Copyright infringement not intended

Picture Courtesy:   MONEY CONTROL

Context:

Sarvam AI has been chosen to develop India’s first homegrown AI foundational model under the ECMS scheme.

News in Detail

The Union government selected Sarvam AI, a Bengaluru-based startup, to create India’s first homegrown AI foundational model. This announcement comes from the Union Minister for Electronics and Information Technology during an event to launch guidelines for the Electronics Component Manufacturing Scheme (ECMS).

This move is part of the IndiaAI Mission, a ₹10,000-crore government program to boost India’s AI capabilities. The goal is to make India self-reliant in AI, reduce dependence on foreign technologies like ChatGPT, and create AI that understands diverse languages and cultures.

What is an AI Foundational Model?

It is a large language model (LLM) that processes vast amounts of data to understand and generate human-like text, answer questions, or even hold conversations.

Sarvam AI plans to build a model with 70 billion parameters—numbers that represent the model’s ability to learn and perform complex tasks. This model will compete with global giants like ChatGPT or China’s DeepSeek.

Unlike foreign models, Sarvam’s AI will focus on Indian needs. It will:

• Speak and understand multiple Indian languages.
• Work well with voice commands.
• Perform advanced reasoning, like solving problems or making decisions.
• Operate securely on a large scale, serving millions of users.

They plan to build three types of models:

1. Sarvam-Large: For complex tasks like reasoning or generating detailed content.
2. Sarvam-Small: For fast, real-time applications like chatbots.
3. Sarvam-Edge: For small devices, like smartphones, to run AI without internet access.

How Will Sarvam AI work?

The government supports Sarvam AI with powerful resources. They provide access to 4,096 high-end Nvidia H100 GPUs (graphics processing units) for six months. GPUs are like supercomputers that help train AI models by processing massive data quickly. These GPUs come from Indian companies like Jio Platforms, Yotta, and Tata Communications, showing India’s growing tech infrastructure.

Sarvam also receives GPU compute credits worth ₹200 crore, instead of direct cash, to ensure the funds are used for building the model. They collaborate with AI4Bharat at IIT Madras, a leader in Indian language AI research, to make their model robust.

Sarvam commits to delivering the model in just six months—a challenging but exciting timeline. Their work has already begun, focusing on a multimodal model that handles text, voice, and possibly images, all customized for Indian users.

Strategic Importance for India

Electronics Component Manufacturing Scheme (ECMS)

It is a ₹22,919-crore government plan launched in April 2025. ECMS aims to make India a hub for electronics manufacturing by supporting companies that produce components like circuit boards.

The scheme will:

• Attract ₹59,350 crore in investments
• Create 91,600 direct jobs
• Boost electronics production to ₹4,56,500 crore

While ECMS focuses on hardware, it connects to the AI initiative because strong electronics manufacturing supports the infrastructure (like GPUs) needed for AI development. 

Must Read Articles: 

BASICS OF ARTIFICIAL INTELLIGENCE

ROLE OF AI IN STRATEGIC AFFAIRS

Source: 

MONEY CONTROL