Our Terms & Conditions | Our Privacy Policy
India has the potential to become the “intelligence capital of the world”, where AI research, development, and deployment can flourish, Vishal Dhupar, managing director, Asia South, Nvidia, tells Surabhi Agarwal.
Dhupar believes India is witnessing a growing adoption of accelerated computing infrastructure but needs more investments. To compete effectively in the AI race and become an AI powerhouse, India must invest heavily in building its own computing capabilities, including servers, storage, and networking, he says.
Edited Excerpts
What role do you see India playing in the AI ecosystem in the future?
India, with its large population, digital infrastructure, and growing tech talent pool, is poised to become a significant player in the AI landscape. Nvidia envisions India as a potential “intelligence capital of the world”, where AI research, development, and deployment can flourish. This would involve leveraging India’s strengths in data, talent, and computational resources to address grand challenges and create global impact.Given the global frenzy for GPUs, how prepared is India to transition into an AI-driven economy and compete on a global scale?
India is witnessing a growing adoption of accelerated computing infrastructure. There are emerging cloud service providers investing heavily in this technology, making it accessible to a wider audience. This increased availability is driving further interest and adoption within the country. Accelerated computing is the future of computing, driven by the need for higher performance and efficiency. This technology, combining CPUs, GPUs, and DPUs, is essential for handling massive datasets and complex workloads. It’s crucial for driving innovation and addressing global challenges.
Discover the stories of your interest
How can India bridge the gap between its strong research foundation in AI and its relatively smaller presence in the global AI industry?
The ratio of funding in the global North is dramatically different than the global South. The ideas that these companies are coming up to disrupt are tremendously novel. More importantly, they are scalable. We need to build the compute in India. Our IT industry is about $250 billion in India. India’s IT industry, while significant, is predominantly export-oriented and software-focused, maybe around $200 billion. With a domestic IT spending of only about 1% of the global total, the country lacks the necessary computational infrastructure to compete effectively in the AI race. To bridge this gap and become an AI powerhouse, India must invest heavily in building its own computing capabilities, including servers, storage, and networking.
How is the cost-effectiveness of accelerated compute compared to traditional computing for training large language models?
Accelerated compute can reduce the cost by 10 times every year due to its efficiency and traditional general-purpose computing only improves efficiency by twice every two years. For example: Training a large language model with $10 million, a general-purpose compute requires 1000 nodes, consuming 13 gigawatts per hour whereas accelerated compute: requires only 10 nodes, consuming 3.2 gigawatts per hour, and can train 44 large models. India should prioritize accelerated computing to drive innovation and address national challenges.
Research shows that there is approximately about $250 billion worth of compute per year that gets purchased. So, let’s assume four years is the depreciation value. That is almost $1 trillion worth of compute. By investing in this technology, India can significantly reduce costs, expedite research, and develop solutions for critical areas like healthcare, education, and climate change. This will not only benefit the nation but also position India as a global leader in AI and technology.
What should be India’s overall approach to maximize AI development and adoption?
As an overall strategy, India should focus on identification of grand challenges that AI can help address at scale, solution & use case development, and building a robust ecosystem and accelerated computing infrastructure. The need is extremely high because it’s transformational. I think India definitely has an opportunity. I believe our energy requirements are going on substantially. A lot of them are being met through renewable energy.
And absolutely GPUs are the tool of information and there’s a demand for it. We are ruthlessly working towards increasing as much of the supplies as possible. In December 2023, when we had made a formal press announcement with Yotta, we had given a number of GPUs of time length, the first tranche in February, the second tranche in June, the final fit out by 2025.
Images are for reference only.Images and contents gathered automatic from google or 3rd party sources.All rights on the images and contents are with their legal original owners.
Aggregated From –
Comments are closed.