AI

The United Nations Development Programme has issued a stark warning that should concern every enterprise leader, policymaker, and technology executive: artificial intelligence is poised to reverse five decades of declining global inequality, potentially triggering a "great divergence" between rich and poor nations that could destabilize the global economic order.

Released December 2, 2025, by UNDP's Asia and Pacific regional bureau, the report titled "The Next Great Divergence" presents the most comprehensive analysis yet of AI's differential impact on developed versus developing economies. The findings challenge the techno-optimist narrative that AI will lift all boats, instead revealing how the technology threatens to concentrate wealth, productivity gains, and strategic advantage in already-wealthy nations while leaving developing countries further behind.

For enterprise leaders operating in global markets, this isn't merely a humanitarian concern. As UNDP chief economist Philip Schellekens warns, "If inequality continues to rise, the spillover effects in terms of the security agenda, in terms of undocumented forms of migration, will also become more daunting." Translation: AI-driven inequality creates business risks through geopolitical instability, supply chain disruptions, and market access challenges.

The Convergence Era Is Ending

For the past 50 years, global inequality between countries has been declining. Developing economies grew faster than developed ones, narrowing income gaps and creating the rising middle class that became the target of every multinational corporation's growth strategy. China's transformation from agrarian economy to manufacturing powerhouse, India's emergence as a technology services center, Southeast Asia's integration into global supply chains—these success stories defined the convergence era.

AI threatens to end this. The UNDP report identifies several mechanisms through which AI amplifies existing inequalities rather than equalizing them:

Productivity Divergence

AI adoption correlates strongly with existing productivity levels, education quality, and governance capacity. Countries with strong institutions, educated workforces, and capital to invest in AI infrastructure will see productivity gains of 40-60 percent in key sectors. Those without these advantages will see minimal gains or actual declines as their workers are displaced without access to upskilling resources.

The math is brutal: if developed economies see 3-4 percent annual productivity growth from AI while developing economies see 0.5-1 percent, the GDP gap widens exponentially. A nation starting at half the per-capita GDP of a developed country will fall to one-third within 15 years, then one-fifth by 2050.

Data Colonialism

AI models require massive datasets for training. The countries and corporations that control these datasets—primarily US and Chinese tech giants—extract value from global data while developing nations provide the raw material with no ownership stake. This mirrors historical resource extraction, except the "resource" is behavioral data, language patterns, and cultural knowledge.

Consider: when a Western AI company trains models on data from developing nations, those models then compete with local workers and businesses in those same markets. The value created flows to model owners in San Francisco, Beijing, and Seattle. The displaced workers remain in Lagos, Jakarta, and Manila.

Infrastructure Inequality

AI requires computational infrastructure that developing nations cannot afford at scale. A single large language model training run can cost 10-100 million dollars in compute. Inference at scale requires data centers with reliable power, cooling, and connectivity. Developing nations face a choice: import AI services from foreign providers (creating dependency and data sovereignty issues) or fall behind competitors who do.

The Asia Pacific region accounts for more than 55 percent of global AI users despite having far less AI infrastructure than North America or Europe. This creates massive compute deficits that will only widen as AI capabilities advance.

Education and Skills Gap

AI's economic benefits accrue primarily to those who can work alongside it—programmers who use AI coding assistants, analysts who leverage AI for insights, executives who make AI-enhanced strategic decisions. These populations are concentrated in countries with strong education systems and technology literacy.

Workers in developing nations face a cruel dynamic: their jobs are among the most vulnerable to AI automation (manufacturing, data entry, customer service), yet they have the least access to education and reskilling programs that would allow them to transition to AI-augmented roles.

Regional Analysis: Asia Pacific at the Epicenter

The UNDP report focuses on Asia Pacific for good reason: the region hosts 55 percent of the world's population and more than half of global AI users, yet exhibits extreme variance in AI readiness across countries. This makes it the ideal lens for understanding AI's differential impacts.

The Winners: Singapore, South Korea, Japan

These nations combine strong governance, high education levels, technological infrastructure, and capital to invest in AI. They're positioned to capture disproportionate shares of AI's economic benefits:

Singapore has invested billions in AI research, established itself as an AI governance leader through its Model AI Governance Framework, and created incentives for AI adoption across industries. Productivity gains from AI could add 20-30 percent to GDP by 2035.

South Korea's semiconductor dominance positions it to profit from AI hardware demand. Its highly educated workforce and government-backed AI initiatives make it a leading AI exporter. Manufacturing automation will eliminate some jobs but create higher-value engineering and robotics roles.

Japan faces demographic challenges that AI can address. An aging population and labor shortages make AI adoption economically imperative. Robotics expertise positions Japan to lead in embodied AI and industrial automation.

The Middle Ground: China, India, ASEAN Core

These economies have scale, growing technical talent, and government support for AI, but face significant challenges:

China has made AI a national priority with massive state investment, but faces US export controls on advanced chips and mounting international distrust. Its AI industry is sophisticated but increasingly isolated from Western technology ecosystems. Internal inequality within China—coastal tech hubs versus inland rural areas—mirrors global patterns.

India has a large technology services industry and growing AI talent pool, but infrastructure gaps and education quality variance create

barriers to broad-based AI adoption. The country could capture 10-15 percent of global AI services revenue but risks seeing those gains concentrated in Bangalore, Hyderabad, and Mumbai while rural areas see minimal benefit.

ASEAN Core (Thailand, Malaysia, Vietnam, Philippines) is integrating AI into manufacturing and services but lacks the research capacity and capital of more developed neighbors. These nations risk becoming AI consumers rather than producers—importing technology dependency along with the models.

The Vulnerable: Bangladesh, Pakistan, Myanmar, Pacific Islands

Nations with limited infrastructure, governance challenges, and educational gaps face the starkest outcomes. AI threatens their competitive advantages (low-cost labor for manufacturing and services) while offering few compensating opportunities. They risk being entirely bypassed by the AI economy.

Consider Bangladesh's garment industry, which employs four million workers. AI-driven automation in textile manufacturing, combined with reshoring trends in developed economies, could eliminate millions of these jobs within a decade. Where do those workers go in an economy with limited technological capacity?

Economic Projections: Trillion-Dollar Opportunities, Existential Risks

The UNDP report includes economic modeling that reveals the stakes:

Potential GDP Gains

If AI is adopted effectively across Asia Pacific, the region could see nearly one trillion dollars in additional GDP over the next decade, with annual growth rates increasing by 2 percentage points. This translates to:

• 15-20 percent GDP increase in developed Asia Pacific economies by 2035
• 8-12 percent increase in middle-income economies with moderate AI adoption
• 2-5 percent increase in low-income economies with limited adoption

These differential gains widen absolute gaps dramatically. If South Korea adds 20 percent to a 1.7 trillion dollar GDP, that's 340 billion dollars. If Myanmar adds 3 percent to a 70 billion dollar GDP, that's 2 billion dollars. The relative gain is 170x larger for South Korea, even though Myanmar's percentage increase seems reasonable.

Current Applications Showing Promise

The report highlights several AI applications already delivering value in the region:

Remote Education: AI tutoring systems are improving educational outcomes in rural schools across Indonesia, Philippines, and Vietnam. Students in areas with teacher shortages now have access to personalized instruction. Early results show 15-25 percent improvement in standardized test scores.

Disease Detection: AI-powered diagnostic tools are expanding healthcare access in underserved areas. Chest X-ray analysis for tuberculosis detection, diabetic retinopathy screening, and malaria diagnosis via microscopy images are being deployed at scale. In India, AI-assisted diagnostics reduced specialist referral wait times from weeks to hours.

Agricultural Optimization: Computer vision and sensor fusion help smallholder farmers optimize irrigation, detect crop diseases, and predict yields. In Thailand, AI advisory systems increased crop yields by 18 percent while reducing water usage by 23 percent.

Credit Access: Alternative credit scoring using mobile data and transaction patterns is expanding access to business loans for small enterprises excluded from traditional banking. In Kenya, AI-driven microlending increased small business credit access by 40 percent.

Disaster Response: AI-enhanced weather prediction, flood modeling, and evacuation planning are strengthening disaster preparedness in cyclone- and typhoon-prone regions. Bangladesh's AI-augmented early warning systems reduced cyclone casualties by 65 percent between 2020-2024.

These applications demonstrate AI's potential to address developing world challenges. But they require infrastructure, data, and expertise that many countries lack.

The Migration and Security Nexus

Schellekens emphasizes that growing inequality won't remain contained within borders: "If inequality continues to rise, the spillover effects in terms of the security agenda, in terms of undocumented forms of migration, will also become more daunting."

Economic models predict that a 15-20 percentage point GDP gap increase between developed and developing Asia Pacific nations could drive:

• 25-40 million additional economic migrants seeking opportunities in developed economies by 2035
• Increased political instability in countries experiencing rapid job displacement without social safety nets
• Trade tensions as developed nations implement AI-driven productivity gains while developing nations struggle to compete
• Resource conflicts over rare earth minerals needed for AI hardware, with developing nations hosting deposits but developed nations controlling processing and manufacturing

This isn't speculation. History shows that major technological transitions (Industrial Revolution, agricultural mechanization, computer age) generated migration pressures and political instability. AI's speed and scale magnify these effects.

Policy Recommendations: Closing the Divergence

The UNDP report doesn't merely identify problems; it proposes actionable solutions. These recommendations should inform both government policy and corporate strategy:

1. Digital Infrastructure Investment

Developing nations need massive infrastructure upgrades to participate in the AI economy. The report calls for:

Regional data centers with renewable energy sources to reduce compute costs and carbon emissions. Shared facilities serving multiple countries can achieve economies of scale that individual nations cannot.

Connectivity expansion prioritizing fiber optic networks to underserved areas. 5G deployment in urban centers means nothing if rural areas lack basic broadband. Infrastructure-as-a-service models can accelerate deployment.

Energy grid modernization to support AI workload power demands. Many developing nations have unreliable electricity that makes sustained AI operations impossible. Solar microgrids and battery storage can address this.

International funding mechanisms through multilateral development banks to finance these capital-intensive projects. The World Bank, Asian Development Bank, and regional institutions must prioritize AI infrastructure as essential economic development.

2. Education and Skills Development

AI competency must become universal, not elite:

Curriculum reform integrating AI literacy from primary school through university. Students need to understand how AI works, its capabilities and limitations, and how to work alongside it. This isn't just programming—it's critical thinking about algorithmic decision-making.

Vocational training programs focusing on AI-augmented work rather than AI-replaced work. Welders who program robotic welders, nurses who operate AI diagnostic systems, farmers who use precision agriculture tools—these are the jobs that will survive.

Teacher upskilling initiatives providing educators with AI competency so they can effectively teach these concepts. Teacher quality is the bottleneck in most developing world education systems.

Open educational resources making AI education materials freely available in local languages. Translation and cultural adaptation are essential—Western AI curricula don't address developing world contexts.

3. Open Models and Technology Transfer

The report explicitly calls for more open-source AI models and technology sharing:

Open-weight models that developing nations can run locally rather than accessing through API calls to foreign companies. This reduces dependency and enables customization for local languages and contexts.

Hardware technology transfer allowing developing nations to manufacture AI chips domestically rather than relying on US, Chinese, and Taiwanese imports. Export controls currently prevent this, creating strategic vulnerabilities.

Collaborative research programs involving researchers from developing nations in AI advancement rather than treating them as passive consumers. Brain drain occurs when talented individuals must emigrate to access cutting-edge research.

Intellectual property frameworks that balance innovation incentives with access needs. Current patent and trade secret regimes concentrate AI capabilities in a handful of corporations and nations.

4. Governance Capacity Building

Effective AI adoption requires institutional capacity that many developing nations lack:

Regulatory frameworks appropriate to local contexts rather than copying Western or Chinese models. AI governance must address local challenges (informal economies, limited data protection) rather than imported priorities.

Public-private partnerships that bring together government, academia, and industry to coordinate AI strategy. Fragmented efforts waste resources that developing nations cannot afford.

Regional cooperation mechanisms allowing neighboring countries to share resources, pool expertise, and coordinate policy. ASEAN, SAARC, and Pacific Islands Forum can serve as platforms for collective AI strategy.

Anti-corruption measures ensuring AI procurement and deployment serves public interest rather than elite enrichment. Governance quality predicts AI adoption success more strongly than technical capacity.

5. Social Safety Nets and Transition Support

Job displacement will occur. The question is whether societies handle it proactively or reactively:

Universal basic income pilots testing whether cash transfers can support workers during transition periods. Several Southeast Asian nations are experimenting with this approach.

Wage insurance programs that bridge income gaps when workers move to lower-paid roles. The alternative—unemployment—is more costly economically and socially.

Retraining subsidies covering education costs for workers seeking to acquire AI-augmented skills. Means-tested programs ensure support reaches those who need it most.

Entrepreneurship support helping displaced workers start businesses using AI tools rather than competing with them. Microloans, business training, and market access programs can facilitate this transition.

Enterprise Implications: Why Businesses Should Care

Corporate leaders might wonder why their companies should concern themselves with global inequality. The answer is straightforward: AI-driven divergence creates direct business risks and opportunities that demand strategic responses.

Market Access Risks

If developing nations fall further behind economically, they become less attractive markets. The billion-person middle class that drove growth strategies for the past 30 years could shrink rather than expand. Companies that invested in emerging markets expecting continued convergence face strategic miscalculations.

Consider automotive manufacturers who built factories in Southeast Asia anticipating growing car ownership. If AI-driven job displacement reduces disposable income in those markets, demand projections fail. The factories become stranded assets.

Supply Chain Vulnerabilities

Many supply chains depend on developing nation labor and resources. AI-driven instability—migration pressures, political upheaval, resource conflicts—threatens these arrangements. Companies face choices:

Reshore operations using AI and automation to bring production back to developed markets. This reduces geopolitical risk but requires capital investment and faces political resistance in source countries.

Invest in stability by supporting education, infrastructure, and governance programs in supplier countries. This creates shared value but requires long-term commitment and multi-stakeholder coordination.

Diversify sourcing across multiple regions to reduce dependency on any single unstable area. This increases complexity and costs but provides resilience.

Talent Competition Intensifies

As AI amplifies productivity differences between workers, competition for AI-literate talent becomes fiercer. Companies in developing nations struggle to retain talent as developed-world firms offer remote positions at much higher wages. Brain drain accelerates.

Tech companies face a choice: pay premium wages to attract global talent (increasing costs), or invest in training local workforces in developing markets (building future capacity). Most will do both, but the mix determines long-term strategic positioning.

Regulatory Fragmentation

The report's emphasis on context-appropriate governance means companies will face increasingly divergent regulatory environments. Rather than a single global AI regulatory framework, we're heading toward regional variations that reflect local priorities and capacities.

This creates compliance complexity but also opportunities. Companies that help developing nations implement effective AI governance—through advisory services, technology platforms, or partnership models—position themselves as trusted partners rather than extractive corporations.

Reputational Considerations

Consumers and employees increasingly expect companies to address societal challenges. AI's role in widening inequality will become a reputational flashpoint. Companies that are seen as profiting from AI while ignoring its social costs face backlash.

Smart enterprises will:

Publish AI impact assessments examining how their AI adoption affects different stakeholder groups globally. Transparency builds trust.

Invest in transition programs helping displaced workers and communities adapt. This isn't charity—it's risk management and market development.

Advocate for policy solutions using corporate influence to support infrastructure investment, education reform, and technology transfer. Business leadership can accelerate or impede needed changes.

What Developed Market Enterprises Should Do Now

For companies headquartered in developed markets, the UNDP report demands strategic reassessment:

1. Audit Your Global Footprint

Map your operations, supply chains, and markets against the report's risk framework. Which countries do you depend on that face high displacement risk and low adaptation capacity? What's your exposure to instability in those regions?

Create scenario plans for different divergence trajectories:

Optimistic case: International cooperation and technology transfer enable broad-based AI adoption. Developing markets continue growing, creating opportunities.

Base case: Current trends continue with modest policy responses. Some developing nations succeed in AI adoption; others fall behind. Market segmentation increases.

Pessimistic case: Divergence accelerates with minimal international coordination. Instability spreads through migration, conflict, and economic disruption. Global supply chains fragment.

Stress-test your strategy against each scenario.

2. Invest in Local Capacity

If you operate in developing markets, invest in local AI capacity rather than just extracting value:

Fund education programs in communities where you operate. Partner with local universities, vocational schools, and training centers to build AI literacy.

Build local research capabilities rather than centralizing all AI development in headquarters. Distributed innovation creates more resilient organizations and demonstrates commitment to local development.

Share technology and expertise through open-source contributions, knowledge transfer programs, and collaborative projects. This builds goodwill and develops future talent pipelines.

3. Advocate for Enabling Policy

Use your policy influence constructively:

Support infrastructure investment through business advocacy groups and industry associations. Push for development bank funding of AI infrastructure in regions where you operate.

Oppose protectionist measures that prevent technology transfer and exacerbate inequality. Export controls on AI capabilities may serve national security goals but often harm developing nations.

Engage in multi-stakeholder initiatives bringing together governments, NGOs, and businesses to coordinate responses. The United Nations, World Economic Forum, and regional organizations provide platforms for this collaboration.

4. Prepare for Talent Mobility

Brain drain from developing nations will intensify as AI amplifies skill premiums. Companies face choices:

Remote work policies allowing you to employ talent globally while they remain in home countries. This reduces migration pressure and builds local capacity.

Competitive compensation acknowledging that high-value AI talent will have global offers. Geographic arbitrage is ending as remote work normalizes.

Talent development programs turning lower-skilled workers into AI-augmented high performers rather than only hiring already-trained experts. This expands the talent pool and demonstrates commitment to inclusive growth.

What Developing Market Enterprises Should Do Now

For companies headquartered in developing markets, the UNDP report is both warning and opportunity:

1. Accelerate AI Adoption

The convergence era relied on labor cost advantages. That advantage is eroding as AI automation reduces labor intensity. Developing market companies must adopt AI aggressively to remain competitive:

Identify high-impact use cases where AI can improve productivity, reduce costs, or enhance quality. Manufacturing process optimization, customer service automation, and supply chain management offer quick wins.

Build internal AI capabilities rather than depending entirely on foreign providers. Even if you use external AI services, understanding the technology is essential for effective deployment.

Partner strategically with AI providers that will share knowledge and build local capacity rather than extracting value. Scrutinize partnership terms—are you building capability or just licensing technology?

2. Lobby for Policy Support

Government policy will determine whether your market develops AI capability or falls behind. Use your business influence:

Push for infrastructure investment in connectivity, computing, and energy that enables AI deployment. Business groups should be the loudest voices demanding digital infrastructure.

Advocate for education reform that produces AI-literate workers. Your future workforce depends on today's schools. Partner with government on curriculum development.

Support open technology policies that allow your company to access AI tools and models without excessive foreign dependency. National security concerns shouldn't become protectionist barriers.

3. Address Workforce Transition

Your employees will be affected by AI displacement. Proactive management reduces disruption:

Upskilling programs teaching existing workers to use AI tools rather than replacing them. Humans augmented by AI remain more flexible and capable than fully automated systems.

Transparent communication about how AI will change roles and what skills will be valued. Uncertainty breeds resistance; clarity enables adaptation.

Gradual implementation that allows workers and systems to adjust rather than sudden switches that create chaos. Speed isn't the same as haste.

4. Build Regional Coalitions

Single developing nations struggle to compete with developed world AI capabilities. Regional cooperation changes the equation:

Shared infrastructure initiatives pooling resources to build data centers, research facilities, and connectivity that no single nation could afford alone.

Cross-border talent mobility within regions, allowing skilled workers to move where opportunities exist while staying within their cultural sphere. ASEAN, SAARC, and African Union frameworks can enable this.

Collective policy advocacy representing regional interests in international forums. A dozen nations speaking with one voice has more influence than a dozen separate voices.

The Path Forward: Convergence or Divergence?

The UNDP report makes clear that AI's impact on global inequality isn't predetermined. The "great divergence" is a trajectory, not an inevitability. Policy choices, corporate decisions, and international cooperation will determine whether AI widens or narrows global gaps.

Three scenarios illustrate possible futures:

Scenario A: Managed Convergence

International cooperation succeeds in enabling broad-based AI adoption. Technology transfer occurs, infrastructure investment flows to developing regions, and education systems adapt. AI's productivity gains are shared widely, though not equally.

In this scenario, global inequality stabilizes or modestly increases but doesn't spiral. Developing nations capture meaningful shares of AI economic value. Migration pressures remain manageable. Political stability holds.

Probability: 25 percent. Requires unprecedented international coordination and willingness of leading nations and corporations to accept reduced relative advantages. Possible but difficult.

Scenario B: Gradual Divergence (Base Case)

Current trends continue with modest policy responses. Some developing nations (China, India, core ASEAN, parts of Latin America and Africa) successfully adopt AI and maintain competitiveness. Others fall behind.

Global inequality widens moderately. Migration pressures increase but remain manageable through border controls and limited legal migration programs. Some political instability but not widespread conflict.

Probability: 50 percent. This is the default trajectory absent major policy interventions or disruptions.

Scenario C: Extreme Divergence

Minimal international cooperation, restrictive export controls, and inadequate infrastructure investment leave most developing nations unable to participate in the AI economy. Inequality grows dramatically.

Mass migration overwhelms border controls. Political instability spreads through affected regions. Supply chains fragment. Developed nations face security threats and loss of market access. The global order strains.

Probability: 25 percent. Would require policy failures and breakdown of international cooperation. Less likely than base case but far from impossible given rising nationalism and geopolitical tensions.

The UNDP report essentially argues for moving from base case toward managed convergence while warning that policy inaction could slide us toward extreme divergence.

Conclusion: The Business Case for Inclusive AI

The UNDP's "Next Great Divergence" report should be required reading for every CEO, CTO, and chief strategy officer in global enterprise. It demonstrates that AI's differential impacts create not just humanitarian concerns but direct business risks through market instability, supply chain disruptions, and talent competition.

The path forward requires:

Proactive investment in AI capacity building in developing markets where companies operate or source from. This isn't charity—it's risk management and market development.

Strategic advocacy for policies that enable broad-based AI adoption: infrastructure investment, education reform, technology transfer, and governance capacity building.

Collaborative approaches involving government, civil society, and business in multi-stakeholder initiatives addressing AI's distributional impacts.

Long-term perspective recognizing that AI's implications play out over decades, not quarters. Short-term profit maximization that ignores social stability is ultimately self-defeating.

Philip Schellekens summarizes the core insight: "We think that AI is heralding a new era of rising inequality between countries, following years of convergence in the last 50 years." Whether that happens depends on choices made now by leaders in government, business, and civil society.

The convergence era lifted billions out of poverty and created the global middle class that drove economic growth. Allowing AI to reverse that progress isn't just morally wrong—it's economically foolish and strategically dangerous. Smart enterprises will recognize this and act accordingly.

The question isn't whether AI will reshape global inequality. It's whether we manage that reshaping proactively or react to its consequences after the damage is done. The UNDP report makes clear that the time to act is now, while there's still an opportunity to influence outcomes rather than merely cope with them.

For further reading on AI's economic implications, see my prediction on AI workforce transformation reaching 40% by 2028, my analysis of enterprise AI adoption challenges, and my examination of AI governance frameworks.

---

Sources: United Nations Development Programme Asia Pacific Regional Bureau "The Next Great Divergence" report (December 2, 2025), interviews with Philip Schellekens (UNDP Chief Economist, Asia Pacific), economic modeling data, Al Jazeera coverage, Tech Startups comprehensive analysis.