Executing AI at population scale: Abhishek Singh, Additional Secretary, MeitY on India’s path to global leadership
Speaking at the Regional AI Impact Summit in Bhopal, Abhishek Singh, Additional Secretary, Ministry of Electronics and IT (MeitY) and Director General, NIC made one message clear: India’s AI journey is no longer about pilots or promises. It is about execution—at scale, across states, and for real-world impact.
Why India Is Central to the Global AI Future
Globally, AI development today is concentrated among a few companies and a few countries. The United States and China remain far ahead in terms of sheer model scale and compute dominance. India currently ranks third—but rankings alone do not tell the full story.
The country’s strength lies in its ability to to deploy technology at population scale. Through Aadhaar, UPI, DigiLocker and the broader India Stack, the country has already demonstrated how digital public infrastructure can transform governance, service delivery, and inclusion. The world is now watching to see whether India can replicate that success with artificial intelligence. According to Singh, this expectation places a responsibility on India—not just to adopt AI, but to shape how it is used.
AI as National Infrastructure, Not Elite Capability
At the core of this approach is the India AI Mission, which Singh described as a deliberate shift away from AI being accessible only to a few large corporations.
The mission is built on seven pillars, but its central idea is simple: the Government of India will provide the common AI infrastructure—compute, datasets, foundational models, skilling platforms, and safety frameworks—so that startups, researchers, states, and institutions can focus on solving real problems.
Access to compute remains one of the biggest barriers globally. More than 90% of advanced AI compute today is controlled by a single vendor, making it expensive and inaccessible. Through an innovative tendering process, the India AI Mission has already provisioned close to 38,000 GPUs, that have been made available to the national AI ecosystem at significantly lower-than-global prices. These resources are open to startups, researchers and innovators across the country.
Creating Shared Data Foundations
Compute alone is not enough. AI needs high-quality, diverse datasets. To address this, India has created a common AI data platform with over 6,000 datasets sourced from government, public sector, and private entities. A key component is Bhashini, which provides language datasets enabling seamless translation and multilingual service delivery across Indian languages.
This focus on language, Singh noted, is essential for a country as diverse as India, where inclusion depends on AI systems understanding local context.
Building AI Models That Understand India
Most widely used large language models today are trained primarily on Western datasets. As a result, they often struggle with Indian languages, governance structures, and cultural nuances.
India is responding by actively supporting the development of indigenous foundational models. Nearly a dozen such projects are currently underway under the India AI Mission, with several expected to be launched ahead of the Global AI Impact Summit.
The objective is not to compete headline-to-headline with global giants, but to build models that are context-aware, culturally aligned, and suitable for Indian use cases across governance, agriculture, healthcare, and education.
From Strategy to Deployed Use Cases
A defining feature of India’s AI strategy, Singh emphasized, is its focus on applications rather than abstractions. Through a structured application development framework, real problem statements are sourced from central ministries and state governments. These span sectors such as agriculture, health and education, urban management, and governance. Startups and research institutions are invited to build solutions, which are then procured and deployed at scale.
AI-based tuberculosis diagnosis systems, agricultural advisory platforms, and GIS-driven planning tools are already being implemented in multiple states. This steady translation of ideas into deployments is what Singh described as India emerging as a global “use-case capital” for AI.
Skilling Beyond Engineers
AI adoption ultimately depends on people. While India has a strong engineering talent base, Singh highlighted the need to broaden particiption. The AI Impact Fellowship, initially designed for engineering students, has now now been expanded to include students from medicine, law, commerce, and humanities. AI, he noted, is no longer a purely technical discipline—it intersects with policy, ethics, law, healthcare, and social sciences.
In parallel, data labs are being established in Tier-2 cities and smaller towns. In Madhya Pradesh alone, 30 data labs have been sanctioned across ITIs and polytechnics. These labs aim to train large numbers of students as data analysts, annotators, and data scientists, ensuring that AI capability is not limited to metropolitan centres.
Strengthening the Startup Backbone
India’s AI ecosystem is being built bottom-up, with startups playing a central role. Centres of Excellence are being set up to incubate startups, provide seed funding, and offer grants. Two such centres have already been approved in Madhya Pradesh. The objective is to enable startups to move move beyond pilots, and ensure creation of production-grade solutions that can scale nationally.
Responsible AI
As AI systems become more pervasive, risks around bias, hallucinations, deepfakes, and privacy grow sharper. Singh outlined India’s balanced approach—one that prioritises innovation while putting guardrails in place.
AI governance guidelines have been issued, and the India AI Safety Institute has been established to focus on bias reduction, machine unlearning, privacy preservation, anonymisation, deepfake detection, and labeling of AI-generated content.
The aim is not over-regulation, but trust—especially when AI is deployed in sensitive public-sector and citizen-facing systems.
A Decentralised National AI Movement
Perhaps the most distinctive aspect of India’s AI journey is how its agenda is being shaped. More than 350 pre-summit events have already been conducted across India, along with engagements in over 30 countries. Inputs from industry, policymakers, startups, academia, and researchers are shaping the summit’s agenda.
The three pillars, People, Planet, and Progress, reflect this broad-based thinking, balancing jobs and skills, sustainability, and inclusive economic growth.
The Road Ahead
India’s AI ambition, Singh concluded, is not about dominance. It is about demonstration. Demonstrating that AI can work at population scale. Demonstrating that innovation and responsibility can move together. And demonstrating that leadership in AI can emerge from real-world deployment, not just research labs.