Artificial intelligence (AI) is often touted as a tool to improve global prosperity and accelerate development in the Global South, yet policymakers and the public share concerns about its role in widening global inequality given that many of its core inputs—computing power, data, and talent—remain highly concentrated in a few countries and firms. For example, only 32 countries host AI-specialized data centers, most of which are found in the Global North, while Africa and Latin America together account for just 3% of global AI compute capacity. Similarly, the Global South represents 88% of the world’s population and generates vast amounts of data, but the lack of infrastructure means much of this is processed abroad.
Given these realities today, a global AI divide will not be defined solely by deployment and use of the technology, rather, it will encompass the fundamental inputs of energy systems, data centers, supply chains, and geopolitical power that make the technology so powerful. Yet such a trajectory is not inevitable; there is still an opportunity to shape the future of AI and promote inclusive and sustainable prosperity for all.
Concentration of AI infrastructure and investment
AI development relies on compute-heavy infrastructure such as data centers, graphics processing units (GPUs), and energy systems. According to McKinsey, the total global spending on data centers could reach $7 trillion by 2030, yet investment so far is highly concentrated in certain areas. For example, of the 23 gigawatts of global data center capacity that was under construction in September 2025, about 75% was in the United States. The International Monetary Fund’s AI Readiness Index scores countries from 0 (least ready) to 1 (most ready) based on indicators covering digital infrastructure, human capital and labor market policies, innovation and economic integration, and regulation and ethics. The index reveals stark gaps: Advanced economies score 0.68 on average, emerging markets score 0.46, and low-income countries score just 0.32, meaning advanced economies score more than double that of low-income countries in overall AI readiness. Among countries that score the highest in AI adoption, a common trend is early and consistent investment in digital infrastructure.
Energy as hidden constraint
AI infrastructure is fundamentally an energy problem. Globally, data centers account for around 1% of global electricity demand, though the impact differs greatly by country. In Ireland, for example, they make up 20% of the country’s electricity consumption, which has prompted some cities to pass moratoriums on new data centers until 2028. Aside from electricity, water is also a central concern. Data centers have become some of the fastest-growing industrial users of water in India, and facilities in Brazil are competing with agriculture for water. Energy constraints and rising temperatures are further complicating the cooling process across the board. Countries that manage to solve energy constraints will gain a decisive advantage in the AI race.
Unequal innovation ecosystems
While AI research and development remains concentrated, innovation ecosystems are also diverging in type. The Global South has focused on “applied AI” and efficiency-based innovation, prioritizing smaller, purpose-built models that can outperform large ones in local contexts with lower costs, less computing power needed, and higher local relevance. For example, InkubaLM, a South-African based compact language model covering African languages such as Swahili, isiXhosa, Yoruba, Hausa, and Zulu, focuses on efficiency by using edge computing for many tasks and prioritizing specific linguistic tasks instead of general-purpose capability.
Data extraction without value creation
The lack of infrastructure to process the large amounts of data generated in the Global South due to the rapid expansion of mobile and subsea cables means that the data is often exported. This rise of “digital extractivism” means that the value accrues elsewhere, often for companies in the Global North.
Governance and power asymmetry
Global AI governance is dominated by the Organization for Economic Co-operation and Development (OECD), G7, and major technology partners, while the Global South is often underrepresented and acts as “rule-takers,” rather than rule-setters. For example, less than 10% of all global AI governance frameworks originate in the Global South.
Reshaping labor markets
AI is reshaping labor markets both within countries and across them. On the positive side, AI has immense potential to be a growth multiplier. For example, by 2030, AI could generate $1.2 trillion for Africa’s economy (about 6% of its GDP). Productivity can improve through the automation of routine tasks and enhanced service delivery, while the technology also opens new industries and sectors like AI startups, digital services, and data economies. Beside these opportunities, though, are several risks, including job displacement, skills mismatch, and unequal access to training. For example, the Global North and Global South already diverge when it comes to AI adoption and usage for the working-age population. As of the second half of 2025, a Microsoft report found that 24.7% of the working population in the Global North use AI tools, while only 14.1% do in the Global South. Across 15 African countries, only 9% of the youth population aged 15–24 have basic computer skills, while the continent will require an additional 23 million STEM graduates to meet anticipated demand by 2030. There are also risks that labor and value chain inequality will deepen. For example, data workers and gig laborers face lower wages and have fewer protections than their counterparts in Global North countries. There is also minimal attention given to AI’s effect on informal workers, even though they make up around 61% of the world’s workers. Informal work is concentrated in the Global South—representing over 80% of the labor force in some places—and these workers will be uniquely impacted by AI, with great potential for technologies to connect informal workers to clients.
There are various opportunities to close the AI divide, including through South-South cooperation and investment in digital public infrastructure
South-South cooperation
South-South cooperation has become more prevalent given the shared constraints Global South countries face. It can lead to stronger shared solutions, whether it is pooling resources or sharing lessons learned. National strategies in Global South countries often include priorities centered on economic development and economic security, including building domestic AI agency and using AI to solve issues related to job creation, both of which open specific opportunities for focuses of South-South cooperation.
Going forward, South-South cooperation should move from diplomacy to implementation by advancing joint infrastructure, shared datasets, and technical working groups.
Renewable energy advantage
Some Global South countries have structural advantages. Brazil, for example, sources 88% of its electricity with renewables including hydroelectric, solar, wind, and biomass, making it a strong contender to become a green AI infrastructure hub. Meanwhile, Kenya and Nigeria lead in the energy and sustainability aspects of the United Nations Development Programme’s (UNDP) AI readiness assessments. Combining AI and renewables can become a competitive advantage to leapfrog legacy fossil-based systems.
Global South countries could align their existing and future AI and energy policies. The G20 could be a coordinating body to lead the way in aligning digital transformation with sustainability, perhaps through a G20 task force on AI-energy synergies that could develop and integrate global metrics for AI-related energy use and carbon emissions as well as coordinate global policies specifically related to these joint rules.
AI opportunities in the informal sector and job creation
AI could unlock greater visibility, productivity, and autonomy to informal workers by bringing new opportunities to workers through recommendation engines that track demand, translation services that open doors to new customers, or voice assistants that manage transportation routes and schedules.
Going forward, engagement with AI should prioritize innovative models that work together with, rather than ignore, the informal economy, given its dominance in the Global South. For example, regional integration efforts that seek to facilitate labor mobility could be boosted by prioritizing investments in digital identities that could allow informal workers to more easily move across regions while predictive analysis could help guide strategies with up-to-date information on skill demand and migration flows. Investing in STEM education, vocational training, and AI-specific capacity building will also be critical. Ensuring that the informal economy is represented and leveraged to meet joint government and public goals should also be a key focus. The UNDP is working to implement new targeted training for IT teams and university innovation centers in Africa that will target the strategic vocational training of the broader support jobs that will be needed, such as electricians, digital plumbers, and technical specialists.
Digital public infrastructure
New public infrastructure models have also emerged. South Africa’s Centre for High Performance Computing (CHPC) offers one regional example; it has a public ownership model that allocates access to researchers and institutions across the Southern African Development Community, pools demand, and shares access. India and Brazil have led the way in building digital public infrastructure (DPI) rails that, when integrated with AI, have become the foundation for digital AI sovereignty and opened new opportunities in areas like digital identity matching or translation tools for medical documents.
Global South countries should continue to build sovereign infrastructure. For example, India is pioneering the move to democratize access to AI infrastructure. Its IndiaAI Mission operates in a way where the government “funds the access to a public compute pool of approximately 18,000 GPUs” and sets the terms for who can access it for what price. Access for domestic startups, researchers, and academic institutions are subsidized, which lowers the barrier to entry. Countries like Saudi Arabia and the UAE are heavily investing in sovereign AI ecosystems. The Saudi Public Investment Fund has already allocated more than $40 billion to AI investments spanning models, software, and physical infrastructure including energy, data centers, semiconductors, and connectivity. Yet many Global South countries do not currently have the level of resources to implement such ventures, requiring different strategies to also achieve sovereign infrastructure. Already, some communities, such as the Indigenous Māori community in New Zealand, have refused to join initiatives aimed at including underrepresented languages in AI, citing distrust in technology companies in using indigenous knowledge for commercial projects. Instead, they are creating a framework that allows them to assert their own data sovereignty and govern their own data in collaboration with the government and private sector.
Redefining global governance and seizing leapfrogging opportunities
Orbital data centers are predicted to become a new strategy to overcome environmental challenges, with U.S. and Chinese companies already announcing plans to build these space-based data centers. While these exciting innovations could provide countries a cheaper path to data center ownership, they also create new governance challenges that will need to be addressed—particularly around data and sovereignty, as data would be moved beyond the reach of local regulators.
A more proactive approach centers the Global South in co-creating governance systems. The AI Impact Summit, hosted in India this year, offered a powerful way to translate the digital infrastructure achievements of Global South countries into tangible AI governance leadership. As countries continue their own sovereign infrastructure journeys, it will be critical to make space for global governance opportunities. For example, building a global architecture on AI which, according to Jake Yu, president of a global managing partner at Peregrine Ventures, could include a constitutional framework that defines common principles for AI, a global operating system of trust that enables interoperability, and a standing council for cooperative intelligence that prioritizes the alignment of national AI strategies, social policies, and private innovation concerns. The space between national strategies and global accountability must be filled with an inclusive, Global-South focused AI governance structure.
The AI divide is not inevitable or unchangeable. However, without strategic intervention to target challenges and leverage opportunities, the divide could harden into structural inequality. With the right strategies, the Global South can leapfrog legacy systems, build sustainable infrastructure, and shape global AI governance. Together, interested parties must work together to make sure that the Global South owns, governs, and defines the future of AI.
The Brookings Institution is committed to quality, independence, and impact.
We are supported by a diverse array of funders. In line with our values and policies, each Brookings publication represents the sole views of its author(s).