NVIDIA unveiled several partnerships and initiatives to accelerate artificial intelligence (AI) development in India at this week’s AI Summit in Mumbai. During the summit, NVIDIA showcased its end-to-end platform and advancements in healthcare, drug discovery, and climate modeling. Here’s a roundup of the key announcements.
Strengthening AI Through Partnerships
NVIDIA is working with Indian companies like Yotta, Tata Communications, E2E Cloud, and Netweb Technologies to build AI factories, allowing businesses nationwide to integrate and scale AI. These AI factories, powered by NVIDIA’s platforms, will provide Indian companies with the computational resources they need for large-scale AI deployments.
Yotta launched Shakti Cloud, the first Indian platform offering NVIDIA’s intelligent management system (NIM) microservices for AI deployment. Yotta is offering $50,000 in credits to startups, with 80 percent of the support going to Indian companies.
Tata Communications is making NVIDIA’s AI software stack available on its infrastructure, enabling more companies to run AI workloads on accelerated computing systems. Meanwhile, Netweb launched its Tyrone Skylus Cloud instances, providing full software and hardware solutions for cloud customers. Netweb is an Indian original equipment manufacturer (OEM) that produces Ng X servers, NVIDIA’s open-source reference design for data center servers.
NVIDIA is also partnering with F5 to improve AI application delivery on India’s sovereign clouds. The partnership utilizes NVIDIA BlueField-3 data processing units (DPUs), which offload key tasks from CPUs, thereby boosting performance and security in data centers. Kubernetes is running on these DPUs to manage containerized AI applications more efficiently. This improves AI performance, reduces latency, and boosts security.
“There’s a fundamental shift in computing taking place. Traditionally, information was accessed from large databases and run through software algorithms written by human programmers. The result was always the same for a given input, and this type of computing was ideal for CPUs. But now, with the rise of AI factories, we’ve seen a different type of computing that’s largely generative. This type of computing involves AI models that train on vast data sets and make predictions on a range of different possible outcomes, and this type of computing is ideally suited for GPUs,”
Said Shar Narasimhan, director of data center GPUs and AI at NVIDIA.
Revolutionizing India’s Language Models
NVIDIA is helping various organizations in India develop large language models (LLMs) that cater to the country’s regional languages, using NVIDIA AI Enterprise software and accelerated computing. NVIDIA’s Neural Modules (NeMo) framework manages and improves these models, increasing their performance compared to typical models.
For example, Flipkart is customizing AI models, such as LLaMA 3.1, to improve the safety of its customer service systems using NeMo. Cunani has developed a multilingual AI model that handles 10 million voice interactions daily for over 150 companies in sectors like banking and insurance. Zoho is using NeMo to develop its own LLMs, which will serve more than 700,000 customers across 50 business applications. Servim AI is also building AI models that support 10 Indian languages for various industries.
Smaller language models are becoming popular because they are efficient while still meeting accuracy needs. NVIDIA introduced the NeMoTron Hindi model, trained with synthetic Hindi data to reflect regional and cultural context. Tech Mahindra is also developing the Indus 2.0 model, which supports Hindi dialects and will be used for AI applications in agriculture, healthcare, banking, and education.
Expanding AI in Healthcare
India’s healthcare sector, expected to grow to $300 billion by 2028, is adopting AI to improve patient care and research. Organizations like IIT Madras and 5C Network use generative AI models to assist in areas like antibiotic research, radiology, and simplifying patient medical reports.
These organizations use NVIDIA’s NeMo Hindi model to process large volumes of medical data and generate more accurate insights. For example, IIIT Delhi uses AI to study antimicrobial resistance, while 5C Network uses AI to help patients receive faster and more precise radiology reports.
Physical AI and Robotics: The Next Frontier
NVIDIA’s vision for physical AI, which combines AI with robotics, is poised to transform industries such as logistics and manufacturing. Physical AI relies on three key components: NVIDIA AI on DGX for training AI models, NVIDIA Omniverse on OBX to teach and validate models, and NVIDIA Jetson to run the AI on robotic systems.
India’s robotics ecosystem is growing, with nearly 30 startups in NVIDIA’s Inception program using these technologies to build robots for various applications. At AI Summit India, companies like Adverb, Autonomy, and ATI Motors demonstrated using NVIDIA’s platforms to create advanced robotic systems.
Adverb, a Reliance company, uses NVIDIA’s technology to build robots for logistics, which companies like Amazon are already deploying. Autonomy is developing robots for last-mile delivery, while ATI Motors is creating autonomous electric vehicles for factories and warehouses, all powered by AI.
Advancing Industrial AI with Digital Twins
NVIDIA’s Omniverse platform is widely used in industrial applications to create digital twins or virtual replicas of physical environments. Indian industrial companies are adopting this technology to optimize their operations.
For example, Ola Electric, one of India’s electric vehicle manufacturers, is using Omniverse to create a digital twin of its future factory, which will help it produce up to 10 million electric two-wheelers annually. Reliance is leveraging Omniverse to integrate its $10 billion green energy “gigafactories,” spread across 5,000 acres, into a virtual environment to optimize operations.
In addition, TCS is offering industrial AI services, including agentic AI for real-time monitoring and Omniverse-based digital twin services for optimizing factory and warehouse operations. Tech Mahindra has built a 300-person team to focus on digital twin technology, providing AI solutions to telecom and automotive companies.
Accelerating AI Innovation and Adoption
To support businesses in building AI-driven customer service solutions, NVIDIA introduced the Nim Agent Blueprint, which helps companies develop AI systems capable of handling complex multi-turn conversations. These AI agents can recall both short-term and long-term interactions, improving the customer experience by reducing the need for repetitive information sharing.
Infosys, TCS, Tech Mahindra, and Wipro are companies using NVIDIA AI Enterprise and its blueprints to create industry-specific AI solutions. Collectively, they are training over 500,000 professionals to work with NVIDIA’s AI tools and deploy them across various sectors, further boosting AI adoption in India.
“We already have interest from customers and partners in other regions in using this blueprint to advance customer service and to optimize performance and the responsiveness of their customer service team. It adds a lot of efficiency to the process because it gives customer service agents, the humans, real-time access to data and recommendations on how to handle customer questions”
Said Anne Hecht, senior director of enterprise AI at NVIDIA.
Meeting Growing Compute Demands
NVIDIA’s new “chain of thought” inference architecture drives exponential growth. The architecture allows AI models to simulate decision trees and explore different outcomes before arriving at the best solution. This is particularly useful for large-scale AI applications, but it requires more computational power for inference, like what was previously needed for training AI models.
To keep up with these growing demands, NVIDIA created the Blackwell platform, which includes chips, systems, and optimized AI containers. NVIDIA also announced that it will release new AI platforms on a one-year cycle, starting with Blackwell, followed by Blackwell Ultra, Vera, and Rubin in the coming years. This roadmap can help customers better plan their own AI initiatives as the updates will happen on a more regular cycle.
“We have been working here in India since 2004, when we made the city of Bangalore our first home. Today, we’re working with over 2000 inception partners, companies in deep tech who embrace NVIDIA to produce solutions, not just for India, but for the globe,”
Said Vishal Dhupar, managing director for South Asia at NVIDIA.
While there are many AI silicon providers, NVIDIA maintains a healthy market share and mind share lead. I’ve been asked many times as to how defensible that position is. What makes NVIDIA unique isn’t the GPU but the holistic way the company looks at accelerated computing. It provides the underlying silicon but also CUDA, its programming framework that makes it easier for developers to build applications. It also provides turnkey systems, such as Drive in automotive and Clara in healthcare, which speeds up the deployment of NVIDIA technology.
With the work it’s doing in India, a country likely to see its AI initiatives explode, the partnerships and work with key businesses should help unlock AI’s potential in the region and make it the partner of choice today and into the foreseeable future.
Get insights from one of the leading voices in tech! Connect with Zeus Kerravala, founder and principal analyst at ZK Research, for expert perspectives on the latest in AI, cloud, and digital transformation.
