Sam Altman is selling a dream again. This time, it’s shaped like the subcontinent. During a recent victory lap, the OpenAI CEO dropped a figure that should make any venture capitalist salivate: 100 million weekly active users in India alone.

That’s a staggering number. It’s also a convenient one.

Let’s be clear about what’s happening here. When Silicon Valley growth starts to plateau and the "AI will save the world" narrative begins to feel a little thin, you look for a massive, tech-hungry population to pad the stats. India is the ultimate validation machine. If you can’t get the suburbanites in Ohio to stop using ChatGPT for low-effort grocery lists, you pivot to the millions of developers and students in Bangalore and Hyderabad who are actually trying to build something.

But 100 million users isn't just a metric. It’s a massive, resource-draining liability.

OpenAI is currently burning through cash like a bonfire in a hurricane. Every time a college kid in Mumbai asks ChatGPT to summarize a textbook or a freelance coder in Pune uses it to debug a React component, a bank of H100 GPUs in a data center somewhere starts screaming. And most of these 100 million users aren’t paying. They’re on the free tier. They’re the "loss" in "loss leader."

The friction here isn't just about the server bill. It’s about the product itself. The $20-a-month price tag for ChatGPT Plus is a rounding error for a software engineer in San Francisco. In India, where the median monthly income for many young professionals hovers around $400 or $500, that subscription isn't a "no-brainer." It’s a luxury. It’s a choice between a premium AI assistant and a week’s worth of groceries.

So, OpenAI is stuck in a classic tech trap. They have captured the attention of a nation, but they haven't figured out how to extract the value. Instead, they’re providing a massive, subsidized utility for a demographic that may never actually convert to paid subscribers. It’s the Spotify problem, but with significantly higher electricity costs and a lot more hallucinations.

There’s also the linguistic wall. Silicon Valley likes to pretend that English is the only language that matters for "global" products. It’s easier that way. But India lives in the gaps between dozens of major languages and hundreds of dialects. GPT-4o might be clever in English, but it’s often a stumbling, incoherent mess when it tries to navigate the nuances of Kannada or Marathi. Altman keeps talking about "broadening access," but he’s really just exporting a Western-centric logic engine and hoping the locals find a way to make it fit.

Meanwhile, local competitors aren't just sitting around. Ola’s Krutrim and startups like Sarvam are trying to build models that actually understand the cultural and linguistic context of the region. They don’t have OpenAI’s multi-billion-dollar war chest, but they don't have its baggage, either. They aren't trying to be a "god-like" intelligence for the entire planet; they’re just trying to work.

OpenAI’s play in India feels less like a strategic expansion and more like a land grab. If you control the interface through which 100 million people interact with the internet, you control the data. You control the habits. You control the future of how work gets done in one of the world’s fastest-growing economies.

But data doesn't pay the power bill. Altman can brag about the 100 million all he wants, but those users are currently a massive drain on a company that still hasn't proven it can be a sustainable business without Microsoft’s hardware handouts. It’s a high-stakes game of chicken. OpenAI is betting that by the time they need to turn a profit, they’ll have made themselves so indispensable that the world—and India—won't have a choice but to pay whatever price they set.

It’s a bold bet. It’s also a desperate one.

How many of those 100 million users will still be there when the "free" sign finally comes down and the real bill arrives?