AI debate India needs: sovereign AI vs. AI's impact on the economy

EDITOR’S NOTE

Dear future-proof humans,

Welcome to another edition of our newsletter. I am especially excited about today’s newsletter. Most weeks, we break down new AI tools or show you how to build practical workflows. But going forward, alongside that, we are also going to write about the bigger picture, what AI means for India as a society and an economy.

Debates often circle around who builds the models—whether India needs a DeepSeek of its own, or if sovereign AI is necessary for strategic security. These questions matter, but they shouldn’t eclipse the deeper one: how will AI change the lives of India’s 1.4 billion people? Regardless of whether we are running models built in the US, China, or elsewhere, the reality is that generative AI is already accessible to the masses. Its impact will be felt in classrooms, hospitals, farms, courts, and offices across the country.

So instead of asking only “whose engine are we using?”, it is time to ask: what does this engine do to India’s society and economy? That is the story we take up in this edition.

Why AI Diffusion Matters for India

Technological revolutions often reshape economies in ways that become clear only in hindsight. Steam engines powered the Industrial Revolution, electricity transformed factories and households, and the spread of the internet redefined communication and commerce. Artificial intelligence is now seen as the next such turning point. Unlike earlier tools, it extends beyond machines performing physical work to software handling cognitive tasks, an ability once thought unique to humans.

Globally, economists compare the spread of AI to the arrival of electricity: a general-purpose technology that will cut across sectors, raise productivity, and alter how businesses and governments operate. But like past shifts, it brings both promise and disruption. Advanced economies are already debating whether AI will revive stagnant productivity or deepen inequality. For emerging markets, the challenge is sharper: they must harness the gains without leaving large parts of their workforce behind.

India sits at the center of this debate. Its scale alone makes it crucial: with over 1.4 billion people and the world’s largest youth population, the stakes of how AI diffuses here will be felt far beyond its borders. India’s economy is also unusually diverse. IT services drive global outsourcing, agriculture still employs over 40% of the workforce, and manufacturing is a government priority under “Make in India.” This mix means the impact of AI will not be uniform, but rather sector-specific, affecting farmers, coders, factory workers, and bureaucrats differently.

At the same time, India has advantages. The country leapfrogged in mobile adoption, created digital public goods like Aadhaar and UPI, and built a thriving startup ecosystem. These successes suggest that India can adapt to transformative technologies when infrastructure and policy align. Yet challenges remain: low female labor force participation, uneven digital access, and a skills gap that could widen if AI adoption accelerates.

There is also precedent for success. India built a world-class IT services industry without inventing microprocessors or programming languages. By applying existing technologies to global business problems, India became an outsourcing hub, created global competency centers, and later adapted internet and smartphones into new e-commerce models. AI could follow the same pattern: adoption, adaptation, and scaling to Indian realities.

Today’s newsletter explores how AI diffusion may reshape India’s economy. It will examine its effects on productivity, growth, and inequality; the pressures and opportunities in the labor market; sectoral shifts; and the policies that could tilt the balance toward inclusive progress.

How does AI impact work and the economy?

But before we turn to growth numbers or job impacts, it helps to ask a basic question: how does AI actually impact work? I want you to think of jobs not as single blocks that vanish or survive, but as bundles of tasks that AI can automate, augment, or leave untouched.

Most occupations combine routine, repetitive activities with tasks that demand judgment, creativity, or interpersonal skills. AI is powerful at automating certain tasks, but far less capable of replacing the full spectrum of a job. This makes the impact more nuanced than the simple story of “machines taking over.”

Take India’s IT services sector, for instance. A software engineer’s job spans writing code, testing, documenting workflows, and working with clients. Generative AI can already accelerate coding and automate documentation. Yet the consultative work of solving client-specific problems remains human-led. The role does not disappear; it changes shape, with more time for higher-value work.

Agriculture presents another angle. Farmers make decisions about weather, crop cycles, and pest control. AI-backed apps can forecast rainfall or warn of infestations, but the labor of cultivation and marketing still rests on people. Here, AI serves as a decision-support tool, especially for smallholders who dominate India’s farming landscape.

Customer service shows the sharper edge of automation. Chatbots can answer delivery queries or reset passwords, reducing demand for routine call-center roles. Yet complex complaints still require human empathy and negotiation, keeping those tasks in the human domain.

Thinking in terms of tasks helps avoid alarmist predictions. It highlights the real challenge for India: helping workers transition from the routine tasks most exposed to automation toward the complementary ones where human expertise continues to matter.

AI in India should be framed less as a Western debate about job losses and more as a strategic opportunity. The real test is not whether AI displaces roles, but whether it can fix long-standing governance and service gaps. India’s challenge is to apply AI in ways that strengthen education, healthcare, justice, and public services, rather than only counting which jobs are automated.

Macroeconomic impact: Productivity, growth, equality

If AI changes work by reshaping tasks, the natural next question is: what does this mean for the economy as a whole? Economists usually focus on three outcomes: productivity, growth, and inequality. On each of these, AI diffusion offers India both opportunities and risks.

Productivity and growth
Global studies suggest that nearly 40% of all jobs worldwide are exposed to AI, with the share rising to 60% in advanced economies. In India, the figure is lower, about 26%, because a large part of the workforce is in low-exposure occupations like farming and crafts. At first glance, this may look like insulation from disruption, but it also means slower adoption of AI’s productivity benefits.

Where adoption does take place, the potential is large. Estimates suggest AI could add $450–500 billion to India’s GDP by 2025, equal to 1.2–1.5%age points of annual growth. This would be comparable to the productivity surge India experienced during its IT-led expansion in the 2000s. Bruegel’s global analysis similarly projects that widespread AI use could lift world GDP by about 7% over the next decade.

Still, history warns us not to expect these gains overnight. Most technologies follow what’s called the “hype cycle”: a burst of inflated expectations, a trough of disillusionment when reality falls short, and then a slower climb toward the plateau of real productivity. AI is likely to be overestimated in the short term but underestimated in the long term. For India, this means that short-term disappointment should not derail investments in the infrastructure and skills needed for the long haul.

Inequality concerns
The distribution of these gains is less even. Unlike past waves of automation, which mostly displaced middle-income routine jobs, AI exposure is actually higher among educated and higher-income workers. In India, urban professionals in IT, finance, and consulting are already adopting AI to raise efficiency. The IT and business process outsourcing sector, which employs 4.5 million people, is a case in point. Up to 60% of coding and testing tasks could be automated, but the firms that embrace AI may expand their global competitiveness. Workers who adapt could see higher wages, while those who do not risk being sidelined.

Meanwhile, rural workers and informal-sector employees remain in low-exposure occupations, especially agriculture, which still employs 43% of India’s workforce. Their jobs face less immediate automation but also fewer productivity gains. IMF data show that higher-income deciles in India are far more likely to be in high-complementarity roles, suggesting that without policy action, AI could widen both wage and wealth inequality.

Global comparisons
Advanced economies face sharper disruption but are better prepared, with high digital penetration and safety nets. India, with lower exposure, has more time to adapt but must act decisively to avoid being locked out of the productivity gains. In many ways, AI diffusion mirrors the story of mobile phones: India lagged in early adoption but later leapfrogged to near-universal coverage. Whether the AI story follows the same arc will depend on how quickly adoption spreads beyond urban elites into the wider workforce.

Labour market impact: Displacement, reskilling, and new expertise development

The macroeconomic story of productivity and growth is incomplete without looking at who bears the adjustment costs. If AI adoption boosts GDP but leaves large groups of workers displaced or excluded, the benefits will be uneven. For India, the labor market impact is central, given the size and diversity of its workforce.

Who is most exposed?
India’s IT and business process outsourcing sector, employing 4.5 million workers, is on the front line. Routine tasks such as coding, testing, and documentation are highly exposed; studies suggest up to 60% could be automated. Yet this is also a sector where complementarity is strong: workers who adapt can use AI tools to enhance productivity and command higher wages.

Manufacturing shows a different pattern. AI-driven quality control, predictive maintenance, and supply-chain optimization may change shop-floor operations. The impact here is less about displacement and more about upskilling machine operators into supervisors of AI-enabled systems.

Agriculture, which employs 43% of Indian workers, remains low-exposure. Most farm tasks are physical and not easily automated by current AI tools. Instead, AI augments decision-making through pest detection, weather forecasts, and market linkages. Displacement risk is low, but productivity gains will be limited unless digital infrastructure reaches rural areas.

Gender divides matter too. Globally, women are more likely to work in high-exposure occupations such as clerical roles. In India, where women are already underrepresented in formal employment, this raises the risk of further exclusion. Without targeted skilling, AI adoption could widen gender gaps.

Reskilling opportunities
The good news is that India’s young workforce is relatively adaptable. Historical data show that college-educated workers are more likely to transition into new high-complementarity roles. For India, this highlights the importance of reskilling programs that help workers shift from repetitive tasks to areas where human expertise, creativity, empathy, and complex problem-solving remain vital. Initiatives like the government’s Skill India Mission and corporate training platforms will need to scale rapidly.

The transition will not be painless. Older workers and those in the informal sector may find it harder to retrain or switch jobs. This makes social safety nets and flexible training pathways critical to avoid deepening inequalities.

As India navigates AI diffusion, the labor market question is not whether jobs will vanish, but how tasks will change, and whether workers can change with them.

Sectoral opportunities and risks in India

If the labor market impact shows us who is most exposed, the next layer is to ask where AI will matter most. India’s economy is a patchwork of sectors with very different levels of digital maturity, so AI diffusion will not unfold evenly. Some areas are primed for rapid gains, while others face barriers that slow transformation.

IT and services
This is the sector most immediately touched by AI. Coding assistants, automated testing tools, and AI-driven customer service systems can cut costs and raise efficiency. For India’s $250 billion IT and BPO export industry, this is both a threat and an opportunity. Firms that adapt may win new global business, but entry-level white-collar roles, the classic pathway into the middle class, are at risk of shrinking.

Agriculture
Employing over 40% of India’s workers, agriculture remains low on direct AI exposure. Yet the opportunity lies in productivity. Tools that deliver localized weather forecasts, pest alerts, and price data can improve yields and incomes. Pilot projects using AI-enabled soil sensors and image recognition for crop health show promise, but adoption depends on affordable access and farmer training.

Manufacturing
India’s “Make in India” push aims to lift manufacturing’s share of GDP. Here, AI applications in predictive maintenance, supply-chain optimization, and quality assurance can reduce downtime and waste. The risk is that many small and medium manufacturers lack the capital and skills to implement these systems, leaving gains concentrated in larger firms.

Healthcare
India faces chronic doctor shortages, particularly in rural areas. AI can bridge gaps through diagnostic support, triaging, and telemedicine. For instance, AI-enabled imaging tools are being piloted to detect tuberculosis and eye diseases. The risk lies in regulation and trust: patients may resist machine-driven advice unless backed by strong clinical oversight.

Education
AI tutoring systems can personalize learning and support teachers in large classrooms. In a country where student-teacher ratios are often 40:1 or higher, this is a clear opportunity. Remember India’s first GenAI teacher in Kerala, named Iris!

But unequal access to devices and connectivity could worsen existing divides between urban and rural schools.

Public administration
India’s judicial backlog is infamous, with millions of pending cases. AI could help by summarizing documents, identifying precedents, and reducing inconsistencies across courts that operate in multiple languages. This goes beyond efficiency; it can widen access to justice. Similarly, AI chatbots could improve government service delivery compared to the “press 1, press 2” systems that frustrate citizens today. The risk is overreliance without adequate safeguards for fairness and bias.

Together, these sectoral snapshots highlight a simple truth: AI in India is not one story but many. Each sector carries its own blend of leapfrog potential and structural constraints.

Four futures for India’s AI decade

If sectoral differences explain where AI matters, scenarios help us think about how the technology might diffuse. The IIC–Bruegel framework lays out a simple two-by-two: the pace of change (incremental vs transformational) and the role of technology (automation vs augmentation). This creates four futures for the decade ahead.

Scenario 1: Transformative Augmentation
This is India’s best-case path. AI enables people, rather than replacing them, to boost productivity across sectors. For India, the advantages include strong growth in tech industries, better governance capacity, and a young population that can adapt quickly. But such a leap would strain education, welfare, and urban systems, especially if millions of service workers are displaced at once.

Scenario 2: Incremental Augmentation
Here, AI adoption is gradual but still focused on enabling workers. Gains are steady, social disruption is moderate, and countries get more time to adjust. For India, this scenario means more breathing room for reskilling and institution-building. The downside is fewer opportunities to leapfrog, risks of rural stagnation, and stronger competition from other developing countries.

Scenario 3: Transformative Automation
This is the nightmare path: rapid AI adoption that replaces workers instead of enabling them. Globally, it could bring broad productivity gains and dramatic innovation, but it is concentrated in a few firms and countries. For India, the costs are severe: an employment crisis, social instability, and even the risk of de-industrialisation if labor-intensive sectors collapse.

Scenario 4: Incremental Automation
This path is slower-moving but still automation-heavy. It brings smoother transitions and more time for adaptation. For India, the pressure on governance is lower, but growth remains sluggish, youth underemployment persists, and traditional migration patterns from farms to cities may be disrupted.

What this means for India
Transformative AI creates “winner-takes-all” dynamics, either inclusive growth through augmentation (Scenario 1) or severe dislocation through automation (Scenario 3). Incremental AI, meanwhile, risks leaving India stuck in a middle-income trap. I too lean toward transformative augmentation as India’s best bet: ambitious enough to maximize its demographic dividend, but with policies to manage education and infrastructure gaps.

Policy directions for India

The next big question is what India can do to tilt outcomes toward inclusive growth. Policy will be decisive. Unlike advanced economies, India cannot rely on existing safety nets or universal high-speed access. It must design its own path for AI diffusion.

Digital infrastructure is the foundation. India’s success with Aadhaar and UPI shows how public digital goods can scale. Extending affordable broadband and cloud access into rural areas will be essential to ensure AI tools do not remain an urban privilege.

Skilling and reskilling come next. With 65% of the population under 35, India has a demographic advantage, but the skills gap is wide. Government missions like Skill India must expand beyond traditional trades to cover AI literacy, coding, and data handling. Equally important is continuous upskilling for mid-career professionals in IT and services, where AI exposure is highest.

Regulation should balance innovation and trust. Overregulation could stifle startups, while lax oversight risks bias and misuse in sensitive areas like healthcare or judicial applications. Clear frameworks for data privacy, algorithmic accountability, and consumer protection will help build confidence.

Safety nets are critical for displaced workers. Unlike Europe or the US, India lacks widespread unemployment benefits. Targeted wage subsidies, conditional cash transfers, or retraining stipends could soften the shock for those whose jobs are restructured by AI.

Private–public collaboration will shape outcomes. Large IT firms are already piloting AI adoption, but small and medium enterprises risk being left behind. Policy can encourage partnerships that extend AI tools to SMEs, much like how fintech expanded through digital payments infrastructure.

Global indices, such as the Oxford AI Readiness Index, rank India in the middle tier, above many peers but behind advanced economies. To close this gap, India will need not just technology adoption but institutions that can manage its social consequences.

India at the crossroads

India’s AI journey sits at a fork in the road. One path leads to inclusive growth, where productivity gains spread across sectors, workers are reskilled, and digital tools reach rural as well as urban citizens. The other risks widening divides, with benefits concentrated among elites while millions remain untouched by transformation.

The outcome will depend on choices made today by policymakers, companies, and individuals.

1. Will AI in India be a leapfrog moment like mobile phones, or will it entrench existing inequalities?

2. Can reskilling efforts keep pace with the speed of automation?

3. And how can India ensure that its 500 million rural workers share in the gains of this new technology?

Hope you enjoyed reading the essay. I want to leave you folks with two exercises for reflection.

• First, revisit the 1970s–80s case of “computers everywhere except in the productivity statistics.” Why did computer adoption fail to show clear gains for so long, and what lessons might that hold for AI today?

• Second, read Mike Judge’s Where’s the Shovelware? Why AI Coding Claims Don’t Add Up. If AI is so ubiquitous, why hasn’t the number of new apps exploded?

Both cases remind us that technology’s promise does not automatically translate into economic outcomes; it takes time, adaptation, and human choices.