Asia has long sold the world a compelling story, the story of convergence. Decades of export-led growth, technology transfer, and regional integration seemed to confirm that poorer economies could catch up with richer ones if they played their cards right. The Asian Development Bank’s latest assessment of artificial intelligence preparedness suggests that the story is about to be rewritten. And not in a hopeful direction.
Key Points
- Infrastructure Deficits: Advanced economies possess superior digital foundations, while developing nations suffer from “cascading constraints” regarding connectivity, computing power, and data accessibility, effectively capping their potential for AI adoption.
- Human Capital Gap: There is a significant disparity in AI-ready workforces; leading economies are aggressively hiring AI-adjacent talent, whereas developing economies struggle to cultivate the necessary skills, leading to an acceleration of the divide.
- Innovation and Sovereignty: Developing nations are largely dependent on importing AI tools designed for foreign contexts, which fails to address local languages and regulatory needs, potentially marginalizing them further in global value chains.
- Institutional Challenges: Weak governance, ambiguous data protection laws, and unpredictable regulatory environments in developing nations act as deterrents to the investment required to build AI capacity.
- Economic Divergence: Economic modeling predicts that by 2030, the GDP growth gap between advanced and developing Asian economies will widen significantly, making it increasingly difficult for the latter to catch up.
The ADB’s findings are stark. Generative AI is poised to deliver enormous productivity gains across the Asia-Pacific region, but those gains will flow overwhelmingly to the economies that are already the most advanced. The gap between winners and laggards is not narrowing. It is being locked in.
The Infrastructure Wall
The numbers tell a blunt story. Advanced economies, Australia, Hong Kong, Japan, South Korea, New Zealand, Singapore, cluster near the global frontier on the ADB’s AI Preparedness Index, scoring an average of 0.19 on the infrastructure component. Meanwhile, Cambodia, India, Myanmar, Papua New Guinea, and the Philippines score below 0.11. That may look like a modest numerical gap. It is anything but.
Digital infrastructure is not merely a prerequisite for AI adoption. It is the ceiling that caps ambition. Limited connectivity means limited access to cloud services. Limited computing capacity means firms cannot train or deploy AI systems at scale. Underdeveloped data infrastructure means the raw material of the AI economy, data itself, remains trapped and underutilized. Each deficiency compounds the next. The ADB describes these as “cascading constraints,” and the language is apt: once you fall behind, the slope steepens.
The uncomfortable truth is that infrastructure gaps of this magnitude cannot be bridged by goodwill or ambition alone. They require capital, coordination, and time, all of which are in short supply precisely in the economies that need them most.
Skills: The Hidden Bottleneck
Infrastructure is only half the problem. Even where connectivity improves, AI delivers nothing without workers who can use it. Here, too, the divide is sobering.
Developing Asia and the Pacific average just 0.13 on the ADB’s human capital and labor market readiness index, compared to 0.17 for advanced economies. Job posting data amplifies the concern: demand for AI-related skills is growing faster and from a higher base in Singapore and South Korea than in India, Malaysia, or the Philippines. This is not a story of developing economies falling behind. It is a story of advanced economies accelerating away.
Firms in leading markets are not waiting for their workforces to catch up organically. They are hiring aggressively for AI-adjacent talent, building organizational capabilities that will make them exponentially more productive in the years ahead. Firms in developing economies, by contrast, are still navigating basic questions of digital readiness. The compounding effect of this difference will be felt for a generation.
The Innovation Ecosystem Gap
Perhaps the most structurally damaging finding in the ADB report concerns innovation capacity. China, Japan, and South Korea benefit from strong government support and substantial corporate R&D investment that enables both AI development and local adaptation, the ability to tailor models to domestic languages, markets, and conditions. This is enormously valuable. A general-purpose AI tool trained overwhelmingly on English-language data is of limited use to a Khmer-speaking smallholder or a Filipino entrepreneur navigating local regulatory complexity.
Developing economies, by contrast, rely heavily on imported AI technologies. They are consumers of tools built elsewhere, for contexts that often do not reflect their own. This dependency is not merely an economic problem. It is a sovereignty problem. Nations that cannot adapt AI to their own languages and conditions will find that the technology reinforces rather than reduces their marginalization in global value chains.
The participation gap in AI-related global value chains, electronics, semiconductors, and computing equipment tells a similar story. Singapore and Hong Kong are deeply integrated; most of developing Asia is not. The technology spillovers that flow through these chains will bypass economies that sit outside them.
Governance: The Overlooked Enabler
One element of the ADB’s analysis deserves particular attention because it tends to be underestimated: institutional quality. Developing Asia scores 0.12 on regulatory and ethics frameworks, compared to 0.20 in advanced economies. Regulatory uncertainty, weak enforcement, and gaps in data governance do not merely create legal risk. They actively suppress investment.
A company weighing where to deploy AI infrastructure will not choose a jurisdiction where data protection is ambiguous, enforcement is unpredictable, and governance frameworks are opaque. The result is a vicious cycle: weak institutions deter investment, which slows technology adoption, which reduces the urgency of building better institutions.
The good news, if there is any, is that governance reform is cheaper than infrastructure investment and faster than educational transformation. Countries that move decisively to establish credible, transparent AI governance frameworks will make themselves meaningfully more attractive to the capital and expertise they need.
The Growth Projections Are a Warning, Not a Forecast
The ADB’s economic modelling is where the stakes become undeniable. Advanced Asia and the Pacific, along with the United States, could see GDP growth increase by 0.6 to 2.1 percentage points by 2030. Developing Asia, constrained by weaker infrastructure, lower skills, and structural exposure concentrated in agriculture rather than AI-amenable services, is projected to gain 0.2 to 1.8 percentage points over the same period.
These are not trivial differences. Compounded over decades, divergences of this magnitude reshape the relative position of economies in ways that are extraordinarily difficult to reverse. The ADB notes that catch-up effects could add up to 0.4 percentage points in some cases, an asterisk that should not be mistaken for reassurance. Catch-up assumes access, capability, and will. For many developing economies, all three are currently constrained.
Notably, China is expected to record the largest gains among developing Asian economies, followed by India. This matters because it suggests that even within the developing world, concentration of benefits is likely. The largest, most institutionally capable emerging economies will capture the most; the smallest and most structurally vulnerable will capture the least.
What Must Be Done
The ADB’s policy prescriptions are sensible: invest in digital infrastructure, reform education systems, build innovation ecosystems, establish credible governance frameworks, strengthen social protection, and integrate into AI-related global value chains. The bank is right that targeted reforms can mitigate short-term displacement risks while enhancing long-term productivity gains.
But the scale of simultaneous action required should not be underestimated, and the window for action is narrowing. AI is not a future technology arriving at a pace that permits leisurely preparation. It is being deployed now, in firms and economies that are already positioned to use it. Every year of delay widens the gap that must subsequently be closed.
The political challenge is equally formidable. Governments in developing Asia face competing demands, such as health, education, poverty reduction, and climate adaptation, that make it difficult to prioritize AI readiness investments that may yield returns only over a decade or more. International institutions, including the ADB itself, have a critical role to play in financing, technical assistance, and creating the multilateral frameworks that allow smaller economies to participate in AI governance discussions rather than simply accepting terms set by others.
A Test of Whether the Asian Century Belongs to All of Asia
The promise of the “Asian Century” was always partly a promise of shared prosperity, of a region rising together rather than merely producing a new hierarchy of winners and losers. The ADB’s findings suggest that the promise is at serious risk.
Generative AI is not inherently a force for inequality. Used well, it can extend high-quality services, education, healthcare, financial advice, and legal information to populations that have historically lacked access to them. The technology’s democratizing potential is real. But potential is not destiny. Whether that potential is realized depends entirely on policy choices, investment decisions, and institutional quality that vary enormously across the region.
The continent is at an inflection point. The path of least resistance leads to a two-speed Asia, where a handful of advanced economies pull further ahead while much of the developing region watches the productivity revolution happen somewhere else. Avoiding that outcome will require urgency, resources, and political will that have not yet been fully marshalled.
The ADB has documented the problem clearly. The harder question is whether the response will match the scale of what is at stake.
Other People are Reading
You must be logged in to post a comment Login