The bottom line: India's IT services industry is not going through a hiring slowdown — it is going through a model change. The bench-and-bill system that hired engineering graduates by the hundred thousand, trained them for nine months, and billed clients for their hours has been replaced by a just-in-time, AI-augmented, outcome-priced system that hires fewer people at higher skill bars. TCS dropped fresher intake from 44,000 (FY26) to 25,000 (FY27). Wipro went from 19,000 (FY22) to 7,500 (FY26). Infosys held flat at 20,000 but tied every hire to demand visibility. The 1.6 million STEM graduates India produces each year now face a market where generic coding skills are commoditised and AI-ready specialism commands a 3x salary premium. The mass-hiring era is over. What replaces it is not fewer jobs — it is a different kind of job.
What was the mass IT hiring model — and why did it work for 25 years?
For over two decades, India's IT services companies — TCS, Infosys, Wipro, HCLTech, Tech Mahindra — operated on a straightforward model: hire large numbers of engineering freshers at ₹3–4 LPA, put them on a bench, train them for 6–9 months in enterprise technologies, deploy them to client projects, and bill clients on a time-and-materials basis (engineers × hours × rate). The model worked because of three structural advantages:
- Labour arbitrage: Indian engineers cost 5–10x less than US equivalents, making the bench economically efficient even at 15–25% utilisation gaps.
- Demand certainty: Clients signed multi-year contracts for maintenance, testing, and low-complexity development work that required baseline headcount, not specialised skills.
- Training pipeline: India's 1.6 million annual STEM graduates produced a steady supply of trainable talent at predictable cost.
TCS alone hired over 100,000 freshers in its FY2022 peak, soaking up campus talent at unprecedented scale. Infosys onboarded over 50,000. Wipro took on 19,000. The system was the largest private-sector employment engine in India's history — the IT industry employs roughly 6 million people and generates $315 billion in revenue (Nasscom, FY2026).
The model was elegant: headcount was the product. More engineers meant more billable hours, which meant more revenue. Growth and hiring were mathematically coupled.
What broke it: the three forces that killed the bench
Force 1 — AI compressed the work that freshers used to do
The tasks historically assigned to freshers — writing boilerplate code, running test suites, maintaining legacy systems, performing manual QA, generating documentation — are now done by AI tools in a fraction of the time. GitHub Copilot, Claude, ChatGPT, and code-generation agents have made a single mid-level engineer with AI tooling roughly 2.5–3x more productive on routine tasks than a team of freshers without it.
The Financial Express reported in April 2026 that IT majors are "moving to outcome-based billing" and "disengaging from headcount-led billing" — the death certificate of the bench model, because if you bill by outcome rather than by hours, headcount is no longer the revenue driver (Financial Express, Apr 2026).
Force 2 — AI infrastructure costs are eating the hiring budget
Zoho founder Sridhar Vembu said it plainly on August 2, 2026: "The money that would have gone to new employees is now going to AI and data centre costs." Zoho has not laid off a single person — it simply stopped creating new roles. The budget that previously funded headcount expansion now funds GPU clusters, AI platform licenses, and inferencing infrastructure (India Today, Aug 3 2026; Indian Express, Aug 2026).
This is not a cyclical slowdown. It is an accounting decision: every rupee spent on an AI platform is a rupee not spent on a fresher's salary. And the AI platform makes the existing workforce more productive, which further reduces the need for additional headcount.
Force 3 — Clients want specialists, not trainable generalists
When billing was time-and-materials, clients did not care whether the engineer assigned to their project was a fresher or a specialist — they paid for hours. In an outcome-based model, clients care deeply: they want the person who can deliver the outcome, not someone who needs nine months of training first. Infosys CEO Salil Parekh confirmed the shift in April 2026: "We have different starting compensation for people who are coming with skills more attuned to AI. We are also building a pool of forward-deployed engineers" (Financial Express, Apr 2026).
The market has split. There is the mass-campus track (₹3.5–6 LPA, shrinking, generic) and the AI-specialist track (₹8–21 LPA, growing, requires demonstrated skills). The middle is hollowing out.
The numbers: how deep is the cut?
The data is unambiguous. This is not a cyclical dip — it is a structural contraction in the mass-hiring track:
| Company | FY22 peak | FY26 actual | FY27 target | Change |
|---|---|---|---|---|
| TCS | ~100,000+ | 44,000 | 25,000 | -75% from peak |
| Infosys | ~50,000+ | 20,000 | 20,000 | -60% from peak |
| Wipro | 19,000 | 7,500 | Not disclosed | -61% from peak |
| HCLTech | Not disclosed | Not disclosed | Not disclosed | N/A |
| Tech Mahindra | Not disclosed | Minimal | Not disclosed | "Pause" |
Sources: TCS hiring data from SightsIn Plus, Apr 2026 and Trak.in, Apr 2026. Wipro data from Financial Express, Apr 2026. FY27 hiring targets from New Indian Express, Jul 2026.
Three of the top five IT firms — Wipro, HCLTech, and Tech Mahindra — declined to give annual fresher hiring targets for FY27 at all. In a sector that used to announce campus quotas as a signal of confidence, the silence is the signal. Tech Mahindra's CFO Rohit Anand said they "took a pause" on fresher hiring because "the demand profile with AI is also changing."
Meanwhile, the work itself is not disappearing. India's IT industry grew to $315 billion in FY2026 and added 135,000 net new jobs. But the character of those jobs changed: AI/ML roles surged 33% year-on-year in July 2026 (Naukri JobSpeak), while entry-level roles fell 44% (Economic Times, Jul 2026; OwnYourCareer, Jul 2026).
The Indian IT industry is investing heavily in AI data centers — the same infrastructure spending that Vembu says is crowding out the entry-level hiring budget.
Why is this a model change, not a hiring slowdown?
A cyclical slowdown means the jobs come back when demand returns. A model change means the jobs that leave do not come back — because the system that created them no longer exists.
Consider the economics. In the 2019 model, an IT firm hired 10,000 freshers at ₹3.5 LPA, accepted 25% Year 1 attrition (net 7,500 retained), and paid an effective cost of ₹4.67 lakh per retained hire. In the 2026 model, the same firm hires 3,000 AI-ready specialists at ₹6–8 LPA, sees 20% attrition (net 2,400 retained), and pays ₹7.5–10 lakh per retained hire. The total cost is lower, the output is the same (or higher, via AI productivity gains), and the bench is smaller. The math works — which is exactly why the shift is irreversible.
TCS's CHRO Sudeep Kunnumal said it in April 2026: the headcount decline came from "better productivity, increased use of automation, and changing business needs" — not from restructuring. "As TCS brings in more advanced tools, including AI-based solutions, some roles are no longer needed, while demand is growing for people with higher and more specific skills" (SightsIn Plus, Apr 2026).
This is also reflected in the financial results. Profits at TCS, Infosys, and HCLTech rose in FY2026 even as headcount fell — a combination that only makes sense if you understand how AI deflation is reshaping India's IT industry. Revenue per employee is rising. Headcount per revenue unit is falling. The system is not shrinking — it is densifying.
What is replacing the bench? Three tracks for 2026 and beyond
Track 1 — The AI-specialist fresher (₹8–21 LPA, growing)
Companies still hire from campuses, but they are paying premium packages for graduates who arrive with demonstrated AI skills: a deployed LLM application, a production ML model, a RAG system with cited sources, or an agentic AI workflow. These hires skip the support-queue grind and go directly into forward-deployed engineering roles. Infosys is explicitly building this pool — Parekh called them "forward-deployed engineers to make sure that we can do more work directly with clients" (Financial Express, Apr 2026).
AI/ML architect demand has grown 11x year-on-year. AI/ML research scientist roles are seeing "explosive growth" (OwnYourCareer, Jul 2026). This track is real, expanding, and accessible — but the minimum bar is a working GitHub portfolio, not a transcript.
Track 2 — The domain-plus-AI career switcher (₹6–16 LPA, emerging)
The most underrated shift in the market: companies increasingly value domain expertise paired with AI literacy over pure computer science credentials. A nurse who understands clinical workflows and can evaluate an AI diagnostic tool's output is more valuable than a fresh CS graduate who can fine-tune a model but has never read a patient chart. This is not a side door — it is a structural change in what the market defines as "qualified."
For the 1.6 million STEM graduates India produces annually, this means the path is no longer "CS degree → campus placement → bench training → deployment." It is "domain knowledge + demonstrated AI capability → direct hire at a premium." India's broader AI talent paradox — world #1 in AI skills but thin on frontier research — is the same paradox at the entry level: lots of CS graduates, not enough AI-ready ones.
Track 3 — The GCC route (₹6–10 LPA, expanding steadily)
India's 2,117 Global Capability Centres employ 2.36 million professionals and generated $98.4 billion in FY2026 (Zinnov-Nasscom, May 2026). GCCs pay 12–20% more than IT services firms for comparable roles and are projected to add 200,000 net employees in FY27. 64% of GCCs foresee up to a 20% increase in fresher hiring, often through hackathons and specialised internships rather than traditional campus recruitment (GCC Journal, Apr 2026; Zinnov-Nasscom, May 2026).
India's GCCs are hiring at 200,000 for FY27 even as traditional IT services firms cut campus quotas — the jobs are not disappearing, they are migrating to a different employer model. GCCs want ownership and decision-making, not just delivery. They hire people who can solve problems, not fill seats.
What should the 1.6 million engineering graduates of 2026 do?
The practical advice for a 2026 engineering graduate is not "learn AI" in the abstract — it is concretely different depending on where you are in the talent stack:
If you are at a tier-1 college (IIT, NIT, BITS, top state engineering colleges)
The campus system still works for you, but the premium has shifted. Companies are paying ₹8–21 LPA for graduates who can demonstrate AI project work. Build a portfolio: a working Streamlit or Gradio app that uses an LLM API to solve a specific problem, with documented design decisions. Deploy it. Get a live URL. This is not about LeetCode — it is about proving you can ship.
If you are at a tier-2/3 college
The mass-campus wing of the funnel — the one that absorbed graduates from tier-2/3 colleges into the ₹3.5 LPA bench pipeline — has structurally contracted. Your addressable market is approximately 25–30% of what it was in 2021 if you are competing on the same track as everyone else with a generic CS degree. The exit is not a higher CGPA; it is a demonstrated-skill portfolio that puts you on the specialist track. Six months of Python foundations, three months of AI project work (LangChain + OpenAI API, a deployed ML model, or an agent system), and a portfolio with live URLs is the concrete path.
If you are a career switcher
The path is domain knowledge + AI literacy, not CS degree + algorithms. One to three months of AI training, three to six months of building a portfolio that applies AI to your domain, and two to four months of targeted job search. The market pays a premium for people who understand a problem space and can deploy AI to solve it — more than for people who know algorithms but have no domain depth.
Is the IT industry itself shrinking?
No. The IT industry is not collapsing — it is restructuring around a higher productivity baseline. Nasscom reports FY2026 revenue of $315 billion, up from $298 billion the prior fiscal year. The industry still employs roughly 6 million people. TCS added 14,000 campus graduates in Q1 FY27 and made a net addition of 9,279 employees, bringing its headcount to 593,798. Indian IT stocks rallied 22% in July 2026, driven by AI-led margin expansion.
But the composition is changing. The fresher share of total hiring dropped from approximately 28% to approximately 15% — a structural rebalancing, not a seasonal dip. The industry is still hiring. It is just hiring different people for different reasons at different price points.
What about the macro picture — is India's demographic dividend at risk?
Yes, and this is the uncomfortable part. India produces approximately 1.6–1.7 million STEM graduates per year. In the mass-hiring era, the IT services industry absorbed a significant percentage of them into the bench pipeline. That absorption mechanism is now operating at roughly 25–40% of its peak capacity. The graduates are still coming. The pipeline that absorbed them is not.
This is the policy challenge Vembu flagged when he said "the real issue facing our nation is how to create jobs for our massive cohort of youth in this very uncertain global landscape" (Indian Express, Aug 2026). India's demographic dividend peaks by 2030. The window to build a new absorption mechanism — whether through GCC expansion, startup ecosystems, apprenticeship programmes, or public-sector digital initiatives — is narrowing.
The Indian government's approach to building AI capacity also matters here. India is pursuing a small AI strategy built on neurosymbolic models that could be more accessible to smaller companies and tier-2/3 cities — but it is early, and the absorption gap is now.
FAQ
Is the mass IT hiring era permanently over, or will it return when the economy recovers?
The mass-hiring model is structurally over, not cyclically. The shift from headcount-based billing to outcome-based billing means companies have no economic incentive to maintain large benches. Even when demand returns, the response will be to hire AI-augmented specialists, not to rebuild the fresher pipeline at FY22 scale. TCS, Wipro, and Tech Mahindra have all explicitly tied future hiring to demand visibility, not headcount targets.
How much did TCS, Infosys, and Wipro cut fresher hiring?
TCS reduced fresher hiring from 44,000 in FY26 to 25,000 for FY27 (and over 100,000 at its FY22 peak). Infosys held steady at 20,000 for both FY26 and FY27. Wipro dropped from 19,000 (FY22) to 7,500 (FY26) and did not disclose an FY27 target.
Can a 2026 engineering graduate still get hired without AI skills?
Yes, but the addressable market has shrunk to roughly 25–30% of its 2021 size. Without demonstrated AI capability, you are competing for a diminishing pool of maintenance, legacy migration, and low-complexity development roles — and you are competing against candidates who do have AI skills and are willing to accept the same offer as a fallback.
What salary can an AI-ready fresher earn in 2026?
AI-ready freshers with portfolio projects (deployed LLM applications, ML models, or AI agent systems) can earn ₹8–21 LPA, compared to ₹3.5–6 LPA for the standard mass-campus track. GCCs pay ₹6–10 LPA for comparable roles, with a 12–20% premium over traditional IT services firms.
How many GCCs are there in India, and are they hiring freshers?
India hosts 2,117 Global Capability Centres as of FY2026, employing 2.36 million professionals and generating $98.4 billion in revenue (Zinnov-Nasscom). 64% of GCCs forecast up to a 20% increase in fresher hiring, often through hackathons and specialised internships rather than traditional campus placement.
What is the difference between the bench model and the new IT hiring model?
The bench model hired freshers in bulk, trained them for 6–9 months, deployed them to time-and-materials projects, and billed clients per engineer-hour. The new model hires fewer people at higher skill levels, deploys them to outcome-priced projects, and uses AI tooling to multiply per-engineer productivity. Headcount is no longer the revenue driver — capability is.
Is India's demographic dividend at risk because of the IT hiring shift?
India produces 1.6–1.7 million STEM graduates annually, but the IT services bench pipeline that absorbed a significant share of them is now operating at 25–40% of peak capacity. The demographic dividend window peaks by 2030, meaning the country has a narrowing window to build alternative absorption mechanisms — GCCs, startups, apprenticeship programmes, and public-sector digital roles.
Every claim here is traced to a primary source, dated, and listed under Sources. Research and drafting are AI-assisted; editing, verification and publication are human decisions, and a person is accountable for what appears on this page. How we work →
Discussion
0 comments