Nvidia AI Summit 2024 Highlights: CEO Jensen Huang, Mukesh Ambani discuss India’s potential as global AI hub

"To have such a large population and large industry of computer scientists at a time when this computing industry is going to become the intelligence industry. Leveraging on everything you have, everything you know, your indigenous advantage, and enormous amounts of data as well as a large population of consumers to drive that flywheel of intelligence to data and data to intelligence, and to have the national will to go do something about it is extraordinary," Huang says on India's advantages in the "intelligence revolution".

On the sidelines of NVIDIA AI Summit 2024, Zoho Corporation has confirmed that it will leverage NVIDIA NeMo, a part of NVIDIA AI Enterprise software to build and deploy a custom large language model (LLM) for its software-as-a-service (SaaS) applications and will invest over 10 million US dollars in the coming year.

How can we help India transform into a centre of not just IT but a centre of AI, Huang asks Ambani.

"From my experience, as our Prime Minister said, this is a new aspirational India. We are one of the few countries in the world with an average age below 35. The Prime Minister's vision has been crucial in driving progress on the ground. Indians possess raw talent, and over the years, India has become home to a wide range of companies. Many energy giants conduct their innovations here, making us a fast-emerging innovation hub," Ambani says.

"We have also been fortunate to have strong connectivity infrastructure. Today, apart from the US and China, India has some of the best digital connectivity, including 4G, 5G, and broadband networks. While our company initially did not operate in this domain, we have now become the largest data company in the world."

"Our volumes are equal to AT&T and Verizon combined," he adds to which Huang retorts that the market size is quite an advantage.

NVIDIA CEO Jensen Huang and Akshay Kumar talk martial arts and growing up in Thailand.

In order to build an AI ecosystem in any country, you have to start with the infrastructure. We will soon have twenty times more compute here in India than anywhere else, Huang says. Once India figures out how to do large language models, you can figure it out in any part of the world, he adds.

The scaling law comes from the fact that the more data you have to train an LLM, the correspondingly large model size, the larger the model has to be. Each year we are increasing the amount of data and model size by a factor of 2 so the computation power has to increase by a factor of 4. We are now moving technology at a rate of four times every year over a course of ten years. We continue to find that AI gets smarter as we scale up the training size.

When we use ChatGPT as a one-shot with a prompt, the AI processes through a very large neural network and provides a sequence of answers. With Strawberry, we realised thinking results in higher quality answers. We've now discovered another scaling law, Huang says.

Billions of dollars have been invested in AI today because of this one instrument, Huang says.

NVIDIA CEO Jensen Huang is on stage right now at the AI summit in Mumbai, India.