India’s pivot to the global race for AI diffusion could grant the country leadership of an important effort

China, however, is shifting its strategy. Its newly approved 15th Five-Year Plan (for 2026–2030) introduces a sweeping ‘AI+ action plan’ to embed artificial intelligence across manufacturing, supply chains, public systems and the entire economy.

The 2026 Stanford HAI AI Index confirms this shift: AI leadership is no longer decided only by who builds the best models. The new battleground is infrastructure plus diffusion—i.e., turning AI into an economy-wide driver of productivity at population scale.

India enters this race with strong structural advantages. We have already built critical digital rails. UPI has scaled from pilot-project stage to over 20 billion monthly transactions.

We possess massive demand: nearly 80 million micro, small and medium enterprises (MSMEs), hundreds of millions of informal workers and stretched public systems where even modest productivity gains create an outsized impact.

Our diversity demands inclusive design—vernacular, easy-to-use solutions that require minimal connectivity and can work anywhere. We have also begun building sovereign capacity through the IndiaAI Mission.

Yet, the risks are significant. The most critical is institutional. If the underlying architecture remains fragmented, AI will amplify fragmentation. If data stays siloed, AI will stay constrained. If workflows are broken, AI cannot deliver meaningful impact.

India must therefore move beyond deploying isolated AI solutions. We must re-engineer the entire system architecture to support broad-based diffusion across our diverse economy.

Coordinating action at this scale, within a limited window of opportunity, demands a deliberate, whole-of-government AI Diffusion Framework built on open architecture, shared intelligence layers, consent-based data exchange and vernacular AI agents embedded into real workflows. Four key pivots are essential.

First, re-engineer institutional architecture for execution at scale: India needs an empowered Council to develop and drive a single national AI Diffusion Roadmap. It must resolve cross-ministerial bottlenecks quickly, conduct regular outcome reviews and dynamically reallocate resources.

Fragmented mandates across ministries have repeatedly undermined India’s technology programmes. The Council must serve as the central integrating authority. It should focus on three priorities.

One, it should establish a National AI Diffusion Implementation Unit, modelled on National Payments Corporation of India’s successful orchestration model. Reporting directly to the Council, this unit would handle programme design, public-private partnerships, real-time monitoring and last-mile coordination.

Two, create a Unified Data and Standards Office (UDSO) to harmonize consent-based data exchange, vernacular standards and interoperability across digital public infrastructure (DPI) layers.

Three, set up Sectoral AI Transformation Councils in priority areas such as agriculture, healthcare, MSMEs, education and urban governance. These councils would clear roadblocks and accelerate diffusion with measurable outcomes. The government has already announced the AI Economic and Governance Council.

This is a welcome recognition that AI diffusion requires high-level coordination. With strong institutional backing and execution discipline, it can dramatically improve our diffusion trajectory.

Second, reform procurement for speed, scale and outcomes: Current procurement is input-based and pilot-centric. India must shift to outcome-linked contracts that reward genuine adoption and results.

For government-funded AI projects, a significant portion of contract value should be tied to verifiable metrics: workflows that go live, active user engagement and tangible efficiency gains such as reduced processing times.

Drawing from models like the UK’s AI Procurement in a Box, India should develop clear standards. Every project must submit a credible diffusion and sustainability plan at the approval stage. Future funds should be released only based on demonstrated outcomes, not on pilots or demonstrations.

Third, strengthen last-mile capacity for deep diffusion: AI must reach panchayats, anganwadis and MSME clusters. This requires the following.

One, we must launch a national cadre of AI diffusion fellows, funded jointly by Skill India, the IndiaAI Mission and states. Their role would be on-ground training, handholding and troubleshooting for gram panchayat officials, anganwadi workers and MSME owners.

Two, we should deploy pre-built vernacular AI agents directly into government and MSME workflows through DPI integration. These agents must support on-device and offline functionality for low-connectivity areas. As the Stanford HAI Index notes, AI systems remain ‘jagged’—powerful in some contexts, brittle in others. Vernacular, context-specific design is thus essential for reliable last-mile deployment.

Three, India must roll out an ‘AI for Every Citizen’ micro-credential programme to foster vernacular AI literacy among millions of frontline users, leveraging grassroots infrastructure like Common Service Centres.

Fourth, redefine success metrics: Success must be measured not by the number of solutions launched or startups created, but by ground-level impact. India should launch a single, real-time Public National AI Diffusion Dashboard tracking three core outcomes across priority sectors and states: adoption rate of AI-enabled workflows, productivity and service delivery impact (such as faster processing and greater coverage) and citizen reach.

Publishing league tables for ministries and states would encourage competitive federalism. A portion of future state AI funding should be tied to dashboard performance.

The 2026 Stanford AI Index reveals that people increasingly believe AI will matter in their lives, yet many remain uncertain about whether institutions can manage it responsibly. India recorded the largest increase in AI nervousness among surveyed countries. A transparent public dashboard can demonstrate to citizens and the world that AI diffusion is being governed with accountability.

If India executes this framework with the same urgency that built UPI, it could become the world’s first large-scale demonstration of democratized intelligence—real-time, relevant AI embedded across every layer of the economy for the benefit of 1.4 billion-plus citizens. Productivity will rise, service delivery will improve and sovereign AI capacity will be secured.

The cost of inaction is strategic dependence: intelligence designed, controlled and priced elsewhere, leading to value leakage and compromised autonomy in governance, security and growth.

The authors are, respectively, chief economic advisor, Government of India, and distinguished fellow, Niti Aayog.