Global CEOs are prone to making effusive comments about the country they are visiting. But it would be churlish to put Nvidia Corp founder and CEO Jensen Huang’s comments on Thursday in the same category. Huang said India has a major role in shaping the future of artificial intelligence (AI) and should move beyond software outsourcing to become a hub for AI innovation. He rightly pointed out that all the ingredients of a successful AI revolution is present here. The task now for the government and industry is to ensure that all the pieces are put together in a cohesive format.
Consider the facts. By 2026, Apple Inc is expected to shift 26% of its global iPhone capacity to India. By the same time, five semiconductor projects worth `1.5 lakh crore would also come on stream. Apple’s requirement of chips for the iPhones it produces in the country is around $12 billion a year. It’s quite possible that by 2026, a large part of this can be sourced from Indian suppliers.
The technology supply chain is quite interdependent. If smartphones, information technology (IT) hardware, and semiconductors start getting manufactured domestically in large scale, AI chipsets and graphics processing units (GPUs), which are essential for creating large language AI models, will automatically follow.
This is what Huang seems to have hinted at when he said that just like software, India will export AI in future. Huang has put money where his mouth is. The string of partnerships which Nvidia has entered into with a range of Indian companies will help it in creating a strong base here. Simply put, just as Nvidia has attained leadership position today in GPUs by being an early starter, the same advantages will accrue to it by sewing up such pacts with firms here. Of course, the domestic firms stand to gain much more by being able to develop models suited for local needs.
At present, the US and China lead in computing infrastructure required for the development of AI technology. With Nvidia dominating the GPU market with about 88% market share, there is a lag of 12-18 months in getting GPUs from the company due to its high demand across the globe. Partnerships with Indian companies will perhaps reduce the wait time.
The other interesting part is that the government is also looking at manufacturing chipsets in the country in partnership with Nvidia. Earlier this year, the government approved a `10,372-crore AI mission to create sovereign capability in computing.
Naturally, any domestic manufacturing or sourcing of GPUs will be done on a public-private partnership mode as the government alone would not be able to do everything. Such an approach will have a two-fold benefit: while the government can focus on building a sovereign capability in AI by setting up a national AI computing centre, industry can focus on their business-centric models. To achieve success in making the country the hub of AI is possible as India has in the past demonstrated this in the case of mobile telephony.
It may have been a late starter in terms of opening the sector to private players, but its model of low average revenue per user and high volumes has been a great success. The rock bottom data rates of `12-13 per gigabyte, against the world average of `295, is surely a model that now needs to be replicated in AI computing too.