AI digital inequality is no longer just about who can open a chatbot. Recent data from Stanford University, the World Bank, the OECD and UNESCO show a more structural split: a few countries and companies hold much of the compute, cloud infrastructure, skills and policy capacity needed to build and govern AI, while many others mostly consume systems designed elsewhere.
The issue is becoming harder to dismiss as AI moves into email, software development, hiring, education, health care, finance and public administration. The people and institutions with fast connectivity, trained staff and reliable cloud access can use AI to extend what they already do. Those without that base are more likely to meet AI as a system that sorts, scores or excludes them.
What is AI digital inequality?
AI digital inequality is the gap between those who can meaningfully build, audit, adapt and govern AI systems and those who mainly receive them as imported tools. It includes access to computing infrastructure, local-language data, technical skills, regulatory capacity and a seat in the forums where standards are set.
Stanford University’s 2026 AI Index report found that the United States hosts more than 5,000 data centers, more than ten times the number in any other single country. The World Bank reported that in 2023 the United States accounted for about 87 percent of global exports of cloud computing and data storage services.
That matters because most AI workloads now run on cloud platforms rather than on local machines. If the infrastructure sits elsewhere, so do many of the commercial dependencies and geopolitical choke points. Countries can still deploy AI, but the engines, pricing, technical standards and update cycles are often controlled outside their borders.
Who has the skills to use AI well?
The skills gap is just as uneven. The OECD says only about 40 percent of adults across its member countries have more than basic digital problem-solving skills. Its survey data also show that AI-related training is concentrated among people with more education: 36 percent of respondents with tertiary education said they had taken AI-related training in the previous year, compared with 18 percent of those with upper secondary education.
UNESCO has reported that governments are moving to put AI into education systems, but support for AI literacy in primary and lower secondary schools, and ethics training for teachers, remains uneven. That sequencing risks giving the largest benefits to students and workers who already have the strongest institutions behind them.
The downside is not theoretical. The Dutch childcare benefits scandal showed how algorithmic welfare systems can harm people when public agencies use opaque risk scoring. Reuters reported in 2018 that Amazon scrapped an AI recruiting tool after it showed bias against women. In both cases, people were subject to automated decisions they had little practical ability to inspect or shape.
South Africa and Indonesia show two different paths
South Africa’s Department of Communications and Digital Technologies released a draft national AI policy in April 2026 that proposed new oversight bodies. The department withdrew it days later after a journalist found that at least six cited academic sources did not exist and appeared to be AI-generated hallucinations. Reuters reported that the minister called the episode “an unacceptable lapse.”
Indonesia has taken a more applied route through its National Research and Innovation Agency, known as BRIN. Rest of World reported that BRIN has built AI tools for underserved communities, including an app that uses satellite data and machine learning to help artisanal fishers find fish, multilingual models for Indonesian and local languages such as Javanese and Sundanese, and chatbots for government services. In August 2025, Indonesia’s Ministry of Communication and Digital Affairs released a national AI roadmap with a goal of training 100,000 AI-skilled workers each year.
African governments are also trying regional coordination. In 2024, African ministers adopted a Continental AI Strategy and African Digital Compact. Participants at the April 2025 Global AI Summit on Africa in Kigali discussed local-language models, public universities, open-source systems and regional governance as ways to reduce dependence on externally built AI.
The divide, then, is not only about access. It is about who gets to decide what AI is optimized for, whose languages it handles, which public problems it is aimed at, and who has the capacity to challenge it when it fails.
This story draws on original reporting from IEEE Spectrum.
