Sarvam AI Likely to Receive INR 220 Cr From Govt to Build Foundational Models

Sarvam AI may become the first startup to benefit from the Government of India’s IndiaAI Mission, aimed at fostering the development of indigenous large language models (LLMs), The Times of India reported.

Sarvam is expected to receive support equivalent to ₹220 crore, primarily in the form of free access to around 4,000 Nvidia H100 chips for six months—essential hardware for training advanced AI models.

The IndiaAI Mission, backed by an allocation of ₹10,300 crore, aims to support key initiatives including the IndiaAI Compute Capacity, Innovation Centre, and Datasets Platform. One major component involves setting up a shared computing facility with 18,693 GPUs.

Last year, Sarvam AI released India’s first open-source foundational model, built completely from scratch. The model, called Sarvam 2B, is trained on 4 trillion tokens of data.

It can take instructions in 10 Indic languages, including Hindi, Tamil, Telugu, Malayalam, Punjabi, Odia, Gujarati, Marathi, Kannada, and Bengali.

“This is the first open-source foundational model trained on an internal dataset of 4 trillion tokens by an Indian company, with compute in India, with efficient representation for 10 Indian languages,” Sarvam AI co-founder Vivek Raghavan said.

To date, the IndiaAI Mission has received over 190 proposals for both large and small language models.

Following an overwhelming response from innovators and researchers, the government of India has extended the submission deadline for foundation model proposals to April 30, 2025.

The move aims to accommodate the surge in interest and ensure broader participation in shaping India’s AI capabilities through cutting-edge foundational research.

"Dates of Submission of Proposals post 30th April 2025 will be announced subsequently. We thank all applicants for their enthusiastic participation and continued support in shaping India’s AI ecosystem," IndiaAI said in a LinkedIn post.