Indian IT and AI: Layoffs, Hiring and the Reskilling Gap

Note: Statistics and figures reflect data available as of October 2026. Verify for latest figures.

In the second week of July, two Indian IT results landed within days of each other. TCS had added more than 9,000 people in a quarter. HCLTech had shed more than 3,000. Read side by side, they raise the question every engineer I know is asking: are we hiring or firing?

The honest answer is both, and that is the story of Indian IT right now. After two decades in Indian IT delivery, I have stopped reading headcount as a single number. The aggregate barely moves. Underneath it, the mix of who gets hired and who gets benched is changing fast.

This post puts the latest results, layoffs and hiring data into one picture, then lays out what I would do at each career stage.

The Q1 FY27 Scorecard

Here is what the four largest firms reported for Q1 FY27 (Apr–Jun 2026), taken from their July results and the coverage around them.

Company	Headcount (30 Jun 2026)	Net change in quarter	Freshers onboarded	LTM attrition
TCS	593,798	+9,279	~14,000 (reported)	13.6%
Infosys	328,062	-532	~4,000	13.0%
Wipro	243,044	+888	None reported	13.9%
HCLTech	223,889	-3,292	1,056	12.7%

Add those up and the top four employ about 1.39 million people and grew by roughly 6,300 in the quarter. That is a rounding error on a workforce that size. Yet the four companies are clearly running four different strategies.

TCS is rebuilding the base of its pyramid. Its Q1 FY27 results showed the largest quarterly net addition in over a year, after a brutal FY26 in which the workforce fell by 23,460 to 584,519, according to People Matters' report on TCS's Q4 FY26 numbers. TCS told Business Today in April that it had made 25,000 campus offers for FY27, and CEO K Krithivasan was reported in July saying he does not expect "drastic change in headcount" because of AI.

Infosys is holding steady and spending its hiring budget differently. Its Q1 FY27 headcount dipped by just 532, and CEO Salil Parekh reaffirmed the target of 20,000 campus graduates for the year (HR Katha's report on the Q1 call). The more interesting number from that call: Infosys plans to build a team of about 6,000 "frontier" or forward-deployed engineers over the next few years, people who sit with clients and push AI work from pilot into production.

Wipro has gone the other way on freshers. Chief HR Officer Saurabh Govil told Business Today in April 2026: "We don't have any target for fresher hiring for next fiscal. It's completely on demand, very volatile environment." Wipro hired about 7,500 freshers in FY26, and media reports on its Q1 FY27 results said it onboarded none in the quarter.

HCLTech shrank by 3,292 people in Q1 FY27 while net profit rose 20.3% year on year, according to its results coverage. Fewer people, more profit. If you want a one-line summary of the AI economics in this industry, that is it.

What the Layoffs Actually Looked Like

The January versions of this post leaned on a table of layoff estimates for every major firm. Most of those numbers came from roundups that I cannot trace back to a company disclosure, so I have cut them. Here is what is on the record.

TCS, July 2025. TCS announced it would release about 2% of its global workforce, roughly 12,261 people, during FY26 (Apr 2025–Mar 2026), "primarily in the middle and the senior grades", as Business Today reported at the time. Krithivasan said explicitly that the cut was about skill mismatch and deployability, "not because of AI giving some 20 percent productivity gains." By April 2026, TCS said the restructuring exercise was complete.

The bench rules changed with it. The same month, TCS put in place a policy requiring a minimum of 225 billable days a year and capping bench time at 35 days, according to Angel One's coverage. That policy matters more to most readers than the layoff itself, and I come back to it below.

Cognizant, May 2026. Moneycontrol reported, and Outlook Business carried, that Cognizant was planning 12,000 to 15,000 job cuts globally, with India expected to take the largest share. This was a media report, not a company announcement, so treat the number as a range rather than a fact.

The quiet version. Most reduction in Indian IT does not come with a press release. It comes through attrition that is not backfilled, longer bench periods and tighter appraisal curves. Infosys lost 8,440 people in Q4 FY26 alone (Outlook Business, April 2026) without announcing a layoff. CFO Jayesh Sanghrajka put it plainly on the Q1 FY27 call: "Headcount is a function of utilisation and the volumes we see."

AI Has Moved From Pilot to Invoice

In January, the industry was still mostly selling AI as proofs of concept: a six-week pilot, an admired demo, then a stall when the model met real data and a real compliance team. Nine months on, AI work is showing up as disclosed revenue:

TCS reported an annualised AI revenue run rate of $2.6 billion at the end of Q1 FY27, up 13.6% from the previous quarter.
Infosys said AI accounted for 8.2% of Q1 FY27 revenue, with Parekh noting it "has been growing at double digits quarter-on-quarter over the last several quarters" (Zerodha's The Chatter summary of the call).
HCLTech reported "Advanced AI" revenue of $171 million for the quarter, up 62.1% year on year in constant currency, in its Q1 FY27 press release. The board also approved up to ₹3,500 crore for AI data centres in India.
Wipro launched an Applied AI centre of excellence with Anthropic, and TCS announced alliances with Anthropic and Mistral in the same quarter.

Zoom out and the scale is still modest. NASSCOM's Strategic Review in February 2026 put total Indian tech revenue for FY26 at about $315 billion, with AI-related revenue estimated at $10 to 12 billion. AI is a growing slice, not yet the main course.

The catch is that this revenue is not purely additive. Infosys CFO Sanghrajka described "AI-led deflation" as a headwind on the Q1 FY27 call: clients are asking for productivity give-backs in the middle of contracts, not just at renewal. So the same AI that creates new projects also shrinks old ones. I have covered that margin squeeze in more detail in the post on agentic AI and IT services margins.

Overall growth is weak while all this happens. Infosys narrowed its FY27 constant-currency guidance to 1.5% to 3.0%. HCLTech guided 1% to 4%. Wipro guided Q2 FY27 at -1.5% to +0.5% sequentially. When the pie is barely growing and AI makes each slice need fewer hands, headcount can only stay flat by shifting who does the work.

Bench Risk Is the Real Layoff

If you take one idea from this post, take this one. In services companies, the bench is where careers end slowly.

Think of a services firm as a bus that only gets paid for seated passengers. For years the bus ran with plenty of spare seats, and nobody minded a few people standing in the aisle between projects. TCS's 35-day bench cap means the conductor now checks the aisle every month.

Utilisation numbers show the pressure. Business Today reported in April 2026 that Wipro's utilisation fell from 86.4% in Q2 FY26 to 83.5% in Q4 FY26. Each point of utilisation at a 240,000-person company is a couple of thousand people not billing.

Which roles land on the bench first? From what I see across teams, and what the companies themselves describe:

Application maintenance on stable systems. Clients now ask for AI-driven productivity credits on exactly this work.
Manual testing and L1/L2 support. Ticket triage, test case writing and log analysis are the first things AI tools are pointed at.
Coordination-heavy middle layers. TCS's own explanation for its 2025 cuts was the shift away from waterfall models that needed layered leadership.

There is one useful nuance from Wipro's Q1 FY27 call: management said AI productivity gains are highest in greenfield projects and "significantly lower for complex legacy code." If you are the person who understands a gnarly 15-year-old mainframe or SAP estate, AI is not coming for your seat as fast as the headlines suggest. If you are on a clean, well-documented greenfield build, it already has.

The Skill Gap, Measured

The old version of this post quoted a skill gap rising from 18% to 25%. I could not find a primary source for it, so it is gone. Here is what is verifiable.

ManpowerGroup's 2026 Talent Shortage Survey (released February 2026, 3,051 Indian employers) found 82% of employers in India struggle to fill roles, against a global average of 72%. Tech and IT services reported 84%. For the first time, AI model and application development (39%) and AI literacy (38%) ranked as the hardest skills to find.
NASSCOM's FY26 review said more than 2 million professionals had been upskilled in AI, including about 300,000 with advanced skills. On a base of 5.95 million workers, that means most people have had a course, and only around one in twenty has skills deep enough to lead the work.

Training hours are not the problem; TCS alone reported 14.6 million learning hours in Q1 FY27 (Outlook Business). The gap sits between "completed a GenAI course" and "can be trusted to ship an agent into a bank's production systems." Companies have plenty of the first and are hiring for the second.

What the Companies Are Betting On

Strip away the press-release language and the bets are clear. TCS cut layers in 2025 and is now refilling the base of the pyramid with AI-trained freshers under a hard bench cap. Infosys is holding headcount flat and building an elite of forward-deployed engineers; more than 80,000 of its employees already use AI coding tools, according to the Q1 call. Wipro hires only on demand. HCLTech is shrinking people and investing in infrastructure.

I compared the TCS and Infosys approaches in TCS Cut 23,000. Infosys Hired 20,000. Who's Right? and the productivity numbers behind them in the FY26 revenue per employee breakdown, so I will not repeat that here.

The Counterargument: Maybe This Isn't About AI at All

I would be doing you a disservice if I pretended the data only points one way.

First, TCS's own CEO said the 2025 cuts were not AI-driven. Second, the industry as a whole still added people: NASSCOM reported net additions of about 135,000 in FY26, taking direct employment to 5.95 million, and its president Rajesh Nambiar called the sector "a net employee generator." Third, the weakest numbers this year line up with weak demand. Guidance below 4% at every major firm reflects cautious client budgets and macro uncertainty more than any robot.

A fair reading is that AI is an accelerant, not the root cause. Slow demand forces cost discipline, and AI gives companies a credible way to deliver the same work with fewer people while they wait for growth. If demand rebounds strongly in FY28, some of today's caution on hiring will reverse.

My view, for what it is worth: even in a rebound, the roles that come back will not be the same roles. Firms that have learned to run a project with 30% fewer people are not going to unlearn it.

A Career Playbook for People Inside Indian IT

None of this is a reason to panic. It is a reason to be deliberate. Here is what I tell my own teams.

Freshers and early career (0 to 4 years)

Pick employers by their fresher intent. TCS (25,000 offers) and Infosys (20,000 target) have stated plans. Wipro has said it has none. That should shape where you put your effort this placement season.
Show one shipped thing. A small RAG app or agent that runs against real data, with a README that explains failure cases, beats five certificates. Interviewers now ask what broke.
Use AI tools on the job, visibly. You are competing with the tool, so become the person who makes it useful for the team.

Mid-career (5 to 15 years)

This is the most exposed group, and TCS's 2025 cuts targeted exactly these grades.

Measure your bench exposure. Count your billable days for the last 12 months. If you are anywhere near 35 days of bench at a firm with that policy, act now, not at appraisal.
Pair your domain with AI. Banking, insurance and healthcare knowledge combined with hands-on AI delivery is the profile clients pay for. Pure technology skills without domain are easier to replace.
Move toward the legacy-heavy or client-facing work. Wipro's own remarks suggest AI gains are smallest on complex legacy code, and forward-deployed roles like the 6,000 Infosys plans to build sit with the client, not in a shared delivery pool.
Budget real hours. Three to five focused hours a week on building, not watching videos, compounds fast over a year.

Senior and leadership (15+ years)

Own an AI delivery number. Leaders who can show a production deployment, a cost saving or a reduction in ticket volume have a story. Those who managed reporting chains have a harder one.
Learn to price AI work. With clients demanding mid-contract productivity give-backs, the person who can structure outcome-based deals is more valuable than the person who manages headcount.
Look at GCCs and product companies. Your delivery experience transfers, and many of them are building AI teams without the bench pressure of a services firm.

If you want a gut check on the broader job market beyond Indian IT, my look at which developer tasks AI is actually replacing separates the panic from the numbers.

What I Am Watching Next

Q2 FY27 (Jul–Sep 2026) results start arriving shortly. I will be watching Wipro's fresher onboarding, whether HCLTech's headcount keeps falling, and whether TCS's AI run rate keeps growing at double digits. If Q1's pattern holds, the split between AI-ready and AI-exposed roles will get sharper before it gets better.

Your move this week: count your billable days, write down the one AI project you could credibly join tomorrow, and if you cannot name one, start building it.