Mon May 05 13:58:23 UTC 2025: ## India’s Ambitious AI Push: Can it Build a Homegrown ChatGPT?

**Bengaluru, India – May 5, 2025** – The Indian government and several startups are racing to develop an indigenous large language model (LLM), similar to ChatGPT or Google’s Gemini. While ambitious, the undertaking faces significant hurdles, primarily high costs and a lack of readily available data.

The success of Chinese firm DeepSeek, which trained a sophisticated reasoning model at a fraction of the cost of Western counterparts, has fueled optimism. However, even DeepSeek’s endeavor reportedly cost over $1.3 billion. This pales in comparison to the billions of dollars being invested annually by major tech companies. The Indian government’s IndiaAI Mission, with a ₹10,372 crore allocation, aims to support this effort through public-private partnerships, but resources remain limited.

Experts are divided on the necessity and feasibility of a homegrown LLM. While some highlight the importance of fostering domestic AI capabilities and enhancing cyber sovereignty, others question the economic viability, citing the high cost of training and limited market size. The potential for job displacement through automation also remains unclear.

A significant challenge lies in the scarcity of high-quality data in Indian languages. Most existing LLMs rely on vast amounts of English-language data, leading to a bias towards Western contexts. While some Indian companies are actively working to address this data deficit by collecting and labeling data in various Indian languages, it is a substantial undertaking.

The article also discusses the brain drain of top Indian research talent to other countries. This, combined with underfunded academic ecosystems, further hinders progress. Ultimately, the success of India’s LLM ambition hinges on addressing these challenges: securing sufficient funding, developing a robust data ecosystem, retaining its top talent, and ensuring that any resulting technology is both affordable and beneficial to all Indians. The question, as one expert notes, is not *if* India should attempt this Herculean task, but *how* it can best orchestrate the resources and expertise to achieve it.

Read More