BharatGen: India’s AI Manhattan Project in the Making

Syllabus: GS3/Science and Technology

Context

• The recent launch of BharatGen marks a transformative mission toward technological sovereignty and culturally rooted artificial intelligence systems, aiming to embed AI into the very fabric of India’s digital future.

About BharatGen

• It is the world’s first government-funded multimodal Large Language Model (LLM) initiative, launched under the IndiaAI Mission, focused on developing foundational AI models in:
  • Text (LLMs for Indian languages);
  • Speech (Text-to-Speech and Automatic Speech Recognition);
  • Vision-language systems (for multimodal understanding);
• It is spearheaded by the Department of Science and Technology (DST) under the National Mission on Interdisciplinary Cyber-Physical Systems (NM-ICPS).
• The project is led by IIT Bombay, with a consortium of premier institutions including IIIT Hyderabad, IIT Madras, IIT Kanpur, IIT Mandi, IIT Hyderabad, and IIM Indore.
  • It is being executed by the Technology Innovation Hub (TIH) Foundation for IoT and IoE at IIT Bombay, which serves as the central coordination hub. It oversees:
    • Model development across modalities;
    • Data collection and curation focused on Indian contexts;
    • Ecosystem partnerships for compute, talent, and deployment;
    • Governance and strategic planning;
• Budgetary Support: BharatGen has secured a staggering ₹988.6 crore in funding from MeitY, making it the largest beneficiary of the ₹1,500 crore national AI budget, under the IndiaAI Mission 2025.

Key Features & Importance

• Language Coverage and Inclusivity: Currently, BharatGen models support 9 Indian languages, and aims to cover all 22 scheduled Indian languages by June 2026.
  • It has already launched Param-1, a bilingual LLM trained on 5 trillion tokens in English and Hindi.
• Real-World Impact: BharatGen has already piloted applications in agriculture, governance, and defence, and these applications aim to be scaled across all states and districts, transforming public service delivery nationwide, once fully deployed.
• Redefining Digital Sovereignty Through AI: BharatGen seeks to redefine digital sovereignty through AI, much like the original Manhattan Project that redefined global power dynamics through nuclear science.
  • BharatGen is not just about building models — it’s about building infrastructure, policy, and public-good ecosystems.
• Supporting Digital Public Infrastructure: BharatGen can be intelligence by participation like Aadhaar was identity by participation.
  • It needs to evolve into a stack: APIs, developer toolkits, open frameworks, and deployment systems that seed innovation across state universities, startups, and grassroots hubs.

Related Concerns & Challenges

• Lack of Robust AI Regulation: India still lacks comprehensive AI-specific legislation, despite the scale of BharatGen.
  • The Digital Personal Data Protection Act (DPDPA) offers broad exemptions to the government, raising concerns about unchecked data processing and surveillance.
• Infrastructure Gaps: Limited access to high-performance GPUs, data centers, and compute resources could slow down development.
• Language and Inclusivity Challenges: Ensuring accuracy, cultural sensitivity, and regional relevance across hundreds of dialects is a massive challenge.
• Talent and Ecosystem Readiness: Deep tech projects need interdisciplinary expertise — from linguistics to ethics to engineering — which is still developing in India.
• Ethical and Governance Frameworks: There are concerns like who controls the models, how decisions are made, and how citizens can challenge AI-driven outcomes.
  • The risk of algorithmic bias, especially in sensitive domains like healthcare and governance, requires rigorous testing and redressal mechanisms.

Government Efforts & Progress To Overcome Above Challenges

• Strengthening AI Infrastructure: BharatGen has already secured 13,640 H100 GPUs under the IndiaAI Mission to support trillion-parameter models.
  • IBM and BharatGen are co-developing open-source, Indic-specific data workflows to streamline model training.
• Building Ethical and Regulatory Frameworks: IBM is helping implement enterprise-grade governance frameworks for responsible model development.
  • Experts recommend involving civil society and academia in shaping AI ethics policies.
• Enhancing Language Inclusivity: BharatGen is developing Domain-specific Small Language Models (SLMs) for agriculture, Ayurveda, legal, and finance sectors.
  • IBM and BharatGen are building systems that switch seamlessly between Indian languages while preserving context.
• Fostering Talent and Ecosystem Growth: BharatGen brings together IITs, IIITs, and IIMs to pool expertise across domains.
  • Open-source models and solution templates allow Indian startups to build AI tools for local contexts.
• Promoting Public Trust and Accessibility: Open-source Release: BharatGen plans to make its models publicly available to ensure transparency and accessibility.
  • IBM and BharatGen are creating templates for sectors like education, governance, and healthcare.
  • AI-powered platforms will support vernacular languages for inclusive public service delivery.

Source: BS