, institutions and priorities are reflected in AI systems. The South Africa case also provides a concrete illustration of the risk of unchecked AI use in drafting public documents – fabricated citations led to the withdrawal of the entire draft policy.
Relevant practical impact
What this means
For a business
Companies offering AI products in global markets should account for the possibility that models trained in narrow linguistic and institutional environments may not serve users outside those environments well, and for the reputational and regulatory risks of unchecked AI use in public documents, as the South Africa case demonstrated.
Risks and complianceCheck the original
Event sources
only one source so far · 1 publisher, 1 independent. We count feeds from the same owner only once.