Key Points
- Government sanctions 93.18 lakh GPU hours for 237 AI projects under IndiaAI Mission
- Purchase order issued for 1.1 EFLOPS high-performance AI compute system in Delhi
- Twenty indigenous foundation models including Sarvam AI and BharatGen receive support
The government has sanctioned 93.18 lakh GPU hours across 237 projects and issued a purchase order for a 1.1 EFLOPS high-performance AI compute system as part of its efforts to address structural constraints in deploying artificial intelligence at scale, the Ministry of Electronics and Information Technology informed the Lok Sabha on Wednesday (12 August).
Jitin Prasada, minister of state for Electronics and Information Technology, told the lower house that India’s AI strategy is based on democratising technology to address India-centric challenges and create economic opportunities.
The government is cognizant of risks associated with limited domestic capabilities in AI research ecosystems and has adopted a multi-pronged approach under the IndiaAI Mission to build technological sovereignty across the AI value chain, he said.
The IndiaAI Mission, approved on 7 March 2024 with a total outlay of Rs 10,371.92 crore over five years, aims to establish a robust and inclusive AI ecosystem aligned with India’s development goals. The ministry has empanelled 15 compute service providers over four rounds to deliver subsidised compute support to Indian developers and researchers, the minister said.
The high-performance AI compute system of approximately 1.1 EFLOPS, a measure of supercomputer performance denoting one quintillion floating-point operations per second, will be installed at the National Informatics Centre Data Centre in Shastri Park, Delhi, Prasada informed.
Talent development
Under the talent development pillar, the ministry has awarded 686 fellowships across undergraduate, postgraduate and doctoral programmes in 178 institutions. The YUVA AI for All programme has been completed by over 26.5 lakh individuals, according to the government’s response.
Twenty-seven India Data and AI Labs have been established, supporting training for more than 3,500 students. The ministry has identified 188 additional labs for establishment across Industrial Training Institutes and polytechnics.
The government has undertaken a structured approach to mainstream AI in governance and priority sectors including agriculture, health and public service delivery, Prasada said. Through the IndiaAI Application Development Initiative, 11 national-level hackathons and innovation challenges have been launched across these sectors.
The IndiaAI Innovation Challenge focused on AI solutions in healthcare, agriculture, governance, learning disabilities and climate change. Thirty solutions were shortlisted in the first stage for pilot implementation with central ministries, states and public sector organisations. Five teams, one per sector, were selected for national-level deployment, with pilots underway across multiple states.
Cybercrime and governance
The IndiaAI I4C CyberGuard AI Hackathon, conducted with the Indian Cyber Crime Coordination Centre, developed AI solutions for classification of cybercrime complaints and identification of emerging crime patterns on the National Cyber Crime Reporting Portal. The shortlisted solution has been integrated with I4C systems.
The IndiaAI Face Authentication Challenge aims to enhance the integrity of public examinations through AI-enabled de-duplication and one-to-many photo verification. Two teams have been selected for deployment at the Union Public Service Commission, and deployment has commenced.
The IndiaAI Intelligent Document Processing Challenge aims to automate extraction, classification, validation and structuring of information from complex multilingual documents. Two shortlisted teams are being deployed at UPSC, with implementation underway.
The Cancer AI and Technology Challenge, in collaboration with the National Cancer Grid, aims to strengthen cancer screening, diagnostics, treatment support and healthcare operations. The top shortlisted teams are undergoing ethical review prior to implementation.
Sovereign AI models
On sovereign AI capabilities, Prasada told Parliament that 58 AI Centres of Excellence have been approved for establishment across states and union territories, of which 22 have been approved and initiated across 13 states and UTs.
By the numbers
- 93.18 lakh GPU hours
- compute capacity sanctioned for 237 AI projects
- Rs 10,372 crore
- total IndiaAI Mission outlay over five years
- 20 models
- indigenous foundation models receiving support
Twenty indigenous foundation model proposals, comprising 12 large multimodal models and eight small language models, have been identified for support. Models already launched include Sarvam AI’s Sarvam 30B and 105B models, BharatGen’s Param2 17B text model and Patram 7B vision language model, Gnani.AI’s 5B speech-to-speech model, and Zenteiq’s BrahmAI 1B tokeniser suite.
Other indigenous models receiving support include Varyaa for image and text to video generation, Avataar’s Varya 14B distilled video generation model, NeuroDX’s MANAS neurological screening models and Shodh AI’s 1.6B multi-scale physics prototype with agentic tool calling framework.
The PMJAY Auto Adjudication Hackathon, in collaboration with the National Health Authority, supports the development of AI-based adjudication frameworks for Pradhan Mantri Jan Arogya Yojana claims, resulting in multiple prototypes and pilot deployments, the ministry informed Parliament.
Your Questions, Answered
How much compute capacity has the IndiaAI Mission sanctioned?
The government has sanctioned 93.18 lakh GPU hours across 237 projects and issued a purchase order for a 1.1 EFLOPS high-performance AI compute system to be installed at the NIC Data Centre in Delhi.
What is the total budget for the IndiaAI Mission?
The IndiaAI Mission was approved on 7 March 2024 with a total outlay of Rs 10,371.92 crore over five years to establish a robust AI ecosystem aligned with India's development goals.
How many indigenous AI models are being supported under IndiaAI?
Twenty indigenous foundation model proposals, comprising 12 large multimodal models and eight small language models, have been identified for support. Notable models include Sarvam AI, BharatGen, Gnani.AI and Zenteiq's BrahmAI.
What talent development initiatives exist under IndiaAI Mission?
The mission has awarded 686 fellowships across 178 institutions, completed YUVA AI training for over 26.5 lakh individuals and established 27 India Data and AI Labs supporting training for more than 3,500 students.
