The recent India AI Impact Summit held in Delhi had its share of hiccups, but it also accomplished many things. Policymakers articulated a vision for India’s AI future, and India received sundry AI-related pledges that could add up to investments in the range of $250 billion in AI-infrastructure and $20 billion in deep tech venture capital.
This needs to be placed in the broader context of what has been attempted for several years. AI and its multitude of potential applications must be usable at scale, under real-world conditions. And, there is no better place to accomplish this.
There is a simple reason why every global IT and consumer electronics major has a large India footprint, and why India features prominently in all their future plans. That is scale. A market with over 700 million smartphone users, the overwhelming majority of whom are young and open to digital experimentation, is a near-perfect test bed for digital ideas and implementations. OpenAI says 100 million Indians use ChatGPT multiple times a week for help with writing, learning and coding, which indicates the enthusiasm for new technology.
The size of the market leads to multinational organisations offering early access to products, and this gives India a headstart when it comes to exposure to new concepts. But transforming exposure into innovation is a tough task. In previous technology iterations such as 4G and 5G, India was just a very big market rather than being at the cutting edge of research and development (R&D).
Policy changes over the past three years have been designed to alter that dynamic. A focus on generating intellectual property (IP) and local capacity has led to initiatives such as the production-linked incentive scheme, R&D on futuristic platforms such as 6G, and the encouragement of start-ups through various policy tweaks. This has started to pay dividends. The AI Summit itself showcased many interesting start-ups operating across spaces as diverse as healthcare, education and many types of enterprise solutions.
But now, where AI is concerned, India is looking to carry out a massive experiment. It will build public compute capacity beyond the existing 38,000 graphics processing units (GPUs), with an additional 20,000 GPUs to be added soon, and more hereafter.
While other nations also have advanced AI infrastructure, the Indian twist is that it is going to implement a new model in terms of offering access to that infrastructure. In most nations, AI infrastructure is concentrated in the hands of a few giant corporations. In contrast, the IndiaAI Mission aims to offer access to compute at a reasonable rent of Rs 65 per hour, allowing it to be used by start-ups, researchers, students and public institutions. This democratisation of technology could spark a widespread culture of innovation across the AI ecosystem, transforming the market into a vast research lab for generating IP.
On the people side of things, India already possesses some of the basic prerequisites. It has a large, young and curious digital population. It has “bench-strength” in its skilled technical workforce. Subsidised access to compute could empower people in ways that are unimaginable.
As AI infrastructure must be backed up by big build-outs in telecom networks and power grids, the creation of manufacturing capacity to support internet of things, and ultimately, the ability to locally design and produce high-end semiconductors, is critical. None of this will be easy. But given political will, the AI Summit indicates that technical skills and financial investments will be available to follow through.
This could be one of the most critical policy experiments of the 21st century.
