The Two Indias

Every major American AI company flew to Delhi last week. Anthropic. OpenAI. Google. Microsoft. Meta. The Bharat Mandapam was packed. $200 billion in investment pledges. Sovereign language models trained from scratch. A signing ceremony for something called Pax Silica, a U.S.-led alliance to secure the global AI supply chain.¹

The headlines were enormous. I kept reading. The number underneath was small. Seventeen.

India's five largest IT companies added 17 net employees in the first nine months of this fiscal year. Not seventeen thousand. Seventeen. The year before? 17,764.²

$200 billion in commitments. Seventeen new hires.

Six weeks ago I wrote about a woman named Jennifer. She works the graveyard shift in Bangalore under an American name, solving billing problems for someone in Ohio at 2 AM. I wondered whether the humans in those jobs would get to change with them.

Last week offered one version of an answer. It depends which India you're asking about.

The investment India is real. Microsoft committed $17.5 billion for cloud and AI infrastructure alone. Indian startups unveiled language models trained from scratch in Indian languages using subsidized government compute.¹ None of this is vapor.

But data centers are capital-intensive. They need engineers, not the 1.65 million people who staff call centers. Sovereign language models create jobs for AI researchers. India produces fewer than 500 AI PhDs a year.³

At the summit, Vinod Khosla was blunt. "By 2030 there will be no such things as IT services or BPO. Those are gone." Former HCL Technologies CEO Vineet Nayar was even more direct. "If you believe they are going to create employment, you must be dreaming."⁴

Same stage. Same week. Same city as the $200 billion number.

The summit did produce a reskilling framework. It was voluntary and non-binding. Meanwhile, in Hyderabad's Ameerpet neighborhood, training centers that used to teach Java and Microsoft Office now offer nine-month AI courses. The price is $1,360, paid by individual workers out of their own pockets.⁵

Nine months. $1,360. No employer picking up the tab.

India's own policy think tank modeled what happens if the bridge doesn't get built. By 2031, the inaction scenario loses 1.5 million tech jobs and up to 700,000 positions in the call centers where Jennifer works.³

Two numbers from the same week. $200 billion flowing in. Seventeen people hired. One India is building the future of AI. The other India built the industry that AI is replacing. And the gap between them isn't closing.

The summit ended. The CEOs flew home. Somewhere in Hyderabad, someone just paid $1,360 for a nine-month AI course, betting on themselves because nobody at the Bharat Mandapam bet on them. That's the India that built the first tech boom. It might have to build the bridge to the next one, too.

---

References

Footnotes

1. Chaudhuri, R. (2026). "Tech Is the Bright Spot in India-U.S. Relations." Foreign Policy ↩ ↩²

2. Computerworld. (2026). "AI Boom, Hiring Bust: Indian IT Firms Add Just 17 Net Employees in Nine Months." Computerworld ↩

3. NITI Aayog. (2025). "Roadmap for Job Creation in the AI Economy." NITI Aayog ↩ ↩²

4. India AI Impact Summit coverage. Khosla via Fortune India (2026). Nayar via Economic Times (2026). ↩

5. Reuters. (2025). "Meet the AI chatbots replacing India's call-center workers." Reuters ↩