India might have a thriving startup ecosystem but it has no AI startups building foundational models, and one of the most prominent Indian voices in the AI space feels that this is a mistake.
Perplexity CEO Aravind Srinivas says that Nandan Nilekani is wrong when he says that Indian companies don’t need to build foundational AI models like ChatGPT or Gemini, which are large generalized models which can complete a wide variety of tasks. “Nandan Nilekhani is awesome, and he’s done far more for India than any of us can imagine through Infosys, UPI, etc. But he’s wrong on pushing Indians to ignore model training skills and just focus on building on top of existing models. Essential to do both,” Srinivas posted on X.
“I feel like India fell into the same trap I did while running Perplexity. Thinking models are going to cost a sh*t ton of money to train. But India must show the world that it’s capable of ISRO-like feet for AI. Elon Musk appreciated ISRO (not even Blue Origin) because he respects when people can get stuff done by not spending a lot. That’s how he operates. I think that’s possible for AI, given the recent achievements of DeepSeek. So, I hope India changes its stance from wanting to reuse models from open-source and instead trying to build muscle to train their models that are not just good for Indic languages but are globally competitive on all benchmarks. I’m not in a position to run a DeepSeek-like company for India, but I’m happy to help anyone obsessed enough to do it and open-source the models,” Srinivas added.
Srinivas was referring to comments that Nandan Nilekani has been making since last year, when he’s said that India doesn’t need to build foundational AI models. He says that these models will ultimately become commoditized, and it might not be worth spending billions of dollars to build them. Nilekani has said that it might be better to build AI applications on top of these models, because that’s where the real value might lie.
Aravind Srinivas seems to disagree. He runs Perplexity, which has been billing itself as an AI-based answering machine which is looking to take on Google. Perplexity is now worth $9 billion, and is perhaps the most prominent AI product made by an Indian-origin founder. Srinivas says that it might not necessarily take billions of dollars to build foundational models — he cites the example of Chinese company DeepSeek, which was founded just in 2023, and has managed to build a state-of-the-art model with its DeepSeek-R1. DeepSeek was founded by an AI-focused hedge fund called High-Flyer, which used AI for researching stocks, but created an AI-focused lab in 2023. DeepSeek had trained an early model, called the V3, for just $5.6 million, and had managed to produce state-of-the-art results in its category.
And it’s not as though India is short of money. Indian startups raised $14 billion in 2024, but most of it was allocated to incremental business models, like building the next quick commerce player or the next ed-tech platform. In November, Zepto alone had raised $350 million to send groceries to people in 10 minutes. AI, on the other hand, could not only upturn the white-collar job market, changing how coding, medicine, law and creative fields operate, but could also be a crucial component in defence and military systems which could in turn end up determining global power structures in the coming years. And with US and China both already taking big leads in building AI models, by choosing to stay away from the model race, India risks losing out having any domestic capabilities in a technology would will likely rearrange the world in the not too distant future.