AI Is Making Money. Is It Closing the First Door into Indian IT?

 

AI Is Making Money. Is It Closing the First Door into Indian IT?

AI revenues are scaling, but the evidence points to a more complicated jobs story: fewer routine tasks, a narrower entry-level funnel and a growing need to redesign how graduates get their first real experience.

INVESTIGATIVE BUSINESS FEATURE  |  INDIA  |  12 OCTOBER 2026

Illustrative artwork: the issue is not whether AI will eliminate every junior role, but whether the first rung of the career ladder is being rebuilt.

The money is beginning to show up in company disclosures. Tata Consultancy Services said its annualised AI revenue crossed US$2.3 billion in the fourth quarter of FY2025–26. That is a revenue run-rate—an annualised pace based on current activity—not a separately reported full-year AI revenue figure. It nevertheless signals that enterprise AI work is moving beyond pilots into paid services.

The labour-market evidence is less triumphant. A survey of 651 Indian IT firms by the Indian Council for Research on International Economic Relations (ICRIER), conducted from November 2025 to January 2026, found that 55 per cent of firms reported lower entry-level employment after AI adoption, while 35 per cent reported an increase. The figures are about the share of surveyed firms reporting a direction of change. They do not mean that 55 per cent of entry-level workers lost jobs.

That distinction is the heart of the story. India is not yet looking at a simple collapse in IT employment. It is looking at a possible weakening of the traditional first job: the supervised work through which a graduate learned to test, document, debug, support and deliver software before moving into more complex roles.

Revenue is growing, but it is not the same as hiring

TCS reported annualised AI revenue above US$2.3 billion in Q4 FY26, while its total FY26 revenue declined 0.5 per cent in reported terms and 2.4 per cent in constant currency. The company also reported a 584,519 employee headcount at year-end. The juxtaposition matters: a fast-growing AI line can coexist with muted overall growth. AI revenue does not automatically translate into more workers, and revenue run-rate cannot be compared one-for-one with recognised annual revenue or employee additions.

AI services may create work in consulting, data engineering, model deployment, cloud infrastructure, cybersecurity, governance and redesign of business processes. But one project can also reduce hours needed for code generation, routine testing, documentation, customer support or first-line analysis. Whether the net result is more jobs depends on the new work created, how much productivity gains reduce prices and expand demand, and whether companies choose to scale teams or protect margins.

The first impact is often task-level. Generative AI can draft code, generate test cases, summarise logs and produce documentation. A junior engineer who once spent weeks on these tasks may be expected to do them faster with AI—or may find that a smaller team can handle the same workload. Human judgement remains essential for requirements, architecture, security, reliability and accountability, but those responsibilities are usually not the first tasks assigned to a fresh graduate.

What the Indian firm survey actually finds

Source: ICRIER–OpenAI survey of 651 firms. Percentages show firms reporting a decline, no major change, or an increase in employment by experience level.

ICRIER found that 65 per cent of firms reported a decline in overall hiring following AI introduction, against 24 per cent reporting an increase and 11 per cent no major change. For entry-level employment specifically, 55 per cent reported a decline and 35 per cent an increase. Mid-level hiring was more positive: 42 per cent reported an increase, while 25 per cent reported a decline. Senior employment was largely stable, with 82 per cent of firms reporting no change.

These are self-reported firm responses, not a controlled experiment isolating AI from client demand, post-pandemic normalisation, interest rates or broader hiring restraint. The study itself cautions that reduced hiring is not equivalent to a fall in the sector’s total employment. Its findings support a credible warning about the mix and pace of hiring; they do not prove AI alone caused every change.

The task mix helps explain why the entry point is exposed. Respondents most often named data-entry and clerical work as seeing reduced demand, followed by manual testing and quality assurance and entry-level programming. Software development and engineering was the business division most often identified as affected by AI. “Affected” does not necessarily mean “eliminated”: teams may use AI to increase output, reduce cost or shift workers to higher-value tasks.

A hiring rebound can still leave freshers behind

Illustrative artwork: AI assistance can speed routine coding, raising expectations for what an entry-level employee should deliver.

Online hiring indicators show the same split. The July 2026 foundit Insights Tracker recorded technology-professional demand up 10 per cent year on year, while the IT sector’s share of fresher hiring fell to 24 per cent from 32 per cent a year earlier. AI-led roles made up 4 per cent of entry-level hiring, up from about 1 per cent. Almost one-third of entry-level openings were internships, contracts or project work, compared with about one in eight two years earlier.

Those measures have limits: job-posting shares are not actual hires, and the tracker does not count every recruitment channel or campus offer. But taken alongside the ICRIER survey, they point toward a tighter conversion from degree to stable first job—even while technology demand and AI-specific openings rise.

September data add a useful qualification: demand is recovering. Xpheno counted 57,000 active openings in the IT-services cohort, an 18-month high. Yet mid-senior and senior positions accounted for nearly 60 per cent of active technology demand. Indeed India also reported that software-development postings mentioning AI had grown 138 per cent between Q2 2024 and Q2 2026, while overall software-development postings were roughly flat. Employers are adding AI capability, but the recovery is weighted toward people able to apply it quickly.

The pattern is not uniform across employers. Large IT services companies have historically hired graduates in cohorts, trained them and deployed them across client work. Global capability centres (GCCs), product companies and technology teams embedded in banks, manufacturers, hospitals and retailers can create other routes in. Yet many specialist AI jobs still ask for experience, domain knowledge or evidence of delivery. A vacancy labelled “AI” is not necessarily an entry-level pathway.

The missing rung in the career ladder

Illustrative artwork: training works best when paired with supervised, paid work on real projects.

For decades, India’s IT-services model used entry-level work as both production capacity and an apprenticeship system. Graduates learned inside teams, often through repetitive tasks that were billable to clients. If those tasks are automated, companies may save time and cost—but the sector must decide how a graduate will acquire the judgement that those tasks once built.

That is the “missing rung” risk. Employers increasingly ask candidates for production-ready skills, internships, portfolios and practical experience. But if companies cut the supervised junior work that produced experience, graduates face a circular barrier: they need experience to get hired, while the jobs that once supplied experience are shrinking or changing form.

TCS’s own disclosures show a potential counterexample to pure substitution: in FY26 it reported 69 million learning hours and more than 270,000 employees improving AI/ML proficiency. Training at this scale matters, but internal upskilling for existing employees does not by itself answer how new graduates enter. The relevant test is whether learning connects to paid project assignments, mentors and a visible route into higher-responsibility work.

Where new entry-level work could grow

The new junior role may be less about producing code line by line and more about using tools safely inside a defined workflow. Plausible early-career openings include validating AI outputs, building evaluation datasets, testing AI-enabled applications, monitoring model behaviour, cleaning and governing data, securing AI systems, integrating APIs and translating a business process into requirements. These jobs need foundational software skills as well as domain knowledge and communication.

But this cannot become a rebranding exercise in which low-paid annotation or short contracts are presented as equivalent to a durable engineering career. The quality question is whether new roles teach transferable skills, pay fairly, provide supervision and lead to progression. Job titles and AI certificates alone will not establish that.

Fresh graduates also need a wider map of employers. Technology work is moving into GCCs and non-IT sectors. A banking technology team may need graduates for data quality, fraud analytics and secure AI operations; a manufacturer may need them for industrial software and cloud systems; a hospital network may need data integration and privacy expertise. These pathways broaden the market, although they do not replace the scale of the old campus-hiring pipeline overnight.

The policy and business response

Who	What to do	What success should mean
IT employers	Publish fresher hiring and onboarding numbers; keep paid graduate cohorts; attach AI productivity to new service growth and internal mobility.	A transparent bridge from training to billable, supervised work.
Colleges	Assess practical problem-solving, software fundamentals, data handling, security and AI verification—not certificates or prompt tricks alone.	Graduates can demonstrate work they can explain and maintain.
Government and industry bodies	Expand paid apprenticeships linked to real employer projects; co-fund placements with outcome reporting and independent evaluation.	More first-job seats and evidence on conversion to regular employment.
GCCs and non-IT firms	Create entry tracks in domain technology, AI operations, testing, cloud and cyber risk.	A wider set of employers absorbs graduates beyond IT services.

A practical compact between employers, education institutions and public skilling programmes.

Public support should reward placements and progression, not just enrolments or course completions. A credible programme would track paid apprenticeship starts, conversion to regular jobs, wages after six and twelve months, retention, and the share of participants from smaller cities and less-resourced colleges. Firms should disclose enough experience-level hiring data to distinguish a real talent pipeline from occasional recruitment campaigns.

The test is the first job, not the AI headline

Illustrative artwork: the opportunity is to rebuild a visible career path from junior roles to AI-enabled specialist work.

AI is creating valuable revenue for technology companies, but the gains do not distribute themselves. The evidence so far does not support a prediction that India’s IT workforce is about to disappear. It does support concern that fewer routine tasks may mean fewer conventional entry-level seats, while the remaining roles demand greater readiness on day one.

India’s answer cannot be to slow down AI adoption or to tell every graduate to become an AI researcher. It is to make the productivity dividend finance a new apprenticeship model: paid, supervised work where graduates learn to build, test, secure and govern systems with AI. If companies can sell AI transformation but cannot explain how the next generation of technologists gets its first year of experience, the industry may be weakening the talent pipeline on which its future growth depends.

Sources and reporting notes

ICRIER, AI and Jobs: This Time Is No Different — survey of 651 Indian IT firms

ICRIER–OpenAI report PDF and methodology

TCS Q4 and FY2026 results: annualised AI revenue, revenue and headcount

TCS Integrated Report 2026: human capital and training disclosures

Foundit Insights Tracker: July 2026 fresher hiring data, reported by Economic Times

Foundit methodology: online job-posting activity

Business Standard: September 2026 active IT-services job demand and experience mix

The ICRIER results are survey responses, not a causal estimate of job losses. Foundit data track online hiring and represent postings, not completed hires. TCS’s annualised AI revenue is a run-rate measure, not full-year recognised AI revenue. Visuals are AI-generated editorial illustrations and do not depict specific people, companies or workplaces.

— Biz Masala Editorial Desk