Self-proclaimed AI experts urged GOI to act decisively to avert a crisis: “We are not high enough on the AI value chain, and we need to act quickly.” Apparently, while Silicon Valley and China are building the future, India is sitting in the basement, labelling images, cleaning datasets and teaching machines the difference between a cow and a buffalo.The remedy, predictably, is that India must “move up the value chain”. This phrase is the typical one CEOs use at strategy meetings and make everyone feels more intelligent. But perhaps India should pause before climbing a ladder whose top rung may disappear tomorrow. What if the data layer isn’t the bottom of the AI value chain at all? What if it is the foundation?We have an unfortunate habit of confusing intellectual glamour with economic value. For decades, India Inc was told services were inferior because somebody else owned the intellectual property. Then products became superior to services. Then platforms became superior to products. Now foundation models are the new Holy Grail.Shifting ladderEvery generation invents a new rung and our strategy promptly starts climbing it. The danger is that by the time we reach the top, the ladder has been replaced. Look at the AI gold rush. Everyone wants to build a foundation or sovereign model because it sounds magnificent at Davos. Nobody wants to admit that training another giant model requires staggering computing power, capital, energy, semiconductor access and top AI talent. The arithmetic is the problem.We don’t need to beat America or China at building frontier models any more than we need to beat Saudi Arabia at producing oil. Competitive strategy requires the discipline of not competing where someone else has already spent ten times your money and have better R&D. Instead, we should ask a Nouveau rich question: Where can we actually make money? The answer may lie in the supposedly humble layers that AI evangelists are eager to escape.Generic data annotation is low value. Proprietary, contextual, multilingual, continuously updated and industry-specific data is different. India has something Silicon Valley cannot manufacture with a few billion dollars: India itself. Our linguistic chaos is a strategic asset: hundreds of languages and dialects, code-switching, accents, slang, regional idioms and wildly uneven digital behaviour.Diversity, our strengthThen there is our economic diversity. A billionaire in Mumbai, a farmer in Punjab, a kirana owner in Karnataka and a student in Kota do not inhabit the same India… or even the same internet. That complexity is precisely what AI needs to understand in the digital sentience era.The world can build another English-speaking chatbot. India can build AI that understands the India those chatbots don’t understand. That is not the bottom of the value chain. The real stupidity would be becoming the world’s cheapest AI data-labelling shop. The opportunity is to become its most sophisticated AI data economy.India should build proprietary datasets around healthcare, agriculture, manufacturing, logistics, courts, education and public services. Turn messy real-world information into structured intelligence businesses can use.The winning Indian AI company may not build “IndiaGPT 17.0”. It may be the one that knows why a tractor fails in rural Maharashtra, why a manufacturer loses production hours in Gujarat, or why a court case remains unresolved for seven years. The futur belongs to those who understand the smallest problem deeply enough to solve it profitably.Corporate vanityThe obsession with “moving up the value chain” can become corporate vanity, not strategy. Every board wants intellectual property. Every CEO wants an AI platform. Every startup wants a proprietary model. Nobody wants to admit that integrating someone else’s model into a customer’s workflow may produce more cash, which is the major reason for the abysmal failure of corporate Gen-AI.In the past few decades we built a formidable technology-services industry and then developed an inferiority complex because Americans owned the software products. Now AI gives us another chance to repeat the mistake: abandon what we are exceptionally good at because the fashionable thing looks more respectable.India’s IT industry understands enterprise customers, complex workflows, regulation, outsourcing, integration and scale. That knowledge is not obsolete because ChatGPT exists. It may become extraordinarily valuable because ChatGPT exists. The opportunity is to turn services into AI-powered IP.If a foreign model is cheaper and better, use it. If domestic capability is strategically critical, build it. AI is moving so quickly that today’s technological summit may become tomorrow’s commodity basement.Enterprises owning a unique dataset, trusted customer relationships and a deeply embedded workflow may quietly collect subscription revenue. The model is not necessarily the moat. Context is. And India has context in industrial quantities. Yes, move up the AI value chain but stop assuming that “up” always means more sophisticated technology. “Up” can mean closer to the customer, or owning the data, understanding the problem, or making money.As the strategy gurus would say, the smartest strategic move is refusing to climb the ladder everyone else is climbing. India doesn’t need to win the AI Olympiad. It needs to make sure it isn’t paying for the stadium while someone else takes the medals.While India Inc. is trying to escape the supposedly low-value data economy, the world is discovering that without good data, the most magnificent AI model is merely a very expensive hallucination toy. Perhaps India’s AI strategy needs one radical idea: Stop trying so hard to look like Silicon Valley and become very, very good at being India. That might turn out to be the most valuable AI advantage nobody else can copy.The writer is a Fortune-500 advisor, start-up investor and co-founder of the non-profit Medici Institute for InnovationPublished on August 21, 2026