# The AI factory: rewiring India's tech industry

> FT documentary on India's two-sided AI economy: low-paid data annotation and robotics training labour, set against AI's existential threat to the IT services sector.

Published: 2026-06-26
URL: https://daniliants.com/insights/the-ai-factory-rewiring-indias-tech-industry/
Tags: geopolitics, ai-adoption, india-tech, career

---

## Summary

FT documentary on India's two-sided AI economy: a vast, low-paid data-annotation and robotics-training labour force in small towns, set against the existential AI threat to India's ~$330-340bn IT services export sector. The core tension is that India is positioning its 1.5bn-person population as cheap "human-in-the-loop" fuel for foreign AI - which generates jobs today but locks the country into an extractive back-office supply chain rather than AI sovereignty.

## Key Insight

**The data-annotation labour economy (the supply side)**

- India is "probably the second largest AI workforce in the world," with the documentary claiming India is in third place globally in the AI race (a contested view).
- **Object Base** (founded 2019, started as data annotation, pivoted to robotics) collects egocentric/gripper/tele-operation data. A textile factory in Karur, Tamil Nadu has workers wearing GoPros/Meta glasses to capture household and manual tasks: 60 of 200 workers, 6-8 hrs/day, paid ~10,000 rupees/month extra. One worker: ~1,000 rupees for 3 hours of capture.
- Stated demand target: ~500 million hours of robotics-training data per single day (a "6-month volume in one day"). The scale signal matters more than the exact figure.
- **NextWave** (started ~15 yrs ago in Malasamudram) built on the insight that 60% of colleges - and 60% of graduates - are in small towns, mostly first-generation graduates. Strategy: "take the work to where the people are, rather than people travel to where the work is." Annotation framed as a factory floor demanding assembly-line accuracy.

**The structural critique**

- AI "by design concentrates power" - and this annotation work is "a supply chain which is extractive by design." Workers are training the very robots/models that may displace them, with no equity in the upside.
- Risk flagged: in some geographies workers will be made to wear cameras with no extra pay, and the captured data can be used against them.
- ~$1.5tn is being "pumped into" LLMs; "AI is fundamentally a marketing term... a way to unlock vast amounts of capital."

**The IT services threat (the demand side that's eroding)**

- India's IT services exports: ~$330-340bn/yr - larger than any single goods export and a major source of dollar income. Built since the early-90s Bangalore boom on English fluency, premium-institute talent, "under-promise, over-deliver," and price ("$1 job done for 20 cents").
- AI is called "the toughest challenge India's IT services sector has faced" - high-volume, repetitive, error-tolerant work is a "prime candidate for AI substitutability." Risk is full displacement, not just transformation. With automation cited at "60% of the work... I need only 40 people, not 100."
- The incumbents' failure: "immensely profitable... they were not investing in AI," not the vanguard of bringing AI to India. India's structural weakness: low private-sector R&D investment.
- **Global Capability Centers (GCCs)** are the next phase - work done in-house for foreign firms inside India. Example: **Tesco** (66bn pound business) runs architectural/civil/mechanical/electrical design, an IoT platform monitoring 100,000+ fridges across 4,500 stores, 30 yrs of data on ~22m customers for personalization, and a "cost intelligence model" that decomposes supplier price hikes (e.g. breaking bread into yeast/water/flour to challenge a 10% increase). One touchless AI system categorizes store issues into ~300 problem types and auto-dispatches vendors.

**The sovereignty paradox**

- India wants to "build our own LLMs/products," but critics say offering population scale as a "carrot to attract foreign tech companies is not a pathway to... sovereignty." It risks becoming the "use case capital of the world" / a market "for the taking by US tech giants" - echoes of the BPO era.
- Capital flight signal: ~$25bn left Indian markets in the first ~3-4 months of the year, money moving to Taiwan/South Korea which produce chips/hardware relevant to the AI race. India "doesn't have that."
- Counter-view in the film: tech-sector job count could rise (more human-in-the-loop demand); India could export AI-trained talent. "The threat to humanity is not AI... it's the inability to learn." Goal frame: developed-country status by 2047.