Artificial Intelligence (AI) is best understood not as a sector but as a general-purpose technology. Like electricity or broadband, it has economic value in lowering costs across many activities rather than transforming a single industry. Countries that gain the most will not simply be those that build powerful models, but those that ensure that technology becomes widely usable. For India, that shifts the real challenge to diffusion: expanding shared computers, institutionalising data platforms, and deploying AI across sectors where productivity and public outcomes can scale.
India enters this phase with an unusually strong base. Its digital public infrastructure already operates at a population scale. Nearly 575 million Jan Dhan accounts, more than 21 billion UPI transactions each month, and direct transfers reaching over 1.86 billion beneficiaries, show that digital rails now underpin everyday economic activity. AI is beginning to sit atop this stack. The issue is not whether the country will adopt technology, but how broadly it can be integrated into firms, research systems and public delivery networks.
Three structural strengths shape this trajectory: data and talent density, techno-legal architecture, and shared infrastructure for computation and research.
Data-Rich Economy: Human Capital
India’s digital scale already produces the inputs AI needs. Connectivity costs have fallen from roughly $4 per GB in 2014 to about $0.2 today, expanding usage across regions and income levels. The country’s linguistic and socio-economic diversity creates training environments that reflect real-world complexity rather than controlled test conditions. Initiatives such as Bhashini are converting this diversity into usable infrastructure by building open speech datasets, translation tools, and language models across Indian languages.
This data advantage is reinforced by workforce depth. India produces engineers at scale and represents a 16 per cent share of the global AI talent pool. The startup ecosystem shows how quickly innovation spreads once inputs become available. The number of startups has grown from roughly 1,000 in 2016 to nearly 200,000 today, producing over 125 unicorns. Domestic model initiatives such as Sarvam and Bhashini, which aim to train foundational systems on Indian linguistic and domain data, illustrate how talent and datasets can combine into local capability.
When dense data environments meet a broad technical base, experimentation spreads across institutions rather than concentrating in a few firms. That pattern has historically defined how general-purpose technologies diffuse.
Adoption depends not only on technical capability but on whether new tools can integrate into systems that already function at scale. India’s digital public infrastructure provides that interface. Identity, payments, authentication, and data-exchange rails support billions of interactions each month. AI systems, therefore, do not need to build user adoption from scratch. They can embed themselves within platforms people already rely on.
India’s governance framework strengthens this advantage. The Digital Personal Data Protection Act (DPDP Act) establishes a statutory basis for responsible data use, while the Data Empowerment and Protection Architecture (DEPA) enables consent-based data sharing in a controlled and auditable way. Together, they create a framework where innovation and safeguards can coexist. AI systems require data flows, but sustained adoption depends on trust. By combining legal protection with operational consent layers, India reduces friction in deployment while maintaining accountability.
Sectoral platforms show how this plays out in practice. Under the Digital Agriculture Mission, more than 7.6 crore farmer IDs and 23.5 crore crop plots have been digitised. An AI-based monsoon forecasting pilot reached 3.88 crore farmers across 13 states, with 31–52 per cent adjusting sowing decisions based on advisories. The Kisan e-Mitra chatbot has answered over 93 lakh farmer queries in 11 languages. In healthcare, the Ayushman Bharat Digital Mission has issued nearly 74 crore ABHA numbers and linked over 49 crore health records nationwide. Education platforms such as DIKSHA have recorded over 495 crore learning sessions and 15 crore enrolments. These are not isolated pilots. They represent incremental cost reductions across large public systems.
Shared Infrastructure:
Lowering AI Cost
Globally, the most immediate constraint on AI development is computers. Training models and running analytics require GPUs, specialised hardware, and sustained research funding. When these inputs remain expensive, innovation narrows and adoption slows.
India has begun to treat this layer as shared infrastructure. GPU availability has risen from roughly 10,000 to more than 38,000 units in a short period. The IndiaAI Mission aims to extend computer access to startups, universities, and research labs so experimentation can occur across the ecosystem. The extension of data-centre tax incentives until 2047 sends a similar signal, treating hyperscale infrastructure as long-term national capacity rather than short-cycle investment.
Hardware policy reinforces this trajectory. Semiconductor incentives and electronics manufacturing schemes aim to strengthen domestic capacity across assembly, packaging, and integration. Allocations of about Rs. 76,000 crore for semiconductor incentives, Rs. 20,000 crore for R&D, and Rs 8,885 crore for electronics manufacturing suggest that computer supply chains are being treated as strategic infrastructure. Without that depth, scaling AI capacity would remain fragile.
Research capital is deepening in parallel. Patent filings have more than doubled over the past decade, and India’s Global Innovation Index rank has improved. Public financing through the Fund of Funds for Startups has mobilised over Rs 22,000 crore, supported more than 1,190 startups, and catalysed a funding pool exceeding Rs 91,000 crore, with a further Rs 10,000 crore recently announced. General-purpose technologies spread fastest where experimentation is continuous, and these trends suggest India’s research base is becoming denser.
Diffusion will Determine AI Gains
These three strengths ultimately point in the same direction. AI’s long-term impact in India will depend less on frontier breakthroughs and more on how widely the technology becomes usable. When computers expand, datasets become institutional, and governance systems support responsible deployment; experimentation spreads across firms, universities, and public agencies.
That diffusion matters because AI’s largest gains will come from incremental improvements across many activities. Agricultural analytics built on AgriStack data can improve productivity. Health systems using ABDM layers can reduce reporting time and improve coordination. Education platforms using multilingual AI tools can expand access. Enterprise systems built on domestic language models can lower operating costs. None of these shifts alone transforms the economy, but together they reshape productivity.
General-purpose technologies do not transform economies because they exist. They transform economies when they become ordinary inputs into production and governance. The outcome of the AI Impact Summit should reinforce this logic that democratic access to AI resources along with the wide-scale diffusion of use cases, can contribute to the scalability, replicability, and adaptability of AI systems across sectors. India has the structural strengths to do exactly that.
