In 2021, under the Programme for Development of Semiconductors and Display Manufacturing Ecosystem in India, a total of Rs 76,000 crore was allotted to give financial incentives and infrastructure support to companies for semiconductor development. 

The Current Landscape of Semiconductor Manufacturing in India

Fast forward to 2024, India has made significant progress in attracting global semiconductor giants to set up manufacturing facilities within its borders. Companies like Advanced Micro Devices (AMD), Micron, and Qualcomm have  committed to investments in India.

Notably, the Tata Group, in partnership with Taiwan's Powerchip Semiconductor Manufacturing Corp (PSMC), is set to establish India's first semiconductor fabrication unit in Dholera, Gujarat, with an investment of Rs 91,000 crore. This facility will produce 300 crore chips annually, catering to various sectors, including automotive, telecom, defence, and consumer electronics.

In addition to the Tata-PSMC fab, other significant projects are underway, including a semiconductor assembly and test unit in Assam, a joint venture between Renesas Electronics (Japan) and CG Power Plant in Sanand, Gujarat, and a DRAM and NAND assembly and test facility in Gujarat by Micron-Tata Projects. 

These upcoming facilities are crucial to the ‘Make in India’ vision of the government, an official working closely on these projects told The Secretariat.

But none of these semiconductor facilities will be producing artificial intelligence (AI) chips, added the official. Not now, and surely not anytime soon.

The Future Of AI Chip Production In India

Despite the progress in semiconductor manufacturing, India's position in production of AI chips, which are also known as semiconductors, remains in its infancy. 

AI chips are specialised semiconductors designed to handle the complex computations required for AI applications. For example, it’s a popular fact that OpenAI used approximately 10,000 Nvidia graphic processing units to train the first iteration of ChatGPT.

Currently, most AI chip design occurs in the United States, while production is concentrated in Asian countries like Taiwan and South Korea. India's capability to produce AI chips is still nascent.

The challenges are multifaceted. Firstly, designing AI chips requires cutting-edge technology and expertise that few countries possess. 

According to an AI expert that The Secretariat spoke to, developing domestic AI chips involves collaboration with global firms, as reference designs must be compatible with a wide range of global software platforms. This process requires extensive collaboration and obviously can be done only when one of the big global AI chip giants agrees to such a deal.

It was in 1976 when the then Cabinet of India, under the leadership of Prime Minister Indira Gandhi, granted its assent to the formation of Semi-Conductor Laboratory (SCL) in Mohali, Punjab.

But by the 1990s, India lost its early lead in semiconductor manufacturing due to several setbacks. A major fire in 1989 damaged the SCL, and while it was partially revived, it only produced a small number of chips for ISRO. 

India needs to form global partnerships to gain the technology and expertise it needs —something it's already doing in semiconductor manufacturing for electronics, but not yet specifically for AI.

Moreover, the path to AI chip production is expensive and time-consuming. The semiconductor production-linked incentives announced by the Indian government have led to the establishment of only one chip fabrication plant (where semiconductors will be manufactured) so far —the Tata-PSMC fab unit. The other three government-led projects are assembly and testing units.

The Tata-PSMC fab unit is expected to become operational in at least three years. This also highlights the long gestation and costly nature of semiconductor manufacturing.

However, there are optimistic signs for the future. Tech entrepreneurs like Bhavish Aggarwal, CEO of Ola Cabs, are stepping into the AI chip arena. Aggarwal's AI startup, Krutrim, plans to design, build, and launch its first AI chip, "Bodhi," by 2026.

Ola's big plan is to dive headfirst into the chip-making game, taking on heavyweights like Nvidia, Intel, Qualcomm, and AMD—who’ve been at it for decades, powering everything AI. 

However, to convert its ambitious dream into a reality, the cab-ride firm is partnering with Untether AI, which doesn’t have a particularly well-established AI tech frontier and works off Nvidia’s playbook.

The challenges for India too are immense, but the country is trying its hand at investing and forging global partnerships. The question remains if it can scale up to emerge as a key player in the global semiconductor and AI chip markets. And how soon?

Jensen Huang, CEO of Nvidia, said at a New York Times event last year that the United States (which is at the forefront of designing chips but not producing them) is around 20 years away from supply chain independence, calling it an extremely challenging feat.

If it takes the US two decades to achieve self-reliance in semiconductors, how long might it take for India to reach its Atmanirbhar Bharat vision in chip making and AI into reality?

Please click link to read another article on AI developments:- How Consumers Can Influence Who Controls Artificial Intelligence And Cut Out Bias