Artificial intelligence for a Viksit Bharat: Paris to Delhi via Washington

Two significant developments last week underscore how progress in artificial intelligence (AI) has become the central focus of tech-diplomacy with India standing at the crossroads. The Paris AI Action Summit 2025 held in France saw Prime Minister Narendra Modi not only co-chair the summit with French President Emmanuel Macron but also hold centre stage at the summit with his speech laying out where India stood in the global debate over AI safety versus AI opportunities.

 

  The Prime Minister’s exhortation to the audience on how each wave of new technology saw newer kinds of jobs being created laid the ground for what followed at the summit, revealing a geo-political divide. The vice president of the United States, JD Vance, in his speech, not only applauded PM Modi’s remarks on AI opportunities over fears but also went further ahead to rebuke the European approach of burdensome regulations that may have stifled innovation and inhibited European competitiveness. There was also clear divergence of views between the United States and the rest of the global community over the extent to which governments should regulate AI, the need for openness and transparency of models and the debate around massive power requirements for AI data centres and the sustainability issues thereof. 

Reflective of how India stood quite literally at the centre of this debate, holding the middle ground between the United States and Europe, was the offer for India to host the next AI Action Summit. If India’s position on AI opportunities was closer to the United States, its views on sustainability were perhaps closer to Europe. Yet, the adeptness with which PM Modi and his team have conducted tech-diplomacy is reflected in the joint statement with the United States released after his visit to Washington DC and his meetings with US President Donald Trump. A substantial portion of the joint statement, while being dedicated to technology, dwelt at length on AI and the infusion of AI data centre infrastructure technology and AI chips from the US to India. It also focused on the need to leverage civil nuclear technology to address the growing power requirements for AI in India, perhaps reviving the Indo-US nuclear deal, and a new focus on small and medium nuclear reactors. With TRUST replacing iCET as the fulcrum of India-US collaboration on emerging technologies it would be interesting to see how India navigates the imperatives of an Aatmanirbhar Bharat with the thrust for American leadership in technology. 

At the heart of these seemingly contradictory bilateral priorities is the IndiaAI Mission with its twin projects to put in place a public-private model of compute capacity for AI development, based mostly on GPUs from Nvidia, and its call for proposals on Indian Foundation Models in the quest for a sovereign large language model (LLM). Prime Minister Modi’s speech in Paris spelled out a LLM out of India as a priority, while information technology minister Ashwini Vaishnaw, in multiple public comments, has indicated a 7- to 8-month timeframe. With DeepSeek out of China making news globally, a chorus of demands has grown on the quest for an LLM out of India. This has put the spotlight on the various public-private efforts, starting with the AI for Bharat program in IIT Madras, which saw the incubation of Sarvam.ai and the BharatGen program in IIT Bombay that has received support from the Department of Science and Technology. PM Modi’s speeches have seen the use of AI to translate into multiple languages, leveraging MeiTY’s Bhashini program, while Anuvadini, a platform that has emerged out of AICTE, has been instrumental in the effort to make textbooks available in multiple languages using AI across levels — from school education all the way up to the university level. While much of the focus of these efforts has been primarily on text and, to some extent, on spoken audio, it is also noteworthy that a few years back, Prasar Bharati had made its entire archive of audio and video of Doordarshan to be used for training in the development of AI models. 

IndiaAI Mission must objectively evaluate its priorities on the kind of foundational models that need sovereign funding. It is clear from the global experience that the differential gap is quite vast between India’s modest aspirations and the install base in both the US and China of compute capacity, clusters of datacentres, power generation capacity and the maturity of LLMs already developed and rolled out. Bridging this gap with general models trained on vast amounts of public information would be a game of shifting the goalposts, with India trying to play catchup along the same path as the US and China. 

Mobile telephony to the digital economy, India’s path to technology transformation has historically been through quantum leapfrogs rather than through linear catchups. It would perhaps be prudent for the IndiaAI Mission to prioritise its funding towards foundational models that are more domain-focused and specialised to the Indian context, with the greatest potential to power a leapfrog rather than try to play catchup with DeepSeek or Open AI on generative AI and general reasoning. One way to narrow down on these priorities would be to determine AI capabilities that are best suited to accelerate India’s developmental goals towards a Viksit Bharat by 2047 as envisioned by Prime Minister Modi while accounting for India’s unique demographic context and scale. As an example, India does not need LLMs and reasoning models to solve math Olympiad problems over the Cloud at the scale of a Billion. However, a Viksit Bharat needs the ability to massively educate, upskill and empower at a scale of Billion through natural interfaces in native languages running locally on a low-cost mobile device. 

When the AI Action Summit next comes to India, it would be a fitting tribute to PM Modi’s vision for a Viksit Bharat, if the sovereign foundational models funded by the IndiaAI mission are able to demonstrate both the ability for a quantum leapfrog in technology by India and also the potential to accelerate India’s development trajectory through societal scale AI digital public goods. 

The writer is former CEO of  Prasar Bharati