Can the Global South Leapfrog in the AI Era? Strategies for Convergence vs. Divergence

Introduction: The Double-Edged Sword of Digital Transformation

The Global South stands at a crossroads. Digital technologies—artificial intelligence, cloud computing, the Internet of Things, and platform economies—offer an unprecedented opportunity for developing nations to bypass traditional stages of industrial development and leapfrog into a data-driven economy. The promise is tantalizing: mobile money in Kenya, drone delivery in Rwanda, and telemedicine in rural India all demonstrate that latecomers can sometimes skip legacy infrastructure and adopt cutting-edge solutions directly.

Yet early evidence tells a more sobering story. As Ciuriak and Ptashkina (2019) documented in their policy brief for the G20, the digital transformation is creating a dual-track dynamic: while developing economies strive to catch up, advanced economies accelerate even faster due to first-mover advantages, high fixed costs, concentrated intellectual property, and deepening skill shortages. The result is not convergence but divergence—a widening digital divide that threatens to entrench structural dependency.

The thesis of this article is clear: success for the Global South requires a strategic mix of technology acquisition and deliberate data governance—not passive adoption. Without active policy intervention, the very tools that promise leapfrogging may instead become new instruments of economic subordination.

[IMAGE: A split-screen image: left side showing a bustling tech hub in Nairobi or Bangalore, right side a high-tech R&D lab in Silicon Valley, with a subtle arrow showing a widening gap.]

The Convergence Promise: Leapfrogging and Knowledge Spillovers

The concept of technological leapfrogging has a long pedigree in development economics. Developing countries can bypass expensive legacy infrastructure—landline telephones, brick-and-mortar banking, traditional retail—and adopt mobile money, digital payments, and e-commerce platforms directly. Kenya’s M-Pesa is the canonical example: it leapfrogged traditional banking infrastructure, bringing financial services to millions of unbanked citizens and spawning a vibrant ecosystem of micro-entrepreneurs.

Digitally Enabled MSMEs as Growth Engines

The policy brief by Ciuriak and Ptashkina emphasizes that digitally enabled micro, small, and medium enterprises (MSMEs) are critical drivers of inclusive growth in the Global South. Cloud platforms like AWS and Alibaba Cloud allow a small business in Lagos or Jakarta to access enterprise-grade computing power, while social media and messaging apps provide direct-to-consumer marketing channels that bypass traditional retail gatekeepers. The OECD’s 2016 report on the digital economy laid the foundational framework for understanding how digital technologies can lower barriers to market entry, reduce transaction costs, and enable new business models.

The Role of FDI and Knowledge Spillovers

Foreign direct investment (FDI) and international professional migration act as powerful conduits for tacit knowledge transfer. When multinational corporations establish R&D centers in Bangalore or Ho Chi Minh City, they bring not only capital but also management practices, quality standards, and exposure to global innovation networks. Similarly, the diaspora of engineers and data scientists who return to their home countries after working in Silicon Valley or Shenzhen carry with them skills and networks that can catalyze local ecosystems.

The key mechanism is knowledge spillovers—the unintended diffusion of technical and organizational know-how that occurs when workers move between firms, when local suppliers learn from demanding multinational clients, and when open-source communities share code and best practices across borders. These spillovers are the lifeblood of convergence.

[IMAGE: Infographic showing a spiral arrow from ‘old technology’ (landlines, cash) directly to ‘new technology’ (5G, digital payments) skipping intermediate steps.]

The Divergence Trap: First-Mover Advantages and Fixed Costs

For all its promise, the digital economy has structural features that tilt the playing field toward early adopters and wealthy nations. Understanding these features is essential for designing effective policy responses.

The Fixed Costs of Data and Compute

Advanced economies have already sunk the enormous fixed costs required for data capture, processing, and AI model training. Consider DeepMind’s AlphaGo Zero, which achieved superhuman performance in the game of Go through self-play. As Silver et al. (2017) documented, training AlphaGo Zero required 4.4 million games of self-play, each involving thousands of computations on specialized tensor processing units. The total compute cost was estimated in the millions of dollars—resources that no single research lab in the Global South can easily mobilize.

This pattern repeats across AI domains: large language models, computer vision systems, and recommendation algorithms all require vast datasets and compute clusters that concentrate in a handful of countries and corporations. The result is a growing asymmetry in AI capability between the Global North and the Global South.

Platform Colonialism and Winners-Take-Most Dynamics

Digital platforms exhibit strong network effects: the more users a platform attracts, the more valuable it becomes, creating a self-reinforcing cycle that concentrates market power. The "winners-take-most" dynamics of platforms like Google Search, Facebook, and Amazon mean that latecomers struggle to compete, even in their own domestic markets.

A stark illustration is Google’s Quayside project in Toronto. As Wylie and Scassa (2018) documented, the company proposed a smart city development where Sidewalk Labs would collect vast amounts of data on residents’ movements, energy use, and behavior. Critics raised concerns about data sovereignty, privacy, and the de facto ceding of urban governance to a private corporation. The project ultimately collapsed, but it highlighted how platform economies can replicate colonial patterns: the Global South provides raw data (the new resource) while the Global North owns the algorithms and the value extraction mechanisms.

Skewed IP Endowments

Intellectual property regimes further entrench divergence. Advanced economies hold the vast majority of AI-related patents, and patent thickets—dense webs of overlapping claims—make it difficult for latecomers to innovate without licensing fees or legal risks. Proprietary algorithms, trade secrets, and exclusive data sets create moats around the technology leaders. For developing countries, importing AI solutions often means accepting black-box systems that cannot be audited, adapted, or improved locally.

If policy interventions are delayed, initial divergence can become structural. The gap in digital infrastructure, human capital, and institutional capacity widens, making catch-up increasingly difficult.

[IMAGE: A scale with ‘Global North’ side weighted down by chips, patents, and data centers, and ‘Global South’ side light but with a small counterweight labeled ‘open platforms’.]

Strategic Interventions: Tipping the Balance Toward Convergence

The dual-track dynamic is not deterministic. Deliberate policy choices can tilt the balance from divergence toward convergence. Drawing on the G20 and OECD frameworks, and building on the insights of Ciuriak and Ptashkina, several actionable strategies emerge.

Open Platforms and Public Digital Infrastructure

One of the most powerful tools for leapfrogging is the development of open digital public goods. India’s Unified Payments Interface (UPI) is a textbook example: a government-backed, open-source platform that allows any bank or fintech to offer interoperable digital payments. UPI has enabled hundreds of millions of Indians to transact digitally, spawned a competitive ecosystem of apps, and kept data governance within the country. Similarly, Rwanda’s use of open-source drone software for medical deliveries demonstrates how public investment in shared infrastructure can bypass proprietary vendor lock-in.

Open platforms reduce fixed costs, enable local innovation, and ensure that data generated by citizens remains under domestic governance. For AI, this means investing in open datasets, shared compute clusters (such as the African Supercomputing Initiative), and open-source models that can be fine-tuned for local languages and contexts.

Strategic FDI and Conditional Technology Transfer

Not all FDI is equal. Developing countries should adopt differentiated strategies that prioritize investments involving genuine knowledge transfer, local R&D partnerships, and data residency requirements. The conditionalities attached to FDI—such as mandatory local hiring of data scientists, joint venture structures, and requirements to publish anonymized datasets—can ensure that spillover effects materialize.

South Korea’s early strategy of "learning by licensing" provides a historical template: firms acquired foreign technology under conditions that required them to build in-house R&D capabilities. Today, a similar approach could be applied to AI, where governments negotiate access to proprietary models in exchange for data-sharing agreements that respect sovereignty.

Data Sovereignty Frameworks

Data is the raw material of the AI economy, and its governance is the linchpin of any convergence strategy. Countries in the Global South must assert data sovereignty—the principle that data generated within their borders is subject to their laws and can be used for their economic benefit. This does not mean autarky; it means creating legal frameworks that ensure fair value capture.

The African Union’s Data Policy Framework and the Personal Data Protection Bill in India represent steps in this direction. Key elements include: requiring data localization for sensitive data (health, finance, biometrics), mandating algorithmic transparency for AI systems deployed in public services, and establishing data trusts or cooperatives that enable citizens and small businesses to collectively bargain with large platforms.

Building Human Capital and Research Ecosystems

Leapfrogging is impossible without skilled people. The Global South must invest heavily in STEM education, particularly in data science, machine learning, and software engineering. But technical skills alone are insufficient—policy entrepreneurs, ethicists, and legal experts who understand the intersection of technology and governance are equally vital.

International partnerships can accelerate capability building. Programs like the African Masters of Machine Intelligence (AMMI) and the Google AI Residency in Africa exemplify how targeted training and research collaboration can create a pipeline of talent. However, brain drain remains a persistent challenge; policies that create attractive local job opportunities, such as government-funded AI innovation labs and startup incubators, are essential to retain talent.

[IMAGE: A world map with arrows showing reverse flows of talent and technology from Global North to Global South, with icons of people, code, and data packets.]

Policy Anchors: G20 and OECD Frameworks

Two international policy frameworks provide important anchors for national strategies.

The OECD’s Digital Economy and AI Principles

The OECD’s 2016 report on the digital economy emphasized the need for inclusive growth, competition policy reform, and international cooperation on data flows. More recently, the OECD AI Principles (adopted in 2019) call for AI systems that are inclusive, transparent, and accountable. For the Global South, these principles offer a baseline for national AI strategies. Countries can use them to demand that AI developers provide explainability and fairness guarantees, especially when AI systems are deployed in critical areas like credit scoring, hiring, and criminal justice.

The G20 and Data Governance

The G20 has increasingly focused on data free flow with trust (DFFT), a concept promoted by Japan’s presidency in 2019. While DFFT aims to facilitate cross-border data transfers, developing countries must ensure that "trust" includes enforceable protections against exploitation. The G20 framework can be leveraged to advocate for capacity-building programs, technology transfers, and dispute resolution mechanisms that level the playing field.

[IMAGE: A diagram showing the G20 and OECD logos connected to arrows representing policy principles flowing into developing country contexts.]

Conclusion: Leapfrogging or a New Form of Dependency?

The AI era presents the Global South with a profound choice. The path of passive adoption—buying AI solutions from foreign vendors, surrendering data to global platforms, and importing algorithms without understanding them—leads to a new form of digital dependency. The same technologies that promise leapfrogging can become tools of structural subordination.

The alternative requires proactive strategy: investing in open digital public infrastructure, negotiating conditional FDI, enacting robust data sovereignty laws, and building a skilled workforce. It requires governments to act as stewards of their national data assets and as architects of inclusive digital ecosystems.

Leapfrogging is possible. The examples of M-Pesa, UPI, and Rwanda’s drone network prove that latecomers can innovate. But success is not automatic. Without deliberate data governance and a strategic approach to technology acquisition, the Global South risks trading one form of dependency for another. The window of opportunity is narrowing. The time to act is now.

[IMAGE: A stylized world map highlighting Africa, South Asia, and Latin America with glowing digital nodes and fiber-optic lines connecting them to developed regions; a faint bridge or leapfrog icon superimposed; warm gradients contrasting with cool blue in the Global North; no text or watermarks, clean 3D render or abstract illustration.]

---

References

• Ciuriak, D., & Ptashkina, M. (2019). Digital Transformation and the Global South: A Policy Brief for the G20.
• OECD. (2016). OECD Digital Economy Outlook 2015. OECD Publishing.
• Silver, D., et al. (2017). "Mastering the game of Go without human knowledge." Nature, 550, 354–359.
• Wylie, B., & Scassa, T. (2018). "Google’s Sidewalk Labs: Smart cities and the problem of data sovereignty." Canadian Journal of Law and Technology, 16(2).