Opinion In the AI race, what India needs to do

India has 4,20,000 AI professionals — a force larger than many nations’ entire tech sectors. A 92 per cent AI adoption rate amongst enterprises — the highest in the world. And a $17-billion market potential. These are the building blocks of a new technological superpower. The government’s commitment is unequivocal; its IndiaAI Mission is a clear declaration of intent. Our developer community is the second largest in the world, behind only the US. India has 240-plus Gen AI startups, of which 70 per cent cater to industry-specific challenges in sectors such as healthcare, education, BFSI, and agriculture. For instance, Sarvam AI is working on building a foundational model with Indian languages. Niramai is a health-tech startup using AI to detect early signs of breast cancer in women. BHASHINI is breaking language barriers with built-in support for 22-plus languages to make communication more inclusive. It has already powered over 100 million inferences, proving its widespread use and reliability.

With such a vibrant ecosystem of innovation, we have the necessary foundation to catalyse both the pace and scale of a new wave of digital transformation. India’s story is one of a billion-plus people fundamentally reimagining its digital architecture in less than a decade. First, look at what we have achieved in financial inclusion — we took bank-account penetration from 30 per cent to over 80 per cent in just seven years. The World Bank and the IMF said this should take 47. By rethinking DPI, we cut the cost of opening a bank account from $23 to just 15 cents. Second, examine the scale of digital payments — $568 billion in monthly UPI transactions, with India handling 49 per cent of global real-time payments. More than just the digitisation of existing payment flows, India has created entirely new economic behaviours and possibilities. This infrastructure has created a completely new generation of technology companies. We built close to 108 unicorns in this period, powered by DPI. From payments, we moved to healthcare startups using the health stack, lending platforms leveraging the account aggregator framework, and commerce companies building on top of open networks. Third, look at how DPI proved transformative during a crisis. During the pandemic, we could instantly transfer $4.5 billion to 160 million people. From construction workers to farmers, from women to rural workers — India facilitated direct, instant, and leakage-free transfers when they needed it most.

On this open-source, interoperable infrastructure, we created an ecosystem that allows both public and private innovation to flourish. The results speak for themselves: 14 million businesses registered on GST, 863 million internet users, and an expected 13.42 per cent digital economy growth rate. With a globally recognised talent pool, relatively low electricity costs, abundant land, a vast reservoir of data, and strong government support for an aspirational entrepreneurial vision, we should lead this race.

A foundational pillar of the AI ecosystem that demands urgent attention is hardware. GPUs are the brain of AI and the key pillar on which machine intelligence is built. The Biden administration, in one of its last orders, issued the AI diffusion rule, thereby putting severe controls and restrictions on the import and export of GPU computing facilities. Despite India being a Quad partner, the US put it under the Tier II category — restricting the import of advanced GPUs as well as large-scale training of frontier AI models. This places a severe handicap on India’s ability to develop and grow as an AI power. India must take decisive steps to build and fortify the hardware backbone powering AI systems, ensuring that they deliver unmatched efficiency, reliability, and scalability. Owning our AI hardware capabilities is a necessity for securing independence in an increasingly tech-driven world. Investments in AI hardware will create jobs and attract significant capital, while also sparking a ripple effect of innovation across industries. By developing robust hardware ecosystems, India can position itself as an indispensable partner in global supply chains, amplifying its influence on the international stage. Nurturing cutting-edge research and cultivating homegrown expertise will unlock breakthroughs that solidify India’s leadership in AI hardware innovation. Finally, championing open-hardware ecosystems will foster collaboration and inclusivity, ensuring accessibility for innovators and developers around the world. To lead the AI revolution, we must understand this truth: Hardware is more than an accessory — it is the engine driving it all.

To maintain our sovereignty in technological advancement and lead from the front, we must not allow ourselves to become a technological colony of the US and China. This requires us to do the following. One, India must innovate in a nimble, less energy-consuming, cost-effective manner. We must do more for less like ISRO. Two, open source is the way forward, but we must create an environment that encourages brilliant and clever engineering. We must converge software with computing power to achieve solutions for pressing challenges like improving learning and health outcomes at the lowest cost. Three, we must build sovereign frontier models that are based on our data sets and do not have any inherent biases. It should be an end-to-end AI ecosystem and not just the application layer. Four, since India has 22 recognised official languages and a vast number of local dialects, we must build multilingual and multimodal foundational models. Five, as a Quad partner, India should secure its position among Tier-I countries in the realm of AI diffusion, and not be subjected to any restrictions or controls. Six, there must be a sense of great urgency and a mission-driven approach.

When future generations write the history of AI, they must mark this decade as the moment when India transformed from a service provider to a global innovator.

The writer is India’s G20 Sherpa and former CEO, NITI Aayog. Views are personal