AI chips: India must get its basics right to meet this great catch-up challenge

Economies that deploy AI across the board would undergo a step change in their global competitiveness. Unlike earlier advances in information technology, whose national origin did not materially affect the ability of people and companies anywhere to use them, AI is seen as an enabler of geopolitical power. 

Bars are in place, denying others access not just to its algorithms, but also other key secrets held within, some of which may help us understand latent biases and the propensity of these tools to hallucinate. Hardware is also being held close. The US has barred the export of advanced microprocessors (GPUs) used for GenAI to China, for example, and put India on a list of countries that can only have rationed access to its cutting-edge chips. 

Also Read: Mint Quick Edit | An AI race shake-up spells an opportunity for India

It is in this light that India’s scramble to shore up its AI capability must be viewed.

As Mint reported on Wednesday, the government has got cracking on a plan to develop AI chips in India, with the Centre for Development of Advanced Computing (C-DAC) leading this chip-design effort. This is an autonomous body under the ministry of electronics and information technology, set up in 1988. It has developed supercomputers based on massive parallel processing. 

Whether it can pull off AI chips that will rival those of world-leader Nvidia cannot be taken for granted, but it is critical to shore up AI-chip capability by recruiting the right talent and giving them the equipment and incentives needed for success. 

Chip fabrication is difficult. Even to make chips, India has only been able to attract projects for wafers with circuit sizes above 20 nanometres, whereas those just a fifth of that size have typically been deployed for AI’s big models. As it’s far from clear if contract chip-makers like Taiwan’s TSMC can fabricate C-DAC’s designs, India will have to massively step up its R&D efforts to create fabs that can produce the kind of chips that run AI.

Also Read: India must wake up on basic R&D for technology before it gets too late

Nor is it only a question of hardware. China’s disruptive AI splash with its low-cost DeepSeek model, while suspected of using highly advanced chips on the sly, was probably enabled by a clever software-engineering workaround. 

That model was developed by the research unit of a Chinese hedge fund and points to highly evolved digital talent in China. That, in turn, depends on an education system that has achieved quality at all levels—from elementary school to specialized research institutions. Basic deep-delves have seen a thousand flowers bloom; AI papers are being discussed. 

Also Read: What America’s technology denial and China’s AI success imply for India

In contrast, Indian school education remains patchy, with excellence an exception rather than the rule. We have the largest pool of young people who could attain the proficiency needed in any field of study for basic breakthroughs. But realizing this potential requires vastly improved governance all across, significant investment in education and remuneration structures that incentivize talent to flow into academia. 

Raising India’s AI capability is an important aim. But it will take a lot more than the exertions of a single agency or ministry. It calls for a broad-spectrum effort, involving academia and risk capital, and also for appropriate regulation—so that AI, like education, serves all of society well.