India is hosting its AI Impact Summit in New Delhi this week. The Wall Street Journal described the country as a "data-rich, capital-poor giant" trying to position itself at the center of the AI era. Seven stories appeared on Techmeme the same day. Together, they answer a question the Summit would rather not ask: is India building AI, or buying it?
The Market
Sam Altman told TechCrunch that India has 100 million weekly active ChatGPT users — the app's largest market outside the United States. One hundred million people in India open ChatGPT every week. They are not using an Indian model. They are not running Indian infrastructure. They are customers of a San Francisco company that is targeting $600 billion in total compensation.
On the same day, Anthropic announced it was opening a Bengaluru office, its second in Asia, saying India revenue had grown. The IndiaAI Mission CEO said India would seek "immense collaboration" with international AI companies. The collaboration is not symmetrical. American AI companies are opening offices in India to hire engineers and capture a market of 1.4 billion people. India is welcoming them because it doesn't yet have an alternative.
The Collapse
Also on February 16, the Financial Times published a look at the fall of Byju's — once India's most valuable startup at $22 billion, now mired in US court proceedings, board battles, and allegations of hiding $533 million from lenders.
-
2022Byju's valued at $22 billion — India's most valuable startup, the poster child of Indian tech.
- Oct 2022 Byju's plans to lay off 12,000 people.
- Sep 2023 Lenders allege Byju's hid $533M in a hedge fund.
- Feb 2024 US unit files for Chapter 11 bankruptcy.
-
NOV 2025A US bankruptcy judge orders Byju's to return funds.
Byju's was edtech. The AI Summit is about AI. But the pattern is the same: international capital floods in, a narrative forms ("India's tech moment"), valuations inflate, and when the capital recedes, what's left is a market — not a producer.
The Infrastructure
The Indian startups that are working in AI tell you what India's actual position is. They are building infrastructure, not models.
Bengaluru-based C2i Semiconductors is building a "grid-to-GPU" power system for AI data centers. Neysa, an AI cloud startup, plans to raise up to $600 million in equity and debt to build compute capacity. Fractal, India's first AI unicorn, does data analytics for enterprises.
Chips. Cloud. Analytics. The three Indian AI startups covered on the same day as the Summit are all building the layers beneath the models — the infrastructure that American and Chinese AI companies will run on. This isn't failure. India's IT services industry built a $250 billion business doing the same thing for the previous era of computing. But it is a specific kind of success: the workshop, not the lab.
The Summit
Data-rich, capital-poor. The Wall Street Journal's phrase is precise. India has the data — 1.4 billion people generating the training signal that AI models consume. It has the talent — Bengaluru produces the engineers that American AI labs hire. And it has the market — 100 million weekly ChatGPT users who are, for now, customers of someone else's technology.
What it does not have is the capital to fund a $100 million training run, or the compute to host one, or the model that results from it. The AI Summit is an attempt to change this. The seven stories from the same day suggest the gap is structural, not temporary.
India's hundred million users are a hundred million data points proving someone else's product-market fit.
The last time India had a tech summit moment, the poster child was Byju's. It was valued at $22 billion. It is now in a US courtroom. The next poster child will be an AI company. The question is whether it will be Indian or just in India.