During the NDTV World Summit 2026 in Mumbai, investor Nikhil Kamath argued that India’s AI prospects lie not in competing for megamodels but in building inference layers that run on open‑source engines.
He pointed out that the United States and China pour billions into AI research, giving their firms a sandbox to experiment, whereas Indian startups face a capital gap that makes high‑risk model development untenable.
Kamath stressed that electricity and data‑centre costs in India are 30‑40% lower than in the West, a margin that could make Indian data‑centres the cheapest place to host inference workloads.
The investor also compared China’s municipal competition with Indian states, suggesting that 28 chief ministers could adopt a ‘state‑as‑CEO’ model to lure tech firms.
He urged the Indian government to draft a fiscal package by June 30 that would subsidise power and cooling for AI data‑centres, a move that could see the first state‑sponsored AI hub roll out early next year.