AI Exacerbates Global Inequalities Without Yet Delivering Economic Gains
Low-income countries capture only 15% of the productivity gains generated by artificial intelligence, compared to 70% for wealthy countries, according to an OECD report published in November 2024. This imbalanced distribution occurs even as AI has yet to generate the massive productivity gains promised in developed economies.
Unlike the mobile revolution of the 2000s, which enabled developing countries to “leapfrog” fixed infrastructure, AI is widening a technological gap that could permanently marginalize one-third of humanity. Four cumulative barriers—infrastructure, skills, capital, and governance—are transforming AI into a factor of exclusion rather than a tool for economic inclusion.
Key Points
- Low-income countries capture only 15% of AI productivity gains compared to 70% for wealthy countries, according to the OECD
- Sub-Saharan Africa lags by 8 years on AI adoption compared to developed countries
- 80% of AI patents are filed by five countries: the United States, China, Japan, South Korea, and Germany
- The Global Initiative for Inclusive AI aims to train 10 million Africans by 2030
Failing Digital Infrastructure Blocks AI Adoption
Sub-Saharan Africa has an internet penetration rate of 43%, compared to 88% in OECD countries. This basic digital divide is compounded by a shortage of data centers and computing power. The African continent hosts less than 1% of global data centers while accounting for 17% of the world’s population.
Bandwidth represents a critical bottleneck. AI applications require high data rates and low latency that African networks struggle to provide. The average cost of a gigabyte of mobile data is $7.12 in Central Africa compared to $0.09 in Western Europe, according to the International Telecommunication Union.
This failing infrastructure contrasts with the mobile experience of the 2000s. Mobile phones had allowed African countries to bypass costly fixed networks. AI, by contrast, requires heavy infrastructure—servers, fiber optics, stable electricity—that few low-income countries can deploy quickly.
The Digital Skills Deficit Deepens the Divide
The gap widens at the level of human resources. OECD countries have an average of 4.2 computer engineers per 1,000 inhabitants, compared to 0.3 in Sub-Saharan Africa. This disparity reflects imbalanced educational investments: R&D spending represents 2.4% of GDP in developed countries compared to 0.3% in Africa.
Higher education is struggling to train the required profiles. Africa produces 350,000 graduates in science and technology annually, compared to 4.5 million in China and 568,000 in the United States. This shortage is worsening with brain drain: 70% of African engineers who graduated from Western universities do not return to their countries of origin.
Local businesses are suffering the full impact of this shortage. A survey of 2,400 African companies reveals that 78% cite the lack of digital skills as the main obstacle to AI tool adoption, ahead of cost or access to financing.
Financial Asymmetry Concentrates AI Innovation
AI investments are massively concentrated in a few hubs. In 2023, the United States and China captured 67% of the $200 billion invested globally in AI, according to Stanford HAI. The entire African continent received 0.8% of these investments, or $1.6 billion.
This financial concentration feeds a vicious cycle. High-performing AI algorithms require enormous volumes of training data and expensive computing power. Companies in wealthy countries have easier access to these resources, widening their technological lead. The cost of training an advanced language model now reaches $100 million, effectively excluding most southern players.
Access to venture capital reflects this asymmetry. Africa raised $3.5 billion in tech venture capital in 2023, compared to $170 billion in the United States. Investment funds remain concentrated in mature ecosystems, limiting the emergence of African technology champions capable of competing with Western or Chinese giants.
Fragile Digital Governance Hinders Local Innovation
Regulatory instability discourages technology investments. 62% of African countries lack a specific legal framework for data protection, compared to 95% of OECD countries. This regulatory gap creates legal uncertainty that frightens international investors and complicates technology transfer.
Administrative corruption raises deployment costs. Transparency International estimates that a technology project costs on average 23% more in Sub-Saharan Africa than in Southeast Asia due to “informal taxes” and bureaucratic burdens. This administrative friction discourages foreign technology companies from establishing operations and handicaps the local ecosystem.
The absence of common technical standards fragments African markets. Unlike Europe, which harmonizes its digital regulations, Africa has 54 different regulatory frameworks for telecommunications and digital technology. This fragmentation limits economies of scale and complicates the development of continental solutions.
Catch-Up Initiatives Struggle to Reverse the Trend
Several programs are attempting to close the AI gap between North and South. The Global Initiative for Inclusive AI, launched by the World Bank in 2023, aims to train 10 million Africans in AI technologies by 2030. It mobilizes $2.4 billion over seven years to develop Africa’s digital ecosystem.
Google announced a $1 billion investment over five years to improve African connectivity, including laying submarine cables and installing local data centers. Microsoft is developing AI centers in South Africa, Kenya, and Nigeria to bring computing power closer to local users.
These efforts remain insufficient given the scale of the challenge. The gap is widening faster than catch-up programs are progressing. While Africa trains a few thousand AI specialists per year, China graduates 200,000 and the United States 65,000. This temporal asymmetry risks permanently anchoring technological dependence in the South.
The experience of companies reinventing skills validation shows that AI can create new skilled jobs. But these opportunities are currently concentrated in countries that already have a mature digital ecosystem.
The Global Redistribution of AI Gains Remains Uncertain
AI could theoretically benefit developing countries through the relocation of digital services or the automation of repetitive tasks. But this redistribution remains hypothetical. The most advanced AI models remain controlled by a handful of American and Chinese companies that can direct their diffusion according to their geopolitical interests.
European AI regulation influences this dynamic. The AI Act adopted in 2024 imposes strict ethical standards that could slow AI adoption in Europe but establish global norms. African countries could benefit from these standards by avoiding the pitfalls of unregulated AI, provided they have the technical capacity to implement them.
The emergence of “frugal” AI solutions adapted to the constraints of the South offers an alternative path. Indian companies are developing algorithms optimized to run on basic smartphones with limited connectivity. These innovations could democratize access to AI without requiring heavy infrastructure.
The question of financing remains open. Traditional development aid mechanisms struggle to keep pace with the speed of technological evolution. AI technology transfer requires new financial instruments capable of rapidly mobilizing private capital while guaranteeing inclusive development.
Sources
- OECD - AI and the global productivity divide
- International Telecommunication Union - ICT Facts and Figures 2024
- Stanford HAI - Artificial Intelligence Index Report 2024
- World Bank - Global Initiative for Inclusive AI 2023
- Transparency International - Corruption Perceptions Index 2024