New Delhi: India sees itself in the first league of artificial intelligence (AI) nations, with its strategy centred on rapid diffusion of AI across the economy, affordable compute access and enterprise-level returns on investment, rather than a race to build the world’s largest models, Union minister for electronics and information technology Ashwini Vaishnaw said at the World Economic Forum (WEF) in Davos.
Speaking at a panel discussion titled “AI Power Play, No Referees”, Vaishnaw said AI power would increasingly be defined by economics and deployment at scale, not by sheer model size. He noted that nearly 95% of AI use cases can be addressed using models in the 20–50 billion parameter range, many of which India already possesses and is actively deploying across sectors.
The minister outlined a five-layer AI architecture—applications, models, chips, infrastructure and energy—saying India is making progress across all layers. At the application layer, he said, India is likely to emerge as one of the world’s largest suppliers of AI-enabled services, driven by productivity gains for enterprises rather than headline-grabbing model breakthroughs.
Highlighting constraints around compute availability, Vaishnaw said the government has enabled a public-private partnership to create a common national AI compute facility with around 38,000 GPUs. The subsidised facility provides access to students, researchers and startups at roughly one-third of prevailing global costs, aimed at accelerating innovation and adoption.
He said India’s AI strategy rests on four pillars: a shared compute infrastructure, a free bouquet of AI models addressing most practical needs, large-scale skilling programmes to train about 10 million people in AI, and enabling India’s IT industry to pivot towards AI-driven productivity for domestic and global clients.
On governance, Vaishnaw said India is adopting a techno-legal approach to AI regulation, combining legal frameworks with technical solutions to address risks such as bias and deepfakes. He stressed that detection systems must meet standards robust enough to withstand judicial scrutiny, signalling a stricter approach to AI accountability.
The comments come as countries worldwide debate AI dominance and regulation amid rapid technological advances. Vaishnaw cautioned against equating geopolitical power with ownership of large models, arguing that sustainable advantage will come from low-cost deployment, resilience and real economic returns.