Vivek Raghavan

Co-founder, Sarvam AI
by
Courtesy of Vivek Raghavan

Can a nation of nearly 1.5 billion people self-determine its AI future without foreign technology? That’s the question that Vivek Raghavan, co-founder of Bengaluru-based Sarvam AI, has spent his career answering. Raghavan cut his teeth on building digital public infrastructure: the technologist worked for over a decade to construct Aadhaar, which now ranks as the world’s largest biometric ID system. 

Raghavan has stressed the importance of AI sovereignty in the subcontinent, and of building a completely homegrown, Indian-language AI stack rather than relying on U.S. tech giants. India is a huge market for many U.S. tech companies, which some say puts the South Asian nation at a disadvantage when it comes to pioneering its own technology. “Many of these companies…provide these services for free. Now that’s a good thing for the people…But the question is, what is really free?” said Raghavan in an interview with Indian news outlet Bharatvaarta. As heavy users of U.S. AI chatbots, Raghavan said Indians are “exporting our data and importing intelligence,” which he compared to the country exporting cotton and importing cloth a century ago.

Alongside co-founder Pratyush Kumar, Raghavan worked on building open-source AI for Indian languages at AI4Bhārat, a research lab at IIT Madras. The duo launched Sarvam AI in August 2023 with the goal of creating multilingual Indian models that could take on global AI players in the West. In March Sarvam released Sarvam 30B and 105B, two open-source, open-weight models capable of advanced reasoning, coding, mathematics, and other high-level tasks, which could pave the way for Indian developers and companies to build their own applications. Sarvam Vision, its frontier vision language model introduced in February, can scan and extract text from documents in 22 Indian languages and English with a higher word accuracy rate than Gemini and ChatGPT, the company says.  In August, Nvidia became a strategic investor in Sarvam AI, joining a roughly $75 million Series B funding round that values the business at about $1.5 billion.

Advertisement