For thirty years, India ran one of the most successful economic models any developing nation has ever constructed, and also one of the most dangerous. It built a $283 billion information technology industry by renting out its brain. It took the world’s most demanding technical universities, produced hundreds of thousands of engineers every year, pointed them westward, and said: here, use this talent, bill these hours, do this work. The West did. And with that talent, it built Google, Microsoft, Adobe, Sun Microsystems, IBM’s research divisions, and half the software stack that runs the modern global economy. The products, the intellectual property, the compounding equity, the category-defining platforms — all of it stayed in California. The salaries and the remittances came back to Bengaluru, Hyderabad, and Pune. India called this a success. It was not wrong, exactly. But it was also not nearly enough. Now the bill has arrived.
The Numbers Are Not a Warning. They Are an Autopsy in Progress.
In the first nine months of 2025, India’s top five IT services firms, TCS, Infosys, Wipro, HCL, and Cognizant, lost over $150 billion in market value. That is not a market correction. That is investors pricing in an existential question. TCS announced 12,000 layoffs in July 2025, the first mass redundancy in its history, citing AI-driven skill mismatches. Hiring volume across Indian IT collapsed by 75%, from 600,000 freshers taken on in FY2022-23 to barely 150,000 in the following cycle. The pyramid that sustained millions of middle-class families, that great wide base of junior engineers who would one day become senior engineers who would one day become managers, is hollowing from the bottom.
On campuses like the Indian Institute of Information Technology, Design and Manufacturing, fewer than 25% of graduating engineers had secured job offers as of early 2026. These are not students at mediocre institutions. These are the children of India’s aspiring middle class, who studied ferociously to get in, took on debt to study, and built their entire life trajectory around a sector that is now being automated at the precise entry point they were trained to occupy.
NITI Aayog’s October 2025 report estimated that in a worst-case scenario, headcount in India’s tech services sector could fall from 7.5 to 8 million in 2023 to 6 million by 2031, while over 60% of formal sector jobs are susceptible to automation by 2030, especially in IT and BPO. A government think tank, in measured policy language, just described the possible elimination of nearly two million technology jobs in eight years.
The Comfortable Trap of the Billable Hour
India’s IT services companies are, structurally, human-rental businesses. A client in New York or Frankfurt needs 200 engineers to maintain a legacy banking system. India provides 200 engineers. The client pays a daily rate. India bills the hours. The model works brilliantly as long as the client cannot find a cheaper alternative and as long as the work requires humans to do it. AI has now attacked both conditions simultaneously.
One AI-powered platform can now do the work of five engineers. The roles most immediately at risk are manual software testing, basic code writing and maintenance, data entry, first-level technical support, and routine compliance documentation. These are precisely the entry-level roles that formed the base of the pyramid, trained the next generation, and gave Indian families their foothold in the economy.
The model carries an additional vulnerability the optimists rarely mention. AI is enabling what analysts call services reshoring. US firms can now automate domestically, bypassing India entirely. The arbitrage that made India indispensable dissolves when no engineer is required at all. The competitive advantage disappears not because India became more expensive, but because the task itself became cheaper than any human, anywhere.
India watched this coming for years. The warnings were issued, the conferences were held, the white papers were commissioned. And then the quarterly targets landed, the clients renewed their contracts, the share prices went up, and the restructuring was deferred for another year. As late as mid-2025, TCS’s annualised AI services revenue stood at $1.8 billion, roughly 5% of quarterly consolidated revenue. For the company that claims to be leading India’s AI transformation, that figure is the most honest measure of how late the pivot actually is.
Three Decades of Building Other People’s Products
Here is what thirty years of the outsourcing model produced in terms of globally dominant technology platforms created in India: almost nothing. WhatsApp was built in California by a Ukrainian immigrant and an American. Zoom was built in California by a Chinese-born engineer. The language models reshaping the world were built in San Francisco, London, and Beijing. India’s contribution to this generation of technology was the talent that built other people’s products, not its own.
This is not an ethnic or cultural limitation. It is an institutional and incentive failure of the first order. The outsourcing model, by design, rewarded execution over invention. A TCS or Wipro project manager was promoted for delivering on time and under budget, not for questioning whether the client’s problem could be solved in a fundamentally different way. The mindset that scaled India’s IT industry, reliable, process-driven, predictable, is precisely the mindset antithetical to the kind of thinking that builds category-defining companies.
Meanwhile, China, with all its political constraints and state-directed distortions, built Alibaba, Tencent, Baidu, ByteDance, and now DeepSeek. South Korea built Samsung. Taiwan built TSMC. When DeepSeek’s R1 model launched in January 2025, outpacing ChatGPT in downloads and sweeping the world, Indian executives asked: why not from Bengaluru? It was a fair question. The answer was uncomfortable: because India spent thirty years building the world’s most sophisticated labour-hire operation and called it a technology industry.
Sovereign Infrastructure: The Foundation That Cannot Wait
The first and most non-negotiable imperative is sovereign AI infrastructure, and India has, belatedly, begun building it. Under the IndiaAI Mission, more than 38,000 high-end GPUs have been made available at subsidised rates of Rs 65 per hour. The national AI platform AIKosh hosts over 7,500 datasets and 273 AI models across 20 sectors. The IndiaAI Mission has selected 12 startups, including Sarvam AI, BharatGen led by IIT Bombay, and Fractal Analytics, to build sovereign foundational models.
Sarvam AI has unveiled Sarvam-30B and Sarvam-105B large language models. Its Sarvam-1 model operates four to six times faster than competing models in Hindi and ten regional languages while running efficiently on mobile phones. This is infrastructure that solves a problem only India has: 1.4 billion people, 22 scheduled languages, hundreds of millions of first-time internet users who will never read a line of English. Any AI company that solves India’s linguistic complexity owns a market larger than the entire European Union. That is not a consolation prize. That is a category.
Semiconductors: The Most Strategic Layer of All
India’s chip market is projected to reach $100 to $110 billion by 2030. Ten semiconductor projects have been approved under the India Semiconductor Mission, committing Rs 1.60 lakh crore of cumulative investment across six states. India already accounts for 20% of the world’s chip design talent. The country that designs the world’s chips but manufactures none of them is leaving the most strategic layer of the technology value chain in other people’s hands. The nation that controls the physical substrate of intelligence controls the most critical chokepoint of the 21st century.
The Research Deficit That Explains Everything
India’s R&D spending fell from 0.8% of GDP in 2008-09 to 0.7% by 2022, among the lowest in the world, below every BRIC peer. A country that invests less in research than Brazil, Russia, and China, while its competitors race to build the next generation of AI, is not competing. It is spectating. The IITs produce world-class engineers whom the world’s best companies immediately recruit away to California and London. India must make it more attractive to build here than to leave. That means research funding, equity incentives for founders, procurement policy that favours homegrown platforms, and the regulatory agility to let Indian startups experiment without being crushed by compliance burdens designed for a different era.
India must also reorient its vast engineering education system away from producing interchangeable coders and towards producing problem-solvers, product thinkers, and domain specialists who can direct AI rather than compete with it. The roles that will survive and flourish are in AI prompt engineering, machine learning operations, AI governance, and the bridging work between raw AI capability and practical business application. None of these were on any engineering syllabus five years ago. Most still are not. Every year India produces graduates trained for roles that AI is eliminating faster than the curricula can be updated.
The Asymmetric Advantage No Other Country Can Buy
There is one thing India possesses that no other country can replicate or purchase, and it has nothing to do with cheap labour or government policy. It is the extraordinary complexity of the Indian context itself.
India is a country of 1.4 billion people, 22 languages, 28 states with distinct governance systems, 600 million farmers, 500 million smartphone users, a functional democracy generating oceanic volumes of structured civic data through Aadhaar, UPI, and DigiYatra, a public health system serving a population larger than Africa, and a financial inclusion story that is the envy of every central bank in the world. Any AI system that can operate effectively in this environment, at this scale, in these languages, under these constraints, for these users, is automatically a system that works everywhere else on earth. The complexity of India is not a burden. It is a training ground.
According to a Google and Inc42 report, India’s AI market could become a $126 billion opportunity by 2030, with a potential GDP impact of $1.7 trillion by 2035. Those numbers will not materialise through the old model of renting talent to foreign companies that harvest the upside. They will materialise only if India builds the products, owns the platforms, files the patents, and captures the equity.
The Window Is Narrow. The Consequences Are Permanent.
India has perhaps five years, possibly fewer, to make the transition from technology supplier to technology sovereign before the current disruption permanently restructures the global AI landscape around the companies and countries that moved first. The outsourcing model will not collapse overnight. There is genuine durability in the complex systems integration work that India’s IT giants do well, and Gartner analysts note that traditional IT services companies will play a pivotal role in enterprise AI adoption by leveraging their client relationships and domain expertise. But that work will shrink in volume and compress in margin with every passing year. It will not sustain 5.67 million technology workers. It will not rebuild the hiring pyramid. It will not generate the tax revenue, the middle-class consumption, or the national confidence that India needs to become the developed economy it aspires to be by 2047.
The generation of engineers now graduating into a market where fewer than one in four will find a job in their chosen field deserves an answer better than reskill. They deserve a country that built something worth working in. That country is not yet India. It could be. The backyard is on fire. The question is whether India builds a new house or just watches.
Subscribe Deshwale on YouTube
