The failure of multiple visits by the Japanese government to 111-year-old Sogen Kato’s house to honour him — he was never home — had a simple explanation. He had been dead for 30 years while his family collected his pension. This triggered a national investigation that identified 3 lakh people who were assumed to be “living” but were dead. India’s Aadhaar programme has reduced such pension scandals, improved the efficiency of government spending (up 117 times since 1991), and catalysed digital payments (42 per cent of the world’s transactions). The next orbit for digital public infrastructure (DPI) lies in AI integration — it will raise state capacity and empower the new tone from the top on Jan Vishwas.
The last few years have been tough for Indian tech. We don’t have a credible horse in the AI generation race, DeepSeek came out of China, and automated code generation has blunted the employment outlook for software services. But a recent conference at the National Council of Applied Economic Research (NCAER) suggested optimism about this digital revolution of AI for three reasons. India is the world’s largest natural data laboratory. Integrating DPI with AI could be the biggest reimagination of the Indian state since 1947. And AI deployment may be more employment-generative than AI generation. Let’s look at all three in detail.
First, India has gone from data poor to data rich in just 10 years because of three bold acts of entrepreneurship in 2016: Jio, UPI, and GST. For decades, India’s economic policy relied on periodic surveys and delayed, uneven and incomplete administrative data. But now, DPI’s digital exhaust makes us the world’s largest natural data laboratory for financial inclusion, welfare targeting, enterprise formalisation, platform economics and digital governance. UPI’s monthly 23 billion transactions create one of the world’s largest behavioural datasets. GST’s 20 crore payments create real-time visibility into supply chains, invoicing, production and consumption. FASTag’s 4.5 billion payment transactions create mobility and logistics datasets at scale. And,more than 27 billion annual Aadhaar authentications enabled a 16-fold expansion in direct benefit transfer beneficiaries. For the first time, a developing country is using population-scale digital data to accelerate development rather than waiting for conventional statistical systems to catch up. The job is hardly done — rigorous academic research, well-funded think tanks to bridge theory and practice, and open public data commons are just emerging — but the future seems very different from the past.
Second, India’s political reimagination since 1947 has created the world’s largest democracy on the infertile soil of the world’s most hierarchical society. But why didn’t we create the world’s largest economy? Besides the delusional 1956 Second Five-Year Plan, weak state capacity is another suspect in our failed transition from farm to non-farm jobs. Historically, governments increased state capacity through more administrative structures and personnel (government spending has increased 53,000 times since the 1947 budget of Rs 198 crore). DPI enables better resource allocation, improved service delivery, earlier fraud detection, and quicker responses. But a reimagined citizen interface has been blunted by a lack of interoperability (digital silos are stronger than physical silos), the government’s internal technology capacity, and a mentality that seems to be satisfied with PDFs, when the aim should be for APIs. AI’s capacity to use unstructured data will help us leapfrog data integration and the interoperability challenges within our strong consent architecture and the recently articulated principles of Jan Vishwas Siddhant. Multiple language interfaces will open up labour and education markets, and AI will improve public health expenditure. India will be the first large country to integrate AI into population-scale DPI — America built AI on private-sector data ecosystems, and China built AI on platform-centric digital ecosystems. DPI without AI is infrastructure, and AI without DPI is intelligence. AI plus DPI is enhanced state capacity.
Finally, India’s angst about AI’s impact on jobs is understandable if it ends up with the winner-take-all characteristics of search or the fiery nationalism that halts access to frontier models — the recently-revoked restrictions on frontier AI models such as Claude 5, for instance. But deploying AI in the daily lives of citizens and companies requires more ingenuity, organisational savviness, and human capital than AI generation itself does. You don’t have to build a car to drive it; let’s build the roads (AI applications) that get people to extract economic and social value. The deployment of AI in DPI could make India the use-case capital of the world, not unlike the American state that used defence spending during the Cold War to catalyse Silicon Valley. India’s enterprise DPI, currently under implementation (universal enterprise number, entity digilocker, API Setu, single source of truth for regulation), could potentially combine with a Universal Lifetime Social Security Account for every citizen (Aadhaar punji) to unleash a revolution in formalisation similar to what NPCI did in payments (private innovation on a non-profit layer). This would reduce information asymmetry, lower transaction costs, improve credit allocation and worker-job matching, enhance supply chain and logistics efficiency, and accelerate the creation of high-wage, private, non-farm jobs.
The Arthashastra suggests that strategy is most powerful when choices in strength, timing, and place are mutually reinforcing. Combining our strengths in DPI with the new tone from the top about Jan Vishwas (NCAER research suggests 66 per cent of connected Indians use the internet for entertainment but only 11 per cent for online government services), could deliver our new tryst with destiny: Higher state capacity, citizen satisfaction, and mass prosperity. And this is an appointment we shall keep.
Sabharwal is co-founder of Teamlease Services and co-author of Made in India. Bapna is Curtis L Carlson Chair Professor in Business Analytics and Information Systems at University of Minnesota
