New Delhi: Calling the ongoing AI-led transformation the “fifth industrial revolution,” Union Minister for Electronics and IT Ashwini Vaishnaw said India’s approach to artificial intelligence is rooted in economics, scale and return on investment rather than a race to build the largest foundational models.
Speaking at the World Economic Forum in Davos, Vaishnaw said AI must be viewed as a multi-layered architecture comprising five distinct layers – the application layer, model layer, chip layer, infrastructure layer and energy layer. India, he said, is making “very good progress” across all five.
At the application layer, Vaishnaw asserted that India is well-positioned to become the world’s largest supplier of AI-driven services. “Go to an enterprise, understand the business… and provide that service using AI applications. That’s where ROI comes from,” he said, adding that returns do not necessarily come from building massive models.
According to him, nearly 95 per cent of enterprise use-cases can be addressed using models with 20 to 50 billion parameters, many of which India already has in deployment across sectors to improve productivity and efficiency.
Emphasising AI diffusion over concentration, Vaishnaw said India’s strategy is focused on widespread adoption. He pushed back against global assessments that place India in a secondary tier of AI readiness, noting that Stanford ranks India third in AI penetration and second in AI talent. “It’s actually in the first,” he said.
On the geopolitical narrative around AI, the minister questioned the assumption that control over large models translates into strategic power.
Even if a country were to “switch off” a large model, India would remain unaffected due to its own “bouquet of models” capable of meeting most domestic needs. “Does creating a large model give you geopolitical power? I don’t think so,” he said, warning that the economics of maintaining such models could even push some developers towards financial distress.
Vaishnaw said the economics of what he calls the fifth industrial revolution will be driven by deploying the “lowest cost solution” for the “highest possible return.” He pointed out that a 50-billion-parameter model can run on a single GPU, while a 30-billion-parameter model – sufficient for 80 per cent of tasks – may not require a GPU at all. With CPUs widely available and custom silicon emerging from multiple countries, dependence on any single nation, he argued, is diminishing.
Addressing India’s biggest constraint in AI- access to GPUs – Vaishnaw highlighted a government-led public-private partnership that has empanelled 38,000 GPUs as a common compute facility. Subsidised by the government, the facility is accessible to students, researchers and startups at nearly one-third the cost seen in many other countries, where big tech firms control access.
Beyond infrastructure, Vaishnaw outlined a four-pronged push: developing fit-for-purpose models, training 10 million people in AI skills, enabling the IT services industry to pivot towards AI-led efficiency, and embedding AI across “every sphere of life and economy.”
On regulation, the minister stressed the need for a “techno-legal approach.” Simple legislation, he said, is insufficient for complex technologies like AI. India is developing technical solutions to mitigate bias, detect deepfakes with court-admissible accuracy, and enable proper “unlearning” before enterprise deployment.
“We are following a techno-legal approach,” Vaishnaw said, underscoring India’s intent to balance innovation with safeguards in the AI era.
However, it should be noted that Vaishnaw also met IBM CEO Arvind Krishna and Meta Chief Global Affairs Officer Joel Kaplan in Davos to discuss India’s growing role in the global technology landscape.
Talks with Meta focused on protecting social media users from deepfakes and AI-generated content, with the company briefing the minister on measures to enhance user safety.
Vaishnaw said global assessments often understate India’s AI position, noting that Stanford ranks the country third in AI penetration and preparedness and among the top in AI talent. India is developing 12 AI models ranging from 50 billion to 120 billion parameters to meet enterprise needs across sectors.
Vaishnaw also pointed to innovations such as sound-to-sound AI and said India’s startup ecosystem, with over two lakh startups, places it among the world’s top three innovation hubs.
Discussions with IBM centred on collaboration in advanced semiconductor technologies, including 7 nm and 2 nm chips, and on strengthening India’s semiconductor talent pool.
Highlighting progress across the semiconductor value chain, he said work is underway at 10 semiconductor plants, three of which have begun pilot production, while four more are expected to enter pilot and commercial production this year. Vaishnaw also pointed to innovations such as sound-to-sound AI and said India’s startup ecosystem, with over two lakh startups, places it among the world’s top three innovation hubs.
Additionally, following his meeting with Google Cloud CEO Thomas Kurian at Davos, the minister underscored the growing confidence of global companies in India as a dependable supply-chain partner. He then stated, “Google is strengthening its commitment to India’s AI ecosystem, including a $15B AI data centre in Vizag & partnerships with Indian startups.”
However, the remarks by the minister at Davos underline India’s bid to shape the fifth industrial revolution through cost-efficient AI, broad-based adoption and strong domestic capabilities, positioning the country as a global hub for applied innovation rather than model-scale dominance.
