Nvidia chief executive officer Jensen Huang said that India should latch on to the new manufacturing industry of artificial intelligence (AI), by harvesting and processing data domestically for the sector -- and exporting the output.
"There's a new manufacturing industry. Don't wait 20 years to get into it. Start now by harvesting and processing data here and then exporting the intelligence... India should generate the AI tokens and export them," Huang said in Mumbai on October 24.
"I believe, in the long term, India will manufacture and export AI. That’s when India will become a massive business opportunity for us," he added.
Advertisement
He further said, "India wants to make money even when India is sleeping. If you are exporting labour, you are not making money when you are sleeping. Buy if you exporting intelligence you can make money even while sleeping."
As the world looks to double the data centre industry size to about $8 trillion, Nvidia believes India has an opportunity to leapfrog into accelerated computing.
While global data centre industry is around $4 trillion which is going to double, India has close to about 1-2 per cent of the total global capacity.
Responding to a question about the high energy resources needed for AI data centres, Huang said, "Mumbai doesn’t have excess power. But India has excess energy. Its not in Mumbai. We have to build the data centres where there is excess energy."
Advertisement
"Our partners in India are joining us to make fundamental infrastructure. In a year's time we will have 20 times more compute in India than a year ago," he added.
Nvidia has expanded tie-ups with India's big firms, such as Reliance Industries, and launched a lightweight artificial intelligence model for the widely-used Hindi language, as it looks to tap a growing market.
Nvidia is also rolling out the new small language model, dubbed Nemotron-4-Mini-Hindi-4B, with 4 billion parameters, for firms to use in developing their own AI models.
"The model was pruned, distilled and trained with a combination of real-world Hindi data, synthetic Hindi data and an equal amount of English data," the company said in a statement.
Indian IT services firm Tech Mahindra, is the first to use the Nvidia offering to develop a custom AI model called Indus 2.0, focused on Hindi and dozens of its dialects, the US company said.
The collaboration of Nvidia with Indian companies along with leveraging country's semiconductor design talent is to make tailored-for-India chips. Additionally, the partnership will also help in meeting the country's make in India's initiative to expand production of semiconductors.
Just one email a week