Bhavish Aggarwal to invest Rs 2,000 crore in AI startup Krutrim, unveils open-source models

Ola founder Bhavish Aggarwal on Tuesday announced a Rs 2,000 crore investment in his artificial intelligence venture Krutrim with a commitment to step it up to Rs 10,000 crore by next year, and said the startup is releasing its AI models to the open-source community.

Aggarwal’s announcements on microblogging platform X comes as the race to develop more efficient and cost-effective AI models heats up, especially after the stir caused by Chinese upstart DeepSeek.

Aggarwal did not provide details on the investment. A person familiar with the matter told ET that the funding would be a mix of equity and debt from Aggarwal.


Aggarwal also unveiled Krutrim AI Lab, with a commitment to “driving cutting-edge AI innovation for India while contributing to the global AI landscape”.

“We’re nowhere close to global benchmarks yet, but have made good progress in one year. And by open sourcing our models, we hope the entire Indian AI community collaborates to create a world class Indian AI ecosystem,” he wrote on X, adding: “We’re still learning to walk before we can run, hopefully within this year!”


The startup is open sourcing Krutrim 2, its new large language model (LLM); Chitrarth 1, a vision-language model built on top of Krutrim 1; Dhwani 1, a speech-language model capable of tasks like speech translation; Vyakhyarth 1, an Indic embedding model for applications such as search and retrieval-augmented generation; and Krutrim Translate 1, a text-to-text translation model, among others.

Aggarwal also announced the deployment of GB200 superchips in partnership with American chip major Nvidia. The system is expected to go live by March, and “we will make it the largest supercomputer in India by end of year”, he wrote on X.

The announcement comes just days after Aggarwal said Krutrim has deployed advanced AI models from DeepSeek on its domestic servers, a move aimed at improving data privacy and reducing AI model training costs.

On January 31, union information technology minister Ashwini Vaishnaw said that DeepSeek models would soon be hosted on Indian servers.

Vaishnaw said also that India would offer the world’s most affordable compute power for high-end chips powering generative AI, at less than $1 per hour, as part of the government’s Rs 10,000 crore IndiaAI Mission.