Bengaluru Startup Sarvam AI Leads India’s Drive for Home-Grown LLM
Sarvam AI, selected from over 400 applicants, will develop India’s first home-grown large language model under the INDIAai Mission.
image for illustrative purpose
India’s government has tapped Bengaluru-based Sarvam AI to develop a home-grown foundational large language model under its INDIAai Mission, selecting the startup from more than 400 contenders.
The Union IT ministry will assign Sarvam AI 4,096 Nvidia H100 graphics processing units for a six-month period, sourcing the hardware through empanelled providers including Jio, CtrlS, Yotta and Tata Communications.
Launched in March 2024 with a budget of Rs 10,372 crore (about $1.25 billion), the INDIAai Mission aims to democratize access to high-performance computing, improve data quality and position India as a global AI leader. To date, the government has invested in a shared facility housing 18,693 GPUs.
According to UNESCO’s 2022 State of the Education Report, India leads the world in relative AI skill penetration. The 2024 Stanford AI Index Report ranks India among the top nations advancing AI research and applications.
Sarvam AI plans to build a 70 billion-parameter model designed for advanced reasoning and voice-first interactions. The startup says the system will support 22 Indian languages alongside English and will be production-ready within six months.
Experts caution that assembling comprehensive datasets covering India’s linguistic diversity—dialects, complex grammars and local idioms—poses a major hurdle. Removing biases related to gender, religion and caste will add further complexity.
The scarcity of specialists in natural language processing, machine learning and multilingual AI threatens to slow progress. India contributes less than 1.5% to global AI research, and its high-performance computing infrastructure and cloud services are still maturing.
To bridge data gaps, the government recently launched AIKosh, a central repository for public-sector datasets. Officials say partnerships with private firms holding vast multimodal data—telecom operators, e-commerce platforms and financial services—and bilateral research exchanges will be essential to accelerate India’s sovereign AI capabilities.
