New Delhi — In a clear signal of its digital ambitions, India is charting a new course in artificial intelligence with a focus on what policymakers are calling “Sovereign AI” — homegrown, cost-effective models designed to meet local needs while keeping data within national borders.
At the recent India AI Impact Summit in New Delhi, government officials and technology leaders made it clear: India does not want the future of its data, languages, and public services to depend entirely on foreign AI platforms such as ChatGPT or Google Gemini, whose servers and core training data are largely based outside the country.
What Is Sovereign AI, and Why Does It Matter?
Sovereign AI refers to artificial intelligence systems that are developed, hosted, and governed within India. The goal is twofold: protect sensitive data and ensure AI systems truly understand India’s linguistic and cultural diversity. With hundreds of millions of users speaking languages like Hindi, Tamil, Bengali, and Marathi, India argues that global models trained primarily on Western data often fall short.
Sam Altman: India Is Becoming a Builder
Sam Altman, CEO of OpenAI, acknowledged this shift, noting that India is no longer just a massive consumer market but is increasingly becoming a builder of AI technology. At the same time, Altman cautioned that developing the world’s most powerful AI systems requires billions of dollars and enormous energy resources — a cost few countries can afford.
Why India Is Betting on ‘Middleweight’ Models
Instead of racing to build the largest AI models in the world, India is focusing on so-called “middleweight” models — typically ranging from 30 to 105 billion parameters. These systems are smaller and cheaper than frontier models but powerful enough to handle most real-world tasks.
Indian startup Sarvam AI has already unveiled models in this range, optimized for Indian languages and use cases. Another domestic player, Krutrim, is pursuing a similar approach, aiming to embed cultural context directly into its AI systems.
Three Pillars of India’s AI Strategy
India’s approach rests on three key pillars:
- Domestic infrastructure: The government is investing in GPUs and making them available to startups at subsidized rates.
- Indian-built models: Startups are developing AI systems trained on local languages, laws, and social realities.
- Global partnerships: India is working with companies like NVIDIA and Microsoft to expand data center capacity within the country.
What’s in It for Ordinary Citizens?
Officials say sovereign AI could transform everyday life. Farmers could receive crop advice in their native languages, citizens could access government services more easily, and healthcare providers could use AI tools to deliver faster, cheaper diagnostics — even in rural areas.
The Challenges Ahead
Significant hurdles remain. AI data centers consume vast amounts of electricity and water for cooling, and the most advanced chips are still imported. Securing these resources at scale will be one of India’s toughest tests.
Still, policymakers remain confident. India’s bet is that it doesn’t need the biggest AI in the world — just the most practical and affordable one. The message is clear: the technology should be Indian, the data should stay Indian, and the benefits should reach Indian citizens first.