Media365
The global narrative on AI has been dominated by the generation of intelligence. Trillions of dollars are being invested in chips, data centres, and the computing infrastructure needed to train and run increasingly powerful models. The stock prices of a handful of leading AI firms in a few countries have surged.
The implicit assumption is that most of the economic value from AI will accrue to the companies that build and own these models, and to those upstream in the semiconductor, data centre, and energy value chains. This has increasingly led investors, firms, and policymakers to classify companies, markets, and even countries as winners or losers in the AI race.
But this may be too narrow a framing. Economic value is ultimately created not only by generating intelligence but by deploying it. The history of general-purpose technologies suggests that many of the largest gains accrue to the downstream users who adopt new technologies, adapt them to local conditions, and diffuse them throughout the economy.
India enters this phase from a position of strength. It already is one of the fastest-growing major economies and has repeatedly demonstrated an ability to deploy technology at population scale. Its experience with digital public infrastructure demonstrates how technology, institutions, and market incentives can combine to reduce costs, broaden access, and create platforms for innovation.
The World Bank Group’s ‘World Development Report 2026: The Promise of Artificial Intelligence’ argues that AI can help sustain growth, boost productivity, improve public services, and expand opportunity. But the largest gains are unlikely to come from winning a race to build frontier models. They will come from applying AI to solve practical problems at scale.
The question of jobs is central. Only 4.5 per cent of existing jobs in low- and middle-income countries appear highly exposed to automation by generative AI, compared with 14.2 per cent in high-income countries. Another 16.2 per cent could experience meaningful productivity gains, not far from the 18.7 per cent estimated for advanced economies. For most workers in developing countries, AI is more likely to complement than replace.
This distinction is important. For developing countries, affordable, task-specific AI tools can raise the capabilities of businesses, households, and governments. Be it helping farmers access advice, assisting health workers with diagnosis, supporting teachers, or improving public administration, the opportunity lies in AI raising productivity and expanding access.
Two challenges loom. The first is concentration. Advanced chips, large-scale compute, data, research talent, and risk capital remain concentrated in a handful of firms and countries. Competition has kept access to leading models relatively inexpensive, but dependence on a small number of providers creates risks and can eventually generate market power.
The second challenge is the lack of complements. AI is only as useful as the ecosystem around it. Its value depends on investments in electricity, connectivity, affordable devices and compute, local-language data, skills, and institutions. People need the skills to use AI effectively and governments need the capacity to evaluate, procure, and regulate it. Technology alone does not generate development gains.
India already has examples of AI tools generating tangible benefits. Applications adapted to local contexts support agricultural extension, improve weather forecasting, strengthen health diagnostics, and expand access to public services. Bharat VISTAAR, the Government of India’s digital platform for farmers, on which multiple states are building their own AI chatbots in multiple languages, and Wadhwani AI’s tuberculosis-screening tools illustrate the potential. Yet the key question is not whether these technologies work. It is whether they can be adopted by millions of people.
Here India’s experience with digital public infrastructure offers useful lessons. Platforms such as Aadhaar, UPI, DigiLocker, and emerging data-sharing systems provide shared public rails that lower costs, enable interoperability, and allow innovation to scale.
UPI is particularly instructive. Its open architecture promoted competition while avoiding dependence on a single provider. Similar principles should guide AI adoption. Governments and firms should favour systems that avoid technological lock-in and preserve portability, flexibility, and competition.
Emerging economies are unlikely to match the scale of investment being made by frontier AI hyperscalers, nor do they need to. More important is building strategic capabilities that expand India’s options to advance AI: Domestic research, Indian-language models, shared compute resources, and a stronger ecosystem for deep-tech investment. Initiatives such as IndiaAI, RDIF, and ANRF are important steps in that direction.
Three principles follow.
First, design for scale from the outset. Across India, AI pilots are everywhere. What is needed is successful deployment at scale. Every pilot should have a credible pathway to broad adoption, including financing, institutional ownership, and integration with existing systems.
Second, test rather than assume. Decisions should be driven by evidence. India’s scale makes rigorous evaluation both feasible and inexpensive. Large-scale rollouts provide opportunities to experiment, learn, and identify what works under real-world conditions.
Third, scale what works and stop what does not. Results, including failures, should be shared across states and ministries so lessons accumulate rather than being repeatedly rediscovered.
For India, the opportunity is not simply to adopt AI. It is to move from isolated applications to economy-wide productivity gains. That is the opportunity that matters: Maximising economic impact and spreading its gains as evenly as possible.
Mishra is World Bank executive director and Lall is director, Development Economics
