Abhishek Singh, CEO of the IndiaAI Mission, has outlined a focused strategy for India's AI development, emphasizing practical, population-scale models over the global race for artificial general intelligence. In an interview, he highlighted India's potential as the world's inference capital and preparations for the upcoming AI Impact Summit in New Delhi. The approach prioritizes sovereign AI solutions tailored to Indian challenges in sectors like healthcare and agriculture.
As global powers race to develop ever-larger AI models in pursuit of artificial general intelligence (AGI), India is charting a more pragmatic course. Abhishek Singh, chief executive officer of the IndiaAI Mission, stated in an interview with Hindustan Times that the country is not competing in the AGI race. Instead, India aims to build AI systems that operate effectively at a population scale, focusing on smaller, task-specific models suitable for limited hardware and decentralized networks.
This aligns with recommendations in the recent Economic Survey, which advocates for efficient models addressing real-world problems in languages, healthcare, education, agriculture, and manufacturing. Singh emphasized that model size, measured in parameters, is less important than their ability to solve specific issues. The first phase of the IndiaAI Mission, backed by ₹10,372 crore, provides subsidized computing resources to 12 startups developing eight large language models and four smaller ones. These startups will showcase their work at the AI Impact Summit, scheduled from February 16 to 20 at Bharat Mandapam in New Delhi.
The summit's objective is to foster sovereign AI—built, hosted, and designed in India for local challenges—with potential for global scaling. Singh noted India's ambition to become the 'inferencing capital of the world,' where AI models run continuously for millions of users. Currently, the mission offers access to over 38,000 GPUs at subsidized rates, but Singh indicated this is insufficient; India may need 100,000 to a million GPUs for widespread use.
Electronics and IT Minister Ashwini Vaishnaw announced that the second phase of the mission will launch in about five months, including additional GPUs revealed at the summit. Recent policies, such as a tax holiday until 2047 for foreign companies routing global cloud services through Indian data centers, are expected to attract investments. Vaishnaw reported $90 billion already secured in AI and data center commitments, with $70 billion underway and projections up to $200 billion.
The summit anticipates over 100,000 attendees, including 20 heads of state, 100 government representatives, and 100 global AI leaders. India seeks more signatories for its joint declaration than previous events, such as the French summit with 58 signatories, despite potential absences from the US and UK. Singh remains optimistic, focusing on collaborative outcomes from seven international working groups. He acknowledged India trails the US and China in the global AI race but believes increased R&D investments, GPU access, and support for foundation models will enable catch-up and leapfrogging.
