India is stepping up its push to become a global hub for AI, with support from NVIDIA across infrastructure, research and startup development.
The announcements were made during the AI Impact Summit in New Delhi, where government officials, industry leaders, universities, and startups outlined how they will work with NVIDIA as part of the government-led IndiaAI Mission.
Expanding AI compute capacity
The IndiaAI Mission is backed by more than $1 billion in public funding to increase computing capacity, build sovereign datasets, support frontier AI models, and promote AI education and startup growth.
Under its compute initiative, India is building a large-scale AI cloud infrastructure powered by tens of thousands of NVIDIA GPUs. NVIDIA says it is working with cloud and data centre operators, including Yotta, Larsen & Toubro and E2E Networks, to expand AI infrastructure.
Yotta is developing “Shakti Cloud,” a sovereign AI cloud platform powered by more than 20,000 NVIDIA Blackwell Ultra GPUs. The company operates campuses in Navi Mumbai and Greater Noida, offering GPU-based AI training and inference services to enterprises and government customers on a pay-per-use basis.
Larsen & Toubro is building gigawatt-scale AI data centre infrastructure aligned with the IndiaAI Mission. Its expansion plans include a 30-megawatt facility in Chennai and a new 40-megawatt site in Mumbai. These facilities are intended to support sovereign cloud workloads and large-scale AI deployments.
E2E Networks is deploying an NVIDIA Blackwell GPU cluster on its TIR platform, hosted at the L&T Vyoma Data Centre in Chennai.
The platform will include NVIDIA HGX B200 systems, NVIDIA Enterprise software, and NVIDIA Nemotron open models to support AI development across sectors such as healthcare, finance, manufacturing, and agriculture.
In addition, Netweb Technologies is launching its Tyrone Camarero AI supercomputing systems built on the NVIDIA Grace Blackwell architecture. The NVIDIA GB200 NVL4 platforms, manufactured in India under the “Make in India” initiative, include four Blackwell GPUs and two Grace CPUs for scientific computing, model training and inference.
Building India’s frontier AI models
Another focus of the IndiaAI Mission is the development of foundation models trained on India-specific data and infrastructure.
India is home to 22 officially recognised languages and more than 1,500 additional languages recorded in the census. Several companies are using NVIDIA’s NeMo open models and NeMo tools to develop multilingual AI systems for government and enterprise use.
Among the organisations involved:
- BharatGen has built a 17-billion-parameter mixture-of-experts model using the NVIDIA NeMo framework.
- Chariot is developing an 8-billion-parameter text-to-speech model for real-time applications.
- Commotion, backed by Tata Communications, has developed an AI operating system for enterprise workflow automation using Nemotron models.
- CoRover.ai has deployed Nemotron Speech models and NVIDIA Riva libraries to support multilingual speech AI for the Indian Railway Catering and Tourism Corporation, handling around 10,000 concurrent users and over 5,000 ticket bookings per day.
- Gnani.ai is building a 14-billion-parameter speech-to-speech model using Nemotron tools, with plans to scale to 32 billion parameters.
- National Payments Corporation of India is exploring training “FiMi,” a financial AI model based on NVIDIA’s Nemon 3 Nano, to enhance multilingual digital payment support across India’s banking ecosystem.
- Sarvam.ai has open-sourced its Sarvam-3 models trained for 22 Indic languages, with parameter counts ranging from 3 billion to 100 billion. Training was conducted on NVIDIA H100 GPUs via cloud partners, including Yotta.
- Soket.ai uses NVIDIA Megatron and NeMo for large-scale model training.
- Tech Mahindra has built an 8-billion-parameter foundation model tailored for Indian languages, designed for educational use.
- Zoho is advancing its Zia LLM platform using NVIDIA NeMo across its business software products.
Developers can use NVIDIA-accelerated infrastructure, like DGX Spark systems, to deploy Nemotron models in India through partners such as PNY, RP Tech India, and Tech Data. DGX Spark can also run models developed by Indian companies, such as Sarvam.ai.
Government and research collaboration
NVIDIA is teaming up with the Anusandhan National Research Foundation to boost AI research in schools and universities. This partnership will give institutions access to NVIDIA’s AI software and offer support through mentorship, workshops, and bootcamps.
Additionally, NVIDIA is collaborating with venture capital firms such as Peak XV Partners and Accel India to identify and support new AI startups. Currently, over 4,000 Indian AI startups are part of NVIDIA’s Inception program.
These efforts aim to strengthen AI in India, enhance model development for multiple languages, and increase funding for research and startups, helping India establish itself as a key player in the global AI field.
