India is positioning itself as a hub of artificial intelligence development through the IndiaAI Mission, which is highlighted at the AI Impact Summit in New Delhi. The government initiative aims to increase compute capacity, develop sovereign artificial intelligence datasets and frontier models, and promote education, startup innovation and trustworthy artificial intelligence frameworks. Under the IndiaAI Compute Pillar, the country is expanding national artificial intelligence cloud offerings with systems that include tens of thousands of NVIDIA GPUs, with capacity earmarked for model builders, startups, researchers and enterprises to train, fine-tune and deploy models inside India.
NVIDIA is working with cloud providers Yotta, L&T and E2E Networks to build advanced artificial intelligence “factories” and sovereign cloud platforms. Yotta is creating Shakti Cloud, a sovereign artificial intelligence infrastructure powered by over 20,000 NVIDIA Blackwell Ultra GPUs across campuses in Navi Mumbai and Greater Noida, offering GPU-dense, high-bandwidth services on a pay-per-use basis for Indian enterprises and public sector users. E2E Networks is deploying an NVIDIA Blackwell GPU cluster on its TIR platform at the L&T Vyoma Data Center in Chennai, incorporating NVIDIA HGX B200 systems, NVIDIA Enterprise software and NVIDIA Nemotron open models to support sectors such as healthcare, finance, manufacturing and agriculture. Netweb Technologies is launching Tyrone Camarero artificial intelligence supercomputing systems built on the NVIDIA Grace Blackwell architecture, with NVIDIA GB200 NVL4 platforms manufactured in India featuring four NVIDIA Blackwell GPUs and two NVIDIA Grace CPUs to power scientific computing, training and inference.
The Innovation Center Pillar of the IndiaAI Mission focuses on foundation and frontier models trained on India-specific data and infrastructure, targeting a multilingual population that includes 22 constitutionally recognized languages and over 1,500 more recorded by the census. NVIDIA Nemotron open models, datasets and tools, including the Nemotron-Personas-India dataset of 21 million fully synthetic Indic personas, are enabling organizations to build speech, language and multimodal models for public, consumer and enterprise use. BharatGen has created a 17-billion-parameter mixture-of-experts model with NVIDIA NeMo and NeMo RL, while companies such as Chariot and Tech Mahindra are developing 8-billion-parameter models for real-time speech and education. Gnani.ai reports a 15x reduction in inference costs by fine-tuning NVIDIA Nemotron Speech models, allowing support for more than 10 million calls per day, and CoRover.ai’s platform supports around 10,000 concurrent users and more than 5,000 daily ticket bookings for railway services. Sarvam.ai is pre-training its Sarvam-3 series at 3B, 30B and 100B parameter scales on NVIDIA H100 GPUs and delivering sovereign models through its Pravah platform, while Zoho, Soket.ai, Commotion and the National Payments Corporation of India are integrating NVIDIA NeMo and Nemotron across enterprise, workflow and financial applications.
To strengthen artificial intelligence research and commercialization, NVIDIA is collaborating with the Anusandhan National Research Foundation to support the AI for Science & Engineering program and future initiatives, providing grantee institutions complimentary access to NVIDIA AI Enterprise software and technical mentorship through the NVIDIA AI Technology Center, along with bootcamps, workshops and hackathons. NVIDIA is also working with venture capital firms including Peak XV, Z47, Elevation Capital, Nexus Venture Partners and Accel India to identify and fund startups building artificial intelligence solutions for domestic and global markets, building on a base of more than 4,000 Indian artificial intelligence startups already in the NVIDIA Inception program. Developers can deploy Nemotron models on NVIDIA-accelerated infrastructure such as NVIDIA DGX Spark, now available in India through local partners and as a “Make in India” manufactured version via Netweb, which also runs sovereign models from Indian builders such as Sarvam.ai.
