Idea
The Coming AI Divide
If the right decisions are not made today, artificial intelligence could deepen inequalities between nations faster and more deeply than the Industrial Revolution ever did.
Abstract
The UNDP's report The New Great Divergence delivers a blunt warning: the AI revolution is global, but its rewards are not. While nearly two-thirds of people in high-income countries regularly use AI tools, fewer than five percent do in low-income ones. A small cluster of economies — the United States, China, and the EU's core — controls the chips, the models, the data centers and the rules, while the rest of the world supplies the data, the users and the risk. This essay traces the new anatomy of inequality — infrastructure, skills, governance — and asks whether countries like Turkey will remain rule-takers in someone else's system, or build the hard infrastructure and soft capacity needed to become rule-makers in their own right.
5-Second Answer
AI is turning from an equalizing force into a dividing one: a handful of countries write the rules and capture the value, while most others supply the data and absorb the risk.
Key Arguments
- For decades the world believed in a convergence story — that globalization and technology would let poorer economies gradually close the gap with richer ones. AI threatens to reverse that story into a faster, more structural, more permanent divergence.
- The numbers are stark: nearly two-thirds of people in high-income countries use AI tools regularly, versus under 5% in low-income countries, with gender gaps in digital access reaching 50% in some regions.
- The Asia-Pacific region alone could see an extra $1 trillion in GDP from AI over the next decade, but the lion's share is projected to be captured by a small group of economies that control chips, model development, data-center capacity and regulatory power.
- The world is splitting into rule-makers, who design AI systems and write the standards, and rule-takers, who import and consume them, hand over their data, absorb embedded biases, and capture only a sliver of the economic value.
- 95% of global content flows through US-based platforms, roughly 70% of large language models operate primarily in English, and three-quarters of the world's data centers sit in just ten high-income countries — meaning the infrastructure of reality itself is concentrated in the hands of those already holding power.
Analysis
"If the right decisions are not made today, artificial intelligence could deepen inequalities between countries. This could happen faster and more deeply than the Industrial Revolution of the 19th century." This is the opening warning of the United Nations Development Programme's latest report, The New Great Divergence — a document plainly written to be a marker for history.
For decades, the world believed in a story of convergence. Globalization, trade and technology, people assumed, would gradually let low- and middle-income countries close the gap with high-income economies. That optimistic narrative has been shaking for a while. With the arrival of artificial intelligence as a new technological threshold, the challenge has intensified. Today the risk is not that the gap between nations is narrowing — it is that the divide is widening rapidly, and doing so faster, more structurally, and far more permanently than before.
The UNDP's warning is unambiguous: the decisions not made today will shape the map of inequality for generations to come.
The data is striking. In high-income countries, nearly two-thirds of the population regularly uses AI tools. In low-income countries, that figure falls below 5%. Income-based gaps in mobile internet use approach 40%, and in some regions women's access to digital tools lags by as much as 50%. The UNDP's core message is clear: the AI revolution is global, but access to its benefits is deeply unequal. Countries are not starting from the same line. At one end sit robust infrastructure, large-scale investment, competent institutions and advanced skill sets; at the other, fragmented connectivity, fragile governance and limited access. In short, AI is rapidly shifting from an equalizing force to a dividing one.
The report estimates that AI could generate more than $1 trillion in additional GDP across the Asia-Pacific region over the next decade. But the lion's share of that value is likely to be captured by a very small group — the United States, China, and perhaps the core economies of the EU. These are the actors who control the chips, the model development, the data-center capacity, the investment flows and the regulatory power.
Developing countries, meanwhile, supply the data, provide the users, and absorb the risks. A handful of global powers collect the value, set the standards, and control the rules. Does this not resemble a new form of technological colonialism? For humanity, this is a familiar equation — it echoes the 19th-century "Great Divergence" triggered by the Industrial Revolution. The UNDP's reference to a "technological dependency spiral" is therefore no coincidence.
The world is increasingly splitting into two camps: countries that develop and govern AI models, and countries that import and consume them. The first group designs the technology and writes the rules. The second group buys the systems, hands over its data, absorbs embedded cultural biases, and captures only a small fraction of the economic value produced.
Countries like Turkey sit closer to this second category. That is not yet destiny. But if it becomes permanent, it risks turning the middle-income trap into a digitally reinforced, concrete structure.
According to the report, 95% of global content flow passes through US-based platforms. Roughly 70% of large language models operate primarily in English. Three-quarters of the world's data centers are concentrated in just ten high-income countries. Security frameworks, ethical standards and governance norms are, likewise, drafted predominantly in these same geographies.
This is no longer merely a technological matter — it touches the infrastructure of reality itself, the cultural frame, and ethical norms. Risk management is part of this process too. And these dimensions are being shaped, to a large extent, by those who already hold power.
The UNDP's report is a clear alarm bell. A second great global divergence is approaching. The future will be shaped not by technology itself but by the collective choices of governments, businesses and societies. Two policy pillars stand out if fragmentation is to be avoided and AI's potential is to be turned into genuine social benefit.
The first is hard infrastructure: affordable devices, reliable connectivity, sufficient computing capacity, and secure digital identity systems. The second is soft capacity: skills development, strong institutions, transparent rules, competitive ecosystems, and governance frameworks that enable meaningful participation.
Ultimately, the goal must be to anchor the AI agenda in human development and social benefit. If that is achieved, artificial intelligence could become not an engine of unjust prosperity, but a cornerstone of a fairer, more inclusive future.
Is it difficult? Yes. Is it impossible? No.
Counterarguments
A skeptic might argue that every general-purpose technology — electricity, the internet, mobile telephony — initially concentrated in wealthy countries before eventually diffusing outward, and that AI will follow the same eventual diffusion curve rather than lock in a permanent divide. The UNDP's report takes this possibility seriously but argues the compute and data-center intensity of frontier AI creates far higher barriers to entry than earlier general-purpose technologies, making organic diffusion less certain. A second objection is that framing this as "colonialism" overstates the case, since developing countries retain sovereign choices about regulation, procurement and industrial policy. That is true in principle, but the report's point is precisely that these choices are being made under conditions of severe asymmetry — in standards, in leverage, and in the ownership of underlying infrastructure — which narrows the practical space for those choices considerably.
Implications
For governments in middle-income countries: treat compute capacity, data-center investment and digital identity infrastructure as strategic assets, not optional upgrades, since they are the hard prerequisites for any seat at the rule-making table. For regional blocs: pooling investment and bargaining power — rather than negotiating individually with a handful of dominant AI providers — is likely the only realistic path to shifting from rule-taker to rule-maker status within a generation. For educators and policymakers: skills development and institutional capacity building are not soft add-ons to an AI strategy; the UNDP frames them as one of exactly two pillars, alongside hard infrastructure, without which the divide becomes structural rather than temporary.
Related concepts
References
- Turkish original — Yapay Zekâ: Büyük Ayrışma Yaklaşıyor
- UNDP — The New Great Divergence (2025)
Signal, not noise.
A monthly briefing on AI, sustainability, and the future of human judgment — filtered from global reports, research, and emerging debates.
Join the BriefingMore ideas
Free Will, or Algorithmic Surrender?
How do you define freedom — the existence of options, or ownership of the decision? Are we deciding, or are we being decided upon?
Greenwashing Is Now a Matter of Proof, Not Morality
As of 1 August, an ordinary billboard reading 'eco-friendly' is no longer just a slogan in Türkiye — it is a legal claim that must travel with an auditable file behind it.