Bhavish Aggarwal’s $230 Million Bet on Krutrim: Pioneering AI for Indian Languages
Ola’s founder, Bhavish Aggarwal, is making waves in the tech sector by investing a whopping $230 million into his own AI startup, Krutrim. As India aims to carve out its niche in the bustling field of artificial intelligence, traditionally dominated by US and Chinese giants, this investment could be a game-changer.
The Ambitious Vision Behind Krutrim
Aggarwal is channeling this massive investment primarily through his family office, as reported by sources familiar with the situation. He announced on X that Krutrim aims to attract a substantial total of $1.15 billion in funding by next year, with plans to source the remaining capital from external investors.
This funding coincides perfectly with Krutrim’s strategic moves, such as releasing its AI models as open source and embarking on an ambitious partnership with Nvidia to develop what they claim will be India’s largest supercomputer.
Krutrim-2: Leading the Charge
Krutrim recently unveiled Krutrim-2, its new 12-billion parameter language model tailor-made for Indian languages. Its performance metrics are impressive, showing a staggering score of 0.95 in sentiment analysis—far superior to its competitors, which are hovering around 0.70. Not only that, but it also boasts an 80% success rate in code generation tasks.
The lab isn’t stopping there; several specialized models aimed at processing images, translating speech, and enhancing text search, all focused on Indian languages, have also been made open source. Aggarwal, highlighting the startup’s commitment to the Indian AI community, remarked, “By open sourcing our models, we hope the entire Indian AI community collaborates to create a world-class Indian AI ecosystem.”
India’s Growing AI Landscape
These advancements come at a crucial time. India is eagerly striving to position itself as a formidable player in the global AI landscape. In fact, just last week, the government praised the strides made by DeepSeek, a Chinese AI lab that has recently released its R1 “reasoning” model. Excitingly, Krutrim’s cloud service began hosting DeepSeek’s models on Indian servers last week.
Furthermore, to bridge the gap in language proficiency assessments, Krutrim developed BharatBench, a unique framework to evaluate AI models’ capabilities specifically in Indian languages. This step addresses the significant oversight found in current benchmarks, which primarily target English and Chinese.
Technical Innovations at Play
Krutrim’s approach sets itself apart with its advanced technical features, including a 128,000-token context window. This innovation allows the AI models to handle more extended texts and intricate conversations, ensuring better user experiences. The results? Krutrim-2 has achieved an astonishing score of 0.98 in grammar correction and 0.91 in multi-turn conversations.
This investment follows the launch of Krutrim-1—a 7-billion parameter model that made headlines as India’s first large language model. Plans for the supercomputer collaboration with Nvidia are set to go live in March, with further expansions anticipated throughout the year.
Conclusion
Aggarwal’s bold investment in Krutrim signals a robust shift towards harnessing AI potential in India, particularly for its rich tapestry of languages and cultures. The country is on the cusp of an AI revolution, fostering homegrown solutions that resonate with local contexts.
The AI Buzz Hub team is excited to see where these breakthroughs take us. Want to stay in the loop on all things AI? Subscribe to our newsletter or share this article with your fellow enthusiasts.
