TLDR: India is on track to launch its own Large Language Models (LLMs) by the end of 2025, driven by the IndiaAI Mission. The government, acting as a facilitator, is backing foundational AI model development, with four entities already identified, including collaborations with IITs. This initiative is part of a broader vision to establish India as a global AI powerhouse by 2047, focusing on models tailored to the nation’s diverse linguistic and cultural needs.
India is making significant strides in its artificial intelligence (AI) landscape, with expectations for the country’s first homegrown Large Language Models (LLMs) to be operational by the close of 2025. This ambitious timeline is a cornerstone of the government’s strategy to foster innovation and achieve self-reliance in critical technologies, ultimately aiming for India to become a global AI powerhouse by 2047.
S. Krishnan, Secretary of the Ministry of Electronics and Information Technology (MeitY), emphasized India’s innovation-driven approach to AI, positioning the government primarily as a facilitator rather than a stringent regulator. Speaking at the Abhay Tripathi Memorial Lecture on ‘AI and Governance’, Krishnan highlighted that existing legal frameworks and industry self-regulation are deemed sufficient to address most regulatory concerns, cautioning against overregulation that could impede technological progress. He stated, “Our view is, use the existing laws, voluntary compliance, and make sure that we do not put in such controls as to make innovation problematic. Also, try and do something in (terms of) the regulation and legislation which will actually promote AI and enable it.”
The development of indigenous generative AI models is a key focus under the IndiaAI Mission. Krishnan confirmed that India is actively establishing its own foundational models, drawing parallels with global leaders like ChatGPT, Gemini, and DeepSeek. Four entities have already been identified to spearhead this effort, with two of them involving collaborations with Indian Institutes of Technology (IITs). The process of setting up these foundational models has commenced, with the goal of having India’s own LLM fully functional by the year-end.
Data protection remains a paramount concern in this AI-driven future. Krishnan underscored the critical role of the Digital Personal Data Protection (DPDP) Act in safeguarding personal data, especially given that AI heavily relies on data. He noted that the draft rules for the DPDP Act are already in the public domain.
Despite the rapid progress, challenges persist. Krishnan pointed out the prohibitive costs and scarcity of Graphics Processing Units (GPUs), which form the hardware backbone for AI compute. He recounted a conversation with Nvidia CEO Jensen Huang, who was primarily focused on selling GPUs, highlighting the contrast with India’s pursuit of more open and accessible AI approaches. Other hurdles include the fragmented nature of India’s data landscape and existing skill gaps within the workforce.
The IndiaAI Mission, launched in 2024, is backed by a substantial budget of over ₹10,000 crore (approximately USD 1.5 billion) annually, allocated to develop India’s AI ecosystem through public-private partnerships. A core aspect of this strategy is to develop and host LLMs within India, specifically tailored to cater to the social, cultural, and business diversity of its 1.46 billion people, who converse in 22 officially recognized languages. This aligns with the concept of ‘Sovereign AI’, which emphasizes a nation’s capacity for self-reliance in AI through its own infrastructure, data, workforce, and business networks, aiming to compete with global leaders like China and the USA.
Several Indian companies are at the forefront of this development. Sarvam AI, which is building India’s first homegrown sovereign LLM, has confirmed plans to open-source its models under permissible licenses through the IndiaAI Mission. Sarvam AI has received a significant subsidy of INR 98.68 crore for access to 4,096 Nvidia H100 GPUs for six months. Other startups selected by the IT ministry to build LLMs under the mission include Soket AI Labs, which is developing a 120 billion-parameter open-source text model; Gnani.ai, working on a 16 billion-parameter Voice AI foundational model; and Gan.AI, creating a 70 billion-parameter multilingual foundation model for ‘Superhuman TTS’ (text-to-speech). To further support innovation, a common compute facility will soon be launched, offering subsidized access to GPUs, with higher-end units costing Rs 150 per hour and lower-end ones at INR 115.85 per hour.
Also Read:
- India Advances Voluntary AI Ethics Framework Ahead of Global Summit
- Goa Unveils Ambitious AI Mission 2027 to Become South Asia’s AI Hub
This concerted effort underscores India’s commitment to building a robust and inclusive AI ecosystem that addresses its unique linguistic and cultural needs while fostering technological independence and global competitiveness.
