IndiaAI Mission: AI Kosha & Common Compute Drive India's AI Revolution

Table of Contents☰

What’s in Today’s Article?

Recently, the IndiaAI Mission launched AI Kosha, a national dataset platform for collecting non-personal data to develop AI models.
It also introduced the Common Compute portal, providing startups and academia access to shared GPUs for AI development.

The IndiaAI Mission was initially launched as a joint initiative between the Ministry of Electronics and Information Technology and Nasscom in 2023.
It has since evolved into a government-led initiative with an outlay of ₹10,371.92 crore.
The mission’s key goals are:
- “Making AI in India” – Encouraging domestic AI development.
- “Making AI Work for India” – Ensuring AI benefits various sectors in the country.
This initiative aims to bolster India’s AI ecosystem by making high-end computing resources accessible to startups and researchers.

A major initiative under the mission is the Common Compute facility, which provides startups and researchers access to Graphics Processing Units (GPUs)—essential for training and running AI models.
Since GPUs are expensive and in high demand globally, IndiaAI is empaneling data centers with GPU resources.

14,000 GPUs have already been acquired and commissioned.
More GPUs will be added quarterly to expand computing resources.
The mission will allocate GPU access to selected entities based on need and availability.

The IndiaAI Mission comprises seven pillars, with a focus on building India’s AI capabilities across various domains.

Develops India-specific datasets to improve AI models.
Reduces bias from Western-trained AI models by incorporating local language and cultural data.

The IndiaAI Mission has invited proposals to develop an indigenous AI foundation model.
67 applications have been received, and the government aims to fast-track the process within months.

AI Kosha is part of the IndiaAI Datasets Platform, aimed at addressing a key challenge in AI development—the dominance of Western-trained models.
Many AI models rely heavily on English-language data from developed countries, making them less effective for India’s diverse linguistic and cultural landscape.

AI Kosha seeks to make India-centric data more accessible for researchers and AI developers.
It includes datasets such as translation models for Indian languages, helping reduce bias in AI training.

Open-source AI models provide a foundation, but local datasets are essential to ensure AI tools are relevant and effective for Indian users.
AI Kosha aims to empower Indian firms to build indigenous generative AI models, reducing reliance on foreign AI tools like OpenAI and Google’s models.