India, July 4 -- To understand the real friction inside India's AI ambitions today, you only need to look at two numbers. 4,096: That is the number of Nvidia H100 computing chips the Government has allocated to Bengaluru-based Sarvam AI. These chips have become the world's most sought-after AI computing infrastructure. Costing up to Rs.30 lakh apiece and manufactured by a single company in Taiwan, they have become indispensable for training frontier AI models. 1%: This is the estimated stake the government is expected to acquire in the company if the subsidy it extended converts into equity when Sarvam completes its ongoing funding round. For the first time since the internet entered our lives, the Indian state is preparing to become more than a regulator of technology. If the proposed conversion takes place, it will become a shareholder in a private AI company. Vivek Raghavan, co-founder of Sarvam AI, declined to comment for this piece. The world over, AI is increasingly being treated as a strategic asset. The United States has tightened access, Europe is pouring public money to build its own AI ecosystem and China has spent years backing domestic technology champions. In that backdrop, New Delhi's support for Sarvam is not just a financial transaction, it reflects a larger belief that India cannot afford to remain dependent on foreign companies for a technology that could shape everything from public services to national security. Amit Ranjan, co-founder of DigiLocker, has no quarrel with the government's objective. In fact, he sees nothing particularly unusual about the proposed equity structure. He points out that the Indian government is already an indirect shareholder in startups through entities such as SIDBI. So, a stake such as this, is simply a more direct version of an existing idea. Ranjan's discomfort lies elsewhere. Think about the technology you use. Few people choose an email service or a video-conferencing platform for nationalistic reasons. They choose it because it works better than the alternatives. So, Ranjan asks, why will people apply different standards to AI? If Sarvam wants to compete with the world's best, it must ultimately be judged by the quality of its models, not by the passport it carries. Harish Mehta, one of the founders of NASSCOM and Founder of Onward Technologies, is, on the other hand, less interested in how Sarvam positions itself than in whether governments should be trying to build national AI champions at all. "India", he says, "has already lost the race to build Large Language Models (LLMs). Not for lack of talent, but because frontier AI has become a race defined by extraordinary concentrations of capital, compute and research capability-advantages that are already heavily clustered elsewhere. Mehta says governments are accustomed to building assets that endure. Roads, airports, power plants and other such become more valuable with use. Frontier tech works differently. An LLM is not a finished asset, but a living system that has to be improved continuously. Its real value does not sit on servers. It resides in the researchers who refine it, the engineers who optimize it, and the ideas they generate. Those people move. When they do, a significant part of the advantage moves with them. "People move from one company to another," Mehta argues. "So how can you have one LLM and expect people to stay on it?" It is an observation that goes well beyond Sarvam. It questions whether AI can ever be treated like conventional infrastructure. Governments can finance research, they can subsidise computing costs, they can even become shareholders, but they cannot own the curiosity, judgement and ambition that keep a frontier model at the frontier. So, the debate over Sarvam is not really about Sarvam. Nor is it about whether the government eventually owns one per cent of a startup. Those details matter because they reveal a larger change already underway. For nearly three decades, governments largely left software to entrepreneurs and the market. AI is changing that relationship. Around the world, states are beginning to treat intelligence as a strategic capability worthy of public intervention. India is set to join that experiment. Should the state actively back a handful of companies it believes can build frontier capability? Or should it focus on creating the conditions in which many competing companies emerge? Amit Ranjan and Harish Mehta arrive at different answers. Yet they agree on one thing: India needs companies capable of competing with the world's best. Their disagreement is over how to build them. That is why the two numbers at the beginning of this story matter. 4,096 is about capability. 1% is about the role of the state. Together, they ask a question India will keep confronting long after Sarvam's funding round is forgotten: When intelligence becomes a strategic national capability, where should the market end and the state begin?...