India has cracked OpenAI’s list of the world’s most advanced artificial intelligence markets, ranking in the top five globally for what the company calls “thinking capability usage per person,” a metric measured through reasoning tokens consumed by ChatGPT Plus users. The numbers are impressive. The reality, however, is considerably more complicated.
In its first “Capability Gap” report for India, OpenAI found a country of fierce contrasts: a sophisticated urban elite pushing AI to its limits in coding, data analysis and complex reasoning, and a vast hinterland that remains largely untouched. The top ten cities account for roughly 50 per cent of all AI users in India, despite housing less than 10 per cent of its population. Delhi NCR leads the country in ChatGPT penetration. Bengaluru, Hyderabad and Chennai follow close behind.
The gap is not merely statistical. It is staggering. Data analysis usage is up to 30 times higher in leading cities than in lagging regions. Coding usage runs four times higher. For developer tools such as Codex, the disparity stretches to ninefold. When OpenAI launched Codex in February 2026, usage in India grew fourfold within two weeks, a figure that flatters the aggregate but masks where that growth is actually happening.
Beyond the metros, a different, quieter story is emerging. In states such as Assam, Odisha, Manipur, Tripura and Chhattisgarh, AI is finding its footing in classrooms rather than coding suites. In Assam, 22 per cent of ChatGPT messages are linked to learning, roughly 20 per cent above the national average. In Jammu & Kashmir, Punjab, Chandigarh, Himachal Pradesh and Kerala, healthcare queries dominate: in Jammu & Kashmir alone, nearly one in ten messages relates to health, some 32 per cent above the national average. It is not the use case the venture capital world writes breathless memos about, but it may prove more durable.
Oliver Jay, managing director for international markets at OpenAI, was candid about where the effort must now go. “The central question now is how quickly the benefits of AI can extend beyond early adopters and leading cities to the wider population,” he said, adding that expanding access and building skills would be central to closing the divide. To that end, OpenAI is deepening partnerships with Tata Consultancy Services and Razorpay, and investing in developer programmes, university tie-ups and localised use cases built around Indian languages and contexts.
The infrastructure, affordability and language support needed to carry AI into India’s smaller cities and rural districts remain formidable challenges. A country that can sit comfortably in the global top five for AI sophistication while leaving the bulk of its population behind is not, in any meaningful sense, an AI nation yet. It is, at best, a collection of AI cities surrounded by an analogue country. Closing that gap is the harder, slower, more important task and the one on which India’s digital future will ultimately be judged.
