For thirty years, Indian IT ran on a formula you could write on a napkin. Revenue equals headcount times billing rate.
More work meant more people. It made Indian IT a mass employer and a reliable middle-class escalator.
In FY26 the napkin stopped working. Indian tech revenue grew 6.1% to cross $315 billion. Headcount grew 2.3%.
The nightmare is not that AI kills Indian IT companies. Revenue is still growing. The nightmare is that the link between revenue and jobs came apart. As of 13 October 2026, that gap is the whole story.
What the numbers actually show
NASSCOM's Strategic Review 2026 confirms the split: sector revenue up 6.1% to about $315 billion in FY26, direct employment up just 2.3%, roughly 135,000 net new jobs.
NASSCOM president Rajesh Nambiar has said the quiet part out loud: five years ago revenue and employment growth moved together. Now they do not.
The market got the memo first. Nifty IT is 2026's worst-performing major sector, though how bad depends on who is counting: down about 24% as of June, around 29% year to date later, closer to 40% from the index peak. Call it a quarter to a third.
What AI is doing to Indian IT companies, specifically
It is eating the bottom of the pyramid.
The pyramid model means a wide base of junior engineers writing and testing code, billed by the hour, under a thin layer of managers.
That is exactly the work AI now does well. Coding assistants like Cursor and GitHub Copilot handle the write-this-function, test-this-module tier that used to absorb hundreds of thousands of graduates a year on the bench. Campus recruitment is the visible casualty.
Here the reporting gets sloppy.
TCS closed March 2026 with 584,519 employees against 607,979 a year earlier, a net fall of 23,460, per company filings. What it announced in July 2025 was different: a restructuring of about 2% of staff, roughly 12,200 roles, mostly at middle and senior levels.
Those measure different things. 12,200 was the announced cut. 23,460 is the net headcount change, which also folds in attrition, retirements and everyone who simply left. TCS's HR head says the fall "cannot be fully attributed" to the restructuring. "TCS cut 23,460 jobs" is a headline, not a fact.
The direction has turned. TCS added 4,258 people in the September 2026 quarter, its third straight quarter of net hiring, and onboarded 10,000 freshers. India's top firms still plan around 80,000 fresher hires this year.
A taxi that drives itself
Indian IT sold hours the way a taxi sells distance. The meter ran, the kilometres added up, the client paid for both.
AI just invented a car that drives itself.
The trips still happen. Nobody pays by the kilometre, and the fleet needs fewer drivers.
That is the linear growth model breaking, without the jargon. Linear growth was the assumption that revenue and headcount climb together in a straight line. In FY26 the line bent.
The other side: record deals, real AI revenue
On their own numbers, the companies are not behaving like victims.
TCS signed $40.7 billion of total contract value in FY26, meaning the full worth of deals booked as revenue over several years. Q4 alone was about $12 billion, and its annualised AI run rate passed $2.6 billion by Q1 FY27.
Infosys won $14.9 billion across 96 large deals in FY26, up from $11.6 billion, with AI at 8.2% of revenue in Q1 FY27 against 5.5% two quarters earlier. HCLTech booked a record $2.4 billion quarter, AI revenue up 62.1% to $171 million.
TCS chairman N Chandrasekaran calls AI enterprise IT's most significant opportunity, not a threat. NASSCOM and brokerages see AI-led services opening an incremental $300 billion to $400 billion market by 2030, against an industry worth about $280 billion.
The bears disagree. Jefferies says AI could shrink application managed services, the business of running a client's existing software for a recurring fee, worth 22% to 45% of revenues at leading firms. Motilal Oswal puts 9% to 12% of IT services revenue at risk within four years.
The catch: signed contracts, missing revenue
Both camps are arguing about a future the accounts have not reached.
Contract values are at records. Turning them into revenue is slow: pilots instead of rollouts, elongated decision cycles, delayed starts.
Infosys did $5,082 million in Q1 FY27, up 2.4% year on year in constant currency. TCS did $7.62 billion, up 2.7%.
Not collapse numbers. Not transformation numbers either.
A sector growing under 3%, hiring a third as fast, sitting on signed contracts that have not become money. Duller and stranger than either story being sold.
What this means for you
If you are an engineering student or a fresher. This is where the damage is real. The old deal, get hired, sit on the bench, learn on the job, is gone, because bench work is what AI does. Firms want proof of work, not a degree tier. Get fluent with the tools doing the automating.
If you already work in IT services. The squeeze is uneven. Routine coding, manual testing, L1 support and basic maintenance compress hardest, the managed-services layer Jefferies flagged. Client context, domain knowledge and architecture do not. Hold that context and you are the moat.
If you invest in or build around the sector. Watch the gap between contract value and revenue conversion. Convert at normal speed and the bear case weakens. Keep signing while revenue stays under 3% and deflation wins. Frontier models like GPT-6 Astra keep raising what one engineer can do.
So, is it a nightmare?
Probably not for the companies. They have the relationships, the deal flow and a big new market to sell into.
It may well be one for the headcount model that made Indian IT a mass employer. Those are two different claims, and most of 2026's coverage has run them together.
Revenue is growing. Jobs are growing a third as fast.
Is AI going to end IT jobs in India?
Did TCS cut 23,460 jobs in FY26?
Why are Indian IT stocks falling in 2026 if revenue is still growing?
Should engineering students in India still target IT services jobs?
