Sarvam AI to Roll Out India’s First Large Language Model Tailored for Indian Languages and Digital Transformation

India is progressing rapidly in the race to construct its own Artificial Intelligence (AI) systems, and a native startup called Sarvam AI is at the center of this immense leap. Sarvam is expected to introduce the country's first "large language model" (LLM) trained on Indian languages and data by the beginning of next year.This is part of the government's big IndiaAI Mission, which is giving developers, students, and startups more computing power.

Union IT Minister Ashwini Vaishnaw noted that the government has increased its resources to 40,000 GPUs, which is four times more than the original goal of 10,000. He stated, "Most likely, Sarvam will be the first to release a well-trained Indian language model that doesn't have the biases that many global models have."

Sarvam has already gotten the most out of the IndiaAI Mission, which has a ₹10,000 crore fund behind it.The Bengaluru-based startup received a record 4,096 NVIDIA H100 GPUs through Yotta Data Services, supported by nearly ₹99 crore in subsidies.

Vivek Raghavan and Pratyush Kumar started Sarvam in July 2023. Before that, they worked on the AI research initiative AI4Bharat. Raghavan is known for contributing to build public digital platforms like Aadhaar. He claimed that the company's main goal is to collaborate with Indian businesses to create AI models that can be used in the real world. Their goal is to have a big effect on a lot of people, especially in sectors like public services where AI can be added to India's digital infrastructure.

Sarvam also raised $41 million in Series A funding in December 2023. Lightspeed Ventures, Peak XV Partners, and Khosla Ventures were the main investors.This money has helped it improve its research and development so that it can make cutting-edge AI solutions.

The government is not just encouraging new ideas, but it is also keeping an eye on AI safety. India has started a "AI Safety Institute" that is collaborating with institutions like IIT Jodhpur to make tools that find errors and biases in AI systems. Vaishnaw claimed that India is taking a "techno-legal" strategy, which means that it is using both technology and the law to make sure AI is safe and reliable.

With Sarvam leading the way and strong government support, India is setting the stage to become a global player in AI innovation. The challenge ahead lies in execution and active participation from the private sector, but the foundation for India’s AI journey is now stronger than ever.

This article is based on information from Money Control