India is pouring its energy into catching up with the US and China on building foundational Artificial Intelligence (AI) models. Government officials are discussing billion-dollar investments in data centres and computing infrastructure. Academics are calling for sovereign large language models (LLMs). Industry groups are lobbying for more public-private collaborations in model training. But all this focus is on yesterday’s problem.
The world doesn’t need more foundational models. It already has dozens. Open-source alternatives like DeepSeek, LLaMA, Qwen, and Mistral are freely available — and improving at a breakneck pace. DeepSeek recently released a model that rivals GPT-4 in reasoning benchmarks. Qwen, developed by Alibaba, has become a top-tier model for multilingual tasks. India doesn’t need to build its own from scratch. It can take these models and run with them. The real opportunity isn’t in recreating what already exists. It’s in doing what India has always done best: Building on top of what’s already there.
That’s exactly how India became a superpower in IT services. It didn’t invent the microchip, but it built billion-dollar firms that helped the world use microchips. It didn’t create operating systems, but it became the global centre for enterprise software development. India didn’t pioneer cloud computing, but it gave rise to SaaS (software as a service) giants like Zoho and Freshworks. It didn’t create banking infrastructure, but it built UPI — the world’s most advanced payments platform.
Now, a similar opportunity is emerging in AI. The next frontier is not more models— it’s platforms. In my book, From Incremental to Exponential, I explained how platforms became Silicon Valley’s biggest strategic advantage. They enabled trillion-dollar companies by allowing others to build on top of their infrastructure— creating digital ecosystems that attracted developers, locked in users, and fuelled unstoppable network effects. India has the opportunity to replicate this success in the AI era. A good platform provides tools, reach, and scale. The best platforms become the backbone of entire industries. And the next wave of platforms will be powered by AI.
And then, there are AI agents. These are autonomous systems that can take a goal — “analyse Infosys’ stock” or “plan a three-city vacation”— and break it down into steps, execute those steps, and deliver results. They browse the web, write code, summarise documents, and interact with other software. When built well, they behave like junior analysts or digital interns. When built brilliantly, they’ll become the infrastructure for future work — operating systems for decision-making, strategy, and execution.
Some early AI agents, like China’s Manus, are already demonstrating what’s possible. But the pace of change is astonishing. Within weeks of making headlines, both Manus and DeepSeek are already being eclipsed by newer entrants and even more powerful tools. The AI space is moving so fast that even OpenAI is scrambling to maintain its lead. Manus doesn’t own a foundational model. It wraps existing models like Claude 3.5 and Qwen and combines them with smart planning, orchestration, and tool integration. It integrates nearly 30 tools — from file readers to web browsers — and delivers outcomes, not just responses. It’s been called a “wrapper”, but it’s really a new kind of platform: One that transforms AI from passive assistant to active problem-solver.
This is where India has historically excelled. Its strengths lie in building on top of existing systems — designing, deploying, and scaling platforms. Think of what Infosys, TCS, and Wipro did for IT. What Zoho and Freshworks did for SaaS, Adar Poonawalla’s Serum Institute did for vaccines — producing billions of doses not by inventing new drugs, but by mastering large-scale production and distribution. India didn’t pioneer drug discovery, but it became the ‘pharmacy of the world’ by perfecting the production of affordable generics. Sun Pharma and Cipla focused on manufacturing and distribution at scale, while Biocon, under Kiran Mazumdar-Shaw, made insulin and biosimilars affordable by optimising existing innovations.
India has the potential to do the same in AI. It could build AI agents tailored to health care, agriculture, logistics, education, and more. It could develop orchestrators for businesses, and public platforms that democratise AI access for small enterprises and underserved communities — not just in India, but globally. Imagine a public AI stack, modelled after UPI, where verified agents can access data, execute tasks, and collaborate securely across domains.
Instead of trying to catch up on yesterday’s models, India can leapfrog into tomorrow’s platforms — and build the same types of trillion-dollar monopolies that Silicon Valley has. Remember that Apple and Microsoft didn’t build any foundational systems, they simply built great solutions on top of what already existed — and they are amongst the most valuable companies in the world.
Moreover, the AI landscape is evolving so rapidly that today’s leaders can be eclipsed in months. LLMs will soon be yesterday’s news and AI agents may dominate the next wave — but something even more disruptive will follow. Innovation is accelerating at an unprecedented pace, and those who build adaptable platforms — rather than static models — will be best positioned to ride these exponential curves. By focusing on building the platforms and applications that others rely on, India could not only leapfrog the current wave but rise above the turbulence of constant disruption.
Vivek Wadhwa is CEO, Vionix Biosciences. The views expressed are personal
