MUMBAI, June 26 — India must build its own foundational artificial intelligence models or risk being relegated to a mere consumer of the technology, according to BharatGen, the country’s only not-for-profit AI entity and the largest single beneficiary of the government’s $1.2-billion AI mission. BharatGen is a consortium of AI labs spread across nine Indian institutes. It was established by the Indian Institute of Technology (IIT) Bombay in December 2024 as a non-profit.
The entity received a government grant of ₹ 1,058.82 crore under the Centre’s AI mission on 18 September 2025. That funding makes it a core pillar of the national push to build sovereign AI capabilities.
The consortium is now developing one of India’s first sovereign large language models — the foundational AI systems that mimic human writing and speech by analysing vast amounts of data. The project is building its first model from scratch, not fine-tuning an existing foreign one. Ganesh Ramakrishnan, chair professor for computer science engineering at IIT-Bombay and founding director of BharatGen, laid out the strategic logic.
“We can only truly understand AI models when we build them from scratch,” he said. He drew an analogy to cooking: knowing the recipe allows you to tailor a dish to your health needs.
Similarly, knowing exactly what goes into an AI model — especially its data — is essential for building effective tools around it. “For AI adoption to scale, users have to move from the last mile to the first mile,” Ramakrishnan added. “An ecosystem built solely around using AI tools will remain at the consumption layer.
We need people who understand, build and improve.” That warning carries weight in a market where most Indian enterprises and startups currently rely on foreign foundational models. The risk, as BharatGen frames it, is that India becomes a downstream user of AI rather than a creator of the technology that will shape the next decade of productivity and economic growth.
Talent deficit looms
However, experts warn that a severe deep tech talent deficit could stall broader industry progress. Building foundational models from scratch requires a rare combination of skills — machine learning research, data engineering, distributed computing and domain-specific knowledge. India’s pipeline of such talent remains thin compared to the scale of the ambition.
The consortium’s structure is designed to address this. By linking labs across nine institutes, BharatGen aims to create a distributed research network that can train PhDs, postdocs and engineers in the craft of model-building.
The non-profit model also means the output is intended to be shared, not locked behind corporate walls. The government has placed a significant bet on this approach. The ₹ 1,058.82 crore grant is one of the largest public investments in foundational AI research anywhere in Asia outside China.
The question is whether the talent base can grow fast enough to match the funding. For now, BharatGen is focused on delivering its first sovereign model.
The entity is building from scratch, which means every layer — from data curation to training architecture to safety alignment — is being designed domestically. That approach is slower than adapting an existing open-source model, but Ramakrishnan argues it is the only path to genuine ownership. The coming months will show whether India can turn its deep tech talent deficit into a surplus through this kind of institutional investment.
If BharatGen succeeds, it could shift the country from a consumer of AI to a builder of it. If the talent gap proves too wide, the government’s billion-dollar bet may yield a model, but not the ecosystem it was meant to create.
