Agentic AI is striking at what may be the most iconic sector of the Indian economy — IT outsourcing — along a transmission chain that runs from cost substitution to layoffs, from layoffs to shrinking housing demand in the tech cities, and from there to newly exposed credit risk on bank balance sheets. This is not a swing of the market cycle; it is a structural substitution of an industrial paradigm. For three decades the world has held up India's IT industry as the textbook case of outsourcing — yet its core business model, labor arbitrage, is now being undermined at the foundation by AI's capacity to generate code. The knock-on effects have already broken through the IT sector itself and spread into the property markets and bank credit systems of tech cities such as Hyderabad, Bengaluru and Pune.
The Paradigm Shock to IT Outsourcing — AI as Industrial Substitute
On May 1, 2026, Nanya Yanjiu Tongxun (South Asia Research Newsletter, a Chinese-language commentary account covering South Asian affairs) republished an in-depth article that had appeared in India's The Economic Times on April 28. The article exposed a structural change already underway: agentic AI is restructuring the global industrial landscape, and India's pillar IT-outsourcing industry is facing a structural shock.
“Agentic AI is restructuring the global industrial landscape, and India's pillar IT-outsourcing industry is facing a structural shock. Over the past three years, India's top five outsourcing firms have cut a combined net total of 85,000 jobs. Generative-AI models such as Claude and Mythos can complete complex work — vulnerability remediation, code writing — at extremely low cost in both time and money, delivering a disruptive blow to the labor-intensive development model.”
The article stressed that this is more than short-term turbulence. In 2025, US tariff policy led overseas clients to postpone technology capital expenditure; in 2026, the US–Iran geopolitical conflict amplified demand volatility further. But beneath these cyclical factors, the shock that agentic AI delivers to the IT-outsourcing business model is structural: in the past, Indian IT companies sold “person-hours” — the hourly price of an engineer — and now AI can complete part of that work for a fraction of the cost. Revenues at several outsourcing firms, including Infosys and HCL Technologies, have come in far below expectations; in March 2026, Oracle's Indian subsidiary cut roughly 12,000 jobs.
The Transmission Chain — From White-Collar Layoffs to a Property Crisis in the Tech Cities
The article laid out the cascading transmission mechanism of the AI shock, and its value reaches far beyond an analysis of the IT sector alone:
IT layoffs → the disappearance of the main home-buying cohort in the tech cities → sluggish sales of commercial housing → slower inventory digestion → rising unsold apartment stock → mounting pressure on investment buyers.
In India's core tech cities — Hyderabad, Pune, Bengaluru, the urban cluster often called “India's Silicon Valley” — the dominant home-buying demographic is precisely the IT white-collar workforce. As this group confronts the occupational risk of AI substitution, real estate — the industry most tightly bound to economic confidence — is the first to feel the chill.
First link (the IT sector): AI substitutes “repetitive high-skill labor” such as code writing and vulnerability remediation, eroding the labor-arbitrage foundation of the outsourcing business model.
Second link (tech-city real estate): the white-collar cohort's income expectations and job stability decline; home-buying demand shrinks and unsold apartment inventory climbs.
Third link (the banking system): mortgages and credit-card loans to IT workers were once classified as low-risk credit; that risk is now expanding steadily.
The Banks' Early Warning — Reshaping the Logic of Credit
The article's most important insight comes from the reaction of India's banking system. The economic-research team of state-owned Canara Bank, headquartered in Bengaluru, has already proposed three risk-management optimizations:
- At the individual-borrower level: for individual borrowers whose jobs are vulnerable to AI substitution, mandatorily reduce the loan-to-value (LTV) ratio from the standard 80% to 60%.
- At the corporate-credit level: bring intellectual property and proprietary data into the scope of supplementary collateral, ensuring that banks hold priority claims over borrowers' AI assets.
- At the redistribution level: for enterprises carrying out AI-driven business restructuring, channel a portion of the labor-cost savings into debt-service reserve accounts.
The common signal in all three measures is this: banks have begun folding “AI substitutability” into their credit-assessment framework. For borrowers in occupations at high risk of AI substitution, banks no longer extend credit on the basis of current income; they are starting to project future income stability instead. This is a paradigm shift from static credit assessment to dynamic assessment of occupational substitution risk.
Canara Bank is no small institution — it is one of India's state-owned banks, and shifts in its risk-management logic tend to be a leading indicator of regulatory direction across the national banking system. Once a state-owned bank begins classifying IT workers' mortgages as a “rising risk,” the property markets of entire tech cities face tighter credit conditions and weaker purchasing-power support.
Even more notable is the trend the article flags at the end: the banking sector is reshaping the logic of credit allocation — “jobs such as healthcare workers and skilled trades show stronger employment resilience and better wage growth; going forward, these groups will enjoy better housing options and more lenient credit-approval standards.” In other words, banks are moving from lending on the basis of job title and salary to lending on the basis of an occupation's resistance to AI substitution — and the way a worker is evaluated by a bank is changing fundamentally.
The Specificity of India's Predicament — The Hazlitt Trap Layered on Top of the AI Shock
The particular way the AI shock lands on India deserves attention. India is a country facing two structural problems at once:
First, the productivity predicament of the “Hazlitt Trap.” The Indian economy has long suffered a structural bias in which policy prioritizes redistribution over production. IT outsourcing is one of the few genuinely high-productivity sectors through which India participates in the global division of labor — and it is precisely this sector that AI is now disrupting at its foundation.
Second, the social risk created by the absence of redistribution mechanisms. India's social-safety net is weak, and workers displaced by AI lack the European-style retraining programs and social protections. The fact that Noida workers staged violent protests over inflation on the same day Nanya Yanjiu Tongxun carried this report is no coincidence — it is the same tide of “cost pressure plus job insecurity” making landfall in different Indian industries at the same moment.
A Reusable Analytical Framework — The “Three-Stage Transmission Model” of AI Shocks
The analytical logic of this essay can be abstracted into a reusable framework for assessing AI shocks to other industries and countries:
Stage one (direct substitution): which core functions of which occupations does AI directly substitute? For IT outsourcing — code writing and vulnerability remediation; for call centers — customer-service conversations; for translation — language conversion.
Stage two (consumption transmission): as the purchasing power of workers in the substituted industry declines, how does that affect the consumer markets tightly bound to them? For India — tech-city real estate; for Silicon Valley in the US — perhaps high-end services and school-district housing.
Stage three (financial feedback): how do banks and the credit system adjust lending standards for occupations at high substitution risk — and how does that in turn accelerate or slow the transmission of stages one and two?
The core judgment of this model is that the endpoint of the AI shock is not the unemployment figure, but the reconstruction of the labor–consumption–credit triangle. Once banks begin lending on the basis of an occupation's probability of AI substitution rather than current income, a self-reinforcing positive-feedback loop can start — tighter credit → weaker consumption → shrinking industry demand → more layoffs → still tighter credit.