Who Really Wins From AI? IMF Finds $2.7 Trillion Gains Flow Mainly to Wealthier Countries

A new IMF–World Bank study estimates that AI is already generating **US$2.7 trillion** in annual labor cost savings, but nearly all benefits remain concentrated in high-income economies and high-skilled occupations. It calls for stronger AI governance, workforce development, multilingual digital content and broader AI adoption to ensure more inclusive and sustainable economic growth.

Artificial intelligence is already creating measurable economic value. Still, its benefits remain highly uneven across countries and occupations, according to a new study by researchers from the International Monetary Fund (IMF) and the World Bank. Using actual AI usage data from more than 100 countries between January 2025 and February 2026, the researchers estimate that AI currently generates US$2.7 trillion in annual labor cost savings, equivalent to around 3.4% of the combined GDP of the 86 countries included in the core analysis. Unlike previous studies that focused on AI's potential, this research measures where AI is already being used and who is benefiting the most. The findings carry important lessons for governments, development agencies and businesses as countries compete to capture AI-driven productivity gains.

AI is creating huge productivity gains, but not for everyone

The study finds that AI adoption has expanded beyond software engineering into education, healthcare, office administration, sales and customer service. Although these occupations generally earn lower wages than software developers, they employ millions more workers. As a result, overall productivity gains continue to rise because AI is helping a much larger workforce.

Annual labor cost savings increased from about US$1.2 trillion in August 2025 to nearly US$2.7 trillion by February 2026, with 84% of the increase driven by the rapid growth in AI adoption rather than improvements in productivity per user. This suggests that the next phase of AI-led growth will depend on expanding AI access across industries rather than relying solely on advances in AI technology.

However, the gains remain concentrated in wealthier economies. High-income countries generate AI productivity gains equal to around 4.2% of GDP, compared with 0.6% in middle-income economies and just 0.1% in low-income countries. Together, advanced economies account for nearly 96% of global AI-related productivity gains, highlighting a widening technology divide.

Governments need policies that spread AI across the economy

The research shows that AI regulatory readiness is one of the strongest factors determining how much economic value countries capture from AI. Countries with stronger governance, clearer AI regulations and better digital institutions generate larger productivity gains and distribute those benefits across a wider range of occupations.

For policymakers, this means AI strategies should extend beyond investing in digital infrastructure. Governments also need policies that encourage AI adoption in education, healthcare, public administration, agriculture, manufacturing and small businesses. Investments in digital skills, workforce training and AI governance will be critical to ensure productivity gains benefit the broader economy rather than only highly skilled professionals.

The study also finds that economies with larger service sectors gain more because generative AI performs particularly well in communication-intensive and knowledge-based jobs.

Development partners have a bigger role than building digital infrastructure

The findings suggest that international development institutions, including multilateral development banks and donor agencies, should expand their AI support beyond internet connectivity projects. Future assistance should increasingly focus on AI governance, regulatory capacity, workforce development and public-sector digital transformation.

One important finding concerns language. Countries where English is an official language experience faster AI adoption across occupations because legal documents, educational materials and business information published in English are better represented in AI training data. This allows AI systems to provide more accurate and locally relevant responses.

For developing countries, this highlights the need to create more high-quality digital content in local languages. Development partners can help by supporting multilingual AI datasets, digital public infrastructure and knowledge systems that make AI useful across a wider range of sectors.

Businesses that adopt AI early could gain a competitive edge

The report also offers important lessons for private-sector stakeholders. Companies that integrate AI across departments—not just software development—are likely to achieve larger productivity improvements. Customer service, finance, marketing, education, legal services and healthcare all present significant opportunities for AI-driven efficiency.

Small and medium-sized enterprises, particularly in developing economies, remain underrepresented in AI adoption. Expanding access to affordable AI tools, employee training and digital infrastructure could help businesses improve competitiveness while creating new opportunities for innovation and growth.

The researchers also caution that the estimated US$2.7 trillion represents the value of labor time saved rather than direct GDP growth. It does not account for infrastructure costs, energy use, AI investments or potential job displacement. Nevertheless, the study concludes that AI is gradually becoming more inclusive. Between August 2025 and February 2026, the number of countries where AI benefits became more broadly distributed increased from 32 countries (30%) to 53 countries (48%). The message for policymakers, development partners and businesses is clear: countries that strengthen AI governance, invest in workforce skills and expand AI adoption across the economy will be better positioned to capture the next wave of productivity-led growth.