AI-related job vacancies have surged ninefold globally since 2021, with middle-income countries accounting for one in five of the new roles, signalling a shift in how artificial intelligence is reshaping labour markets beyond advanced economies, the World Bank said.
The findings are contained in the World Bank’s ‘Digital Progress and Trends Report 2025: Strengthening AI Foundations’, which shows that while high-income countries continue to dominate AI innovation and capital flows, emerging economies are increasingly becoming active participants in the fast-growing AI workforce, driven largely by the rapid adoption of generative AI tools.
More than 40 per cent of global ChatGPT traffic now originates from middle-income countries such as Brazil, India, Indonesia and Vietnam, the report said, underscoring how consumer-level adoption of generative AI is translating into rising demand for AI-related skills in these markets.
The World Bank estimates that one in five GenAI job postings worldwide is now located in middle-income economies, a sharp increase from just a few years ago.
Still, the report cautions that the momentum could stall without targeted investment in digital infrastructure and skills.
High-income countries, home to just 17 per cent of the global population, continue to host 87 per cent of notable AI models, 86 per cent of AI startups and attract 91 per cent of global AI venture capital, reinforcing structural imbalances in the global AI economy.
Read also: Thousands of jobs in Lagos are posted online, yet underemployment persists- LSETF
To help developing nations convert rising AI adoption into sustainable employment and economic growth, the World Bank outlined what it calls the “Four Cs” of AI readiness: Connectivity, Compute, Context and Competency.
Connectivity remains a major constraint. While 93 per cent of people in high-income countries are online, only 54 per cent of lower-middle-income countries and 27 per cent of low-income countries have internet access, limiting participation in cloud-based AI platforms, remote work and online training.
Compute capacity is another bottleneck. Middle- and low-income countries together account for just 23 per cent of global data centre capacity, compared with 77 per cent in high-income economies, restricting the ability of local firms to develop, train and deploy AI models at scale.
Beyond infrastructure, the World Bank stressed the importance of “context”, the availability of local data, languages and cultural relevance. Most AI models are trained predominantly on English-language text, reducing their effectiveness in local settings. However, newer data formats such as audio and video are opening pathways for emerging economies to build AI tools tailored to domestic needs in sectors such as health, agriculture, finance and education.
Human capital remains the most critical gap. Fewer than five per cent of people in low-income countries have basic digital literacy, compared with 66 per cent in high-income nations, the report said. Without investment in digital skills, AI literacy and advanced technical training, countries risk seeing AI adoption expand without delivering broad-based employment gains.
The report also pointed to the rise of affordable, lightweight “small AI” applications as a potential equaliser. These tools, which can run on everyday devices without extensive data centre infrastructure, are already being used to analyse health data, support small businesses and improve access to services, offering a path for developing economies to bypass some traditional infrastructure barriers.
“The Four Cs provide a roadmap for countries seeking to unlock AI’s transformative potential. Targeted investment in connectivity, compute, context and competency will allow developing nations to participate meaningfully in the global AI economy, create jobs and drive sustainable development,” the World Bank said.
While the ninefold surge in AI job vacancies highlights the scale of opportunity, the World Bank warned that the window for catching up may be narrow.
Without sustained policy focus on infrastructure, skills and locally relevant AI applications, developing countries risk remaining users of AI technologies rather than producers of the next wave of digital innovation.
