Indian IT Revenue Is Rising and Hiring Isn't Coming Back. Here's Why

Short answer: Indian IT’s revenue and its headcount have come apart. TCS reported Rs 72,275 crore in Q1 FY27 (April-June 2026), up 13.9% year on year — and over the same twelve months its workforce went from 613,069 to 593,798: 19,271 fewer people. That did not happen through mass firing. TCS says the layoff phase is over, and it added 9,279 people that same quarter. The mechanism is non-replacement. The cause is not AI, it is the business model: for thirty years Indian IT sold effort per person per hour, and clients now buy outcomes. Which makes the standard advice — “the market is bad, just wait it out” — actively wrong. The old model didn’t break. It got replaced, and the replacement runs on fewer people.

Two numbers on one press release

They were never printed side by side, and side by side is the only way they mean anything.

TCS, twelve months to 30 June 2026
Revenue (Q1 FY27)	Rs 72,275 crore, +13.9% YoY
Order book	USD 9.5 billion
Workforce, 30 June 2025	613,069
Workforce, 30 June 2026	593,798
Net change	-19,271 people

More money. Fewer people. Same company, same document.

First, let’s kill the version going around

The internet’s account of this quarter is that the big four bled people. That is simply false, and repeating it makes the real problem harder to see.

Company	Q1 FY27 net headcount change
TCS	+9,279
Wipro	+888
Infosys	-532
HCLTech	-3,292
Combined	+6,343

Across a workforce of 13,88,793. That is not hiring — 6,343 on that base is a rounding error. But it is not a bloodbath either.

So if nobody was fired at scale and the headcount didn’t collapse, why does it feel like the door shut?

Because of the seat that emptied and was never filled again. That is what non-replacement looks like from inside an office, and it never generates a headline.

The pyramid was the product

Here is the part that explains everything else.

Indian IT services was built as a pyramid: a wide, inexpensive base of junior engineers under a thin senior layer. That shape was not an accident of hiring — it was the product, because clients bought effort measured per person per hour. More work meant more people, so revenue growth and headcount growth were the same movement.

Then the client changed the question. Not “how many people will you put on this” but “by when, and for how much”.

The moment the unit of sale stops being a person-hour, the pyramid stops being an asset and starts being a cost. The rope between revenue and jobs snaps — and a snapped rope does not recover. It has to be tied again, from the other end.

But isn’t this just the usual downturn?

The strongest version of the counter-argument deserves stating properly, because half of it is right.

The 2022-23 bench bubble really is deflating. Hiring really did freeze after the dot-com crash and again in 2008-09, and it really did come back bigger both times. Attrition is around 13%. Projections still put growth at 12-15%.

Three things that argument cannot absorb:

1. Last time the pyramid was rebuilt. This time it is being replaced. AI is absorbing precisely the routine coding, testing, maintenance and documentation work that the wide base existed to sell.

2. Hiring didn’t fall so much as move. Naukri JobSpeak’s June 2026 data: overall IT hiring -3% YoY, AI hiring within IT +16%, AI/ML hiring +25% across 14 industries, led by insurance and consumer goods. Info Edge’s Hitesh Oberoi has flagged the shift toward senior and specialised talent. A business cycle reduces demand. A business cycle does not relocate demand into insurance.

3. A revenue line exists now that didn’t in 2008. Infosys reported AI-related revenue at 8.2% of total in Q1 FY27 — revenue that does not scale with headcount.

And the demand story kills the recession framing outright: TCS’s USD 9.5 billion order book and HCLTech’s highest-ever Q1 net-new booking at USD 2,407 million — booked with 3,292 fewer people.

So is AI eating IT jobs?

No, and the lazy version of that claim is doing real damage to people making career decisions.

Record bookings. TCS onboarding roughly 14,000 campus graduates in the quarter. AI roles up 16-25%. This is a re-sort, not an annihilation.

What has genuinely changed is the entry door. TCS hired 44,000 freshers in FY26; in FY27 it had made 25,000 offers as of April 2026, with the CEO saying the number may rise on better demand visibility and the HR head pointing at roughly 40,000. TCS’s own stated reason is a changing delivery model, not AI.

The company has not announced a cut to its target. What it has done is stop announcing the number a year in advance. The campus door’s width is now set quarterly. It used to be a number; it is now a “depends”.

What to actually do

The advice to wait is the one thing that is now measurably wrong. Waiting used to be caution. It is now a price, paid in months.

1. Stop searching “IT job”. Search AI and ML roles outside IT services — insurance and consumer goods are where the growth actually showed up.

2. Target a GCC rather than a services firm. Different economics, different reason for your seat to exist.

3. If you are in a Tier-2 or Tier-3 city, the free capacity is real and mostly empty. IndiaAI Mission has approved 58 AI Centres of Excellence, 543 Data & AI Labs in ITIs and polytechnics, and 686 fellowships across 178 institutions. The deliberately non-metro design here deserves credit. The scale does not: NASSCOM expects AI job demand to cross 1 million by 2026 against roughly 16% of Indian IT professionals being AI-skilled. The design is right and it is short by a full zero.

4. Leave the automatable middle on purpose. L1 support, manual QA, legacy application maintenance, routine testing and documentation. If that is your day, that is your notice period. Internal transfer is the cheapest move you will ever make.

Note what is missing from that list: “upskill”. It is an empty word that tells you to do something without telling you what, where, or how you would know it worked.

The part for parents paying fees

Placement used to arrive with the degree. It now arrives with whatever you added on top of it. That is a harder deal, and pretending otherwise helps nobody.

What broke is not the market. It is the rope that tied revenue to jobs — and a broken rope doesn’t recover. It gets tied again, from the other end.