It’s been two years since ChatGPT blew everyone’s mind. Even China’s DeepSeek and Manus are turning heads by showing it’s possible to build powerful AI on a budget.
Meanwhile, India — with all its tech talent and startup energy — is still playing catch-up in the race to create its own foundational AI model.
Let’s be real: India isn’t short on brainpower. Around 15% of the world’s AI workforce is from here. But the issue is, most of that talent is leaving. Stanford’s AI talent migration report shows a steady brain drain, with researchers heading abroad where deep-tech R&D environments actually exist. Back home, very few breakthroughs are coming out of Indian universities or corporate labs.
Sure, the government has finally woken up. It says a made-in-India foundational AI model could be ready in under 10 months. It’s also giving startups and researchers access to high-end chips. But experts aren’t convinced.
Building something like DeepSeek isn’t just about throwing chips and cash at the problem — it needs long-term investment, a rich academic ecosystem, and serious policy support. And India’s AI mission? It’s worth $1 billion.
Sounds like a lot until you compare it to the US’s $500 billion Stargate project or China’s $137 billion AI masterplan.
India does have over 200 generative AI startups, and global players like Microsoft and Nvidia are betting big on the country. Even OpenAI’s Sam Altman, who once doubted India’s role in foundational AI, has now changed his tune. India is actually OpenAI’s second-biggest user market.
But there’s a big gap between building apps on top of existing platforms and creating your own base-level model like ChatGPT or DeepSeek. Most Indian startups are tweaking open-source models — which is smart and necessary — but it’s not the same as leading the charge.
And let’s not forget the data issue. Training a multilingual AI in a country like India is tough without good datasets in Hindi, Tamil, Marathi, and dozens of other regional languages.
Experts say India needs to learn from the UPI playbook — where government, industry, and academia worked together to revolutionise payments. Something similar needs to happen for AI.
The country also needs to boost its chip-making game if it wants to run powerful models in-house. Otherwise, we’ll be stuck depending on imports — and that’s a risky game geopolitically.
Bottom line? India has the people, the startups, and the hunger. But unless the big players step up — both in policy and investment — it might be years before we see an Indian-made DeepSeek.
