At this year’s Davos, Union Minister Ashwini Vaishnau delivered a masterclass on strategic clarity. When IMF Managing Director Kristalina Georgieva placed India in the “second tier” of the AI economy, Vaishnow was unfazed. He completely restructured the conversation.
His argument was accurate. The power of AI cannot be measured only by the size of the model. It must be evaluated across five layers: application, model, chip, infrastructure, and energy. India is making progress on all five. But where India has an advantage is in the application layer, where AI brings real value to businesses and citizens.
“At the application layer, India will probably be the world’s largest service provider,” Vaishnau said. This is not a wish. It is a strategic position rooted in economic reality. And he’s right.
Application layer benefits
The application layer is where AI stops being a research paper and starts to be useful. This is where models are deployed into enterprise workflows to solve real-world problems. This is where the return on investment is generated.
Vaishnaw’s key insights are worth noting. “Ninety-five percent of the work can be done using models with 20 billion or 50 billion parameters.” Most companies don’t need frontier models with billions of parameters. They need AI to understand their specific business processes. You need a solution that integrates with your existing systems. Not a breakthrough in research, but a measurable increase in productivity.
This is India’s sweet spot. It’s no coincidence that this country’s $254 billion IT-BPM industry has become the world’s back office. It’s built on decades of understanding enterprise needs, designing business processes, and delivering solutions at scale.
The same capabilities that made India essential for digital transformation also make it essential for AI transformation.
Consider the challenges faced by global banks deploying AI for fraud detection. Maybe 10% of the problem is in the model. The remaining 90% is understanding the bank’s data architecture, integrating with legacy systems, managing regulatory compliance across jurisdictions, training operations teams, and iterating based on actual performance. This is exactly what Indian IT services companies have been doing for decades. AI changes tools. The content of the work remains unchanged.
what actually happens
Currently, over 10,000 radiology scans are processed every day across 1,500 healthcare facilities in India, and our AI is not meant to replace radiologists. It strengthens them. Our model detects 234 pathologies across imaging modalities. But the value we provide is not in the model itself. It’s in the clinical workflow that we’ve built around it.
AI prioritizes urgent cases so important discoveries reach clinicians faster. Audio transcription eliminates the need for manual reporting. Quality assurance personnel catch errors before the report reaches the patient. None of these required us to build a model larger than OpenAI or Google. To do so, we needed to understand how radiology actually works and incorporate AI precisely into those workflows.
This pattern repeats in all sectors. healthcare, financial services, manufacturing, retail, agriculture. The winner is not the one with the biggest model. They will be the ones most effectively implementing AI into the chaotic reality of running a business. India has 5.4 million IT professionals who are well aware of that troubling reality.
Why model size is the wrong metric
The insistence on frontier models reflects researchers’ views on AI, not economists’ views. The construction of GPT-5 is intellectually impressive. But the economic value of AI comes from large-scale adoption, not benchmark scores.
Vaishnau cited Stanford University’s rankings that put India third in the world in terms of AI penetration and second in terms of AI talent. These metrics are more important than model size because they measure actual adoption. Measure whether AI is transforming productivity across the economy.
The government’s IndiaAI mission will enhance this advantage. With 38,000 GPUs now available at a subsidized rate of around Rs 65 per hour, startups, researchers and businesses can now access world-class computing without exorbitant capital expenditures. This democratization of infrastructure will accelerate application layer innovation across the country.
India is also building large-scale indigenous language models through initiatives such as Sarvam, Gnani, and BharatGen. These models are designed for India’s linguistic diversity, trained on Indian data, and optimized for Indian use cases. Sovereignty is important. However, the actual development is more important.
Positioning for the fifth industrial revolution
Vaishnow positioned AI as a driving force in what he called the “Fifth Industrial Revolution.” Its economics, he argued, depends on return on investment, or whether it provides the best value at the lowest cost. This is not a game that favors those who spend the most on computing. Companies that implement technology most effectively will have an advantage.
India’s comparative advantage is clear. It’s a deep corporate relationship built over decades. Employees trained in process engineering and business transformation. A cost structure that makes AI services accessible not only to Fortune 500 companies, but also to midsize businesses around the world. Cultural fluency across Western and emerging markets.
Western countries’ heavy investments in frontier models are producing diminishing marginal returns. Each new model requires exponentially more computing to gradually improve the benchmark. However, there is still a large gap between research capabilities and corporate recruitment. Someone has to fill that gap. India is in a position to be a bridge to the world.
a grounded vision
What I find most appealing about Vaishno articulations is their down-to-earth presentation. This isn’t some hype about India matching the US’s computing budget or building the world’s largest model. It is a clear-eyed assessment of where value is created and how India can capture it.
This strategy focuses on diffusion rather than invention. Population-wide developments rather than laboratory breakthroughs. Economic viability, not vanity metrics. These are long-term nation-building priorities.
If well executed, continued investment in computing infrastructure, skills development and public-private partnerships will move India beyond simply rejecting the ‘second-class’ label. This defines what AI leadership in the Fifth Industrial Revolution actually means. It’s not the biggest model, but it has the biggest impact.
The application layer is where AI meets reality. India is ready.
