Tata Boss Says AI Agents Will Replace Half Its Tech Jobs — Is He Right and What Happens Next?

Introduction: The Most Consequential Prediction of 2026

Natarajan Chandrasekaran has run the Tata empire for years. He is not given to dramatic statements. When he sat down at TCS's shareholder meeting on June 9, 2026 and said what he said, the room — and then the internet — stopped.

"The company will have an equal number of AI workers — we call them AI agents — as there are employees. If the company has half a million employees, the day is not far when the company will have half a million AI agents working alongside them." He then went further: he predicted that AI agents would eventually replace around half the roles that TCS currently employs humans to fill. TCS, Asia's biggest software services firm, plans to reduce hiring in coming years as it steps up the use of AI across a company that serves multinationals around the world.

To understand why those words carry the weight they do, you need to understand what TCS is. It is not a small tech startup making bold predictions for investor attention. It employs roughly 600,000 people, making it India's largest private-sector employer. It is the anchor of a $315 billion software services industry that transformed India's economy over three decades. When its chairman says half those jobs will go to AI agents, he is not speculating about someone else's workforce. He is describing his own plan for his own company — and, implicitly, the plan every company like it will be forced to follow.

What TCS Actually Does — and Why AI Threatens the Whole Model

For decades, the Indian IT outsourcing industry ran on a simple and extraordinarily profitable formula. Western companies needed software written, systems maintained, data processed, and customer queries answered. Indian engineers could do that work at a fraction of the cost of hiring locally in the United States or Europe. The model is called labour arbitrage, and it created one of the most remarkable economic transformations of the twentieth century — lifting millions of Indian graduates into a stable, well-paying professional class and anchoring the growth of cities like Bengaluru, Hyderabad, and Pune.

The core of that model is the billable hour. A client hires TCS to build a system or maintain infrastructure, TCS assigns a team of engineers, and it charges for the hours those engineers spend. More engineers on a project means more revenue. A large client contract means a large team. Growth meant hiring more people, which meant more hours, which meant more revenue.

As agentic AI automates the man-day billing model, the $300 billion outsourcing industry faces a brutal pivot. Junior roles that execute repetitive, standardised tasks are most vulnerable. Talathi believes up to half of traditional outsourcing work is directly exposed to extinction by AI, with areas like contract reviews, compliance documentation, data processing, routine coding, and testing being the most vulnerable.

The billable hour is the model that dies when AI agents arrive. An AI agent does not bill by the hour. It completes the task. A piece of code that took a junior engineer two days now takes an AI system two minutes. The client pays for the output, not the time. TCS sees this clearly: as clients adopt agentic AI, systems that can make and execute decisions autonomously, TCS is also adapting its pricing approach. Some clients are moving toward "outcome-based" models where the firm is paid for results rather than billable hours, signalling a fundamental change in how IT services are delivered and valued.

The Jobs Most at Risk Inside TCS and the Broader Sector

Chandrasekaran explicitly ruled out mass layoffs. He confirmed, however, that traditional large-scale graduate hiring will permanently slow. Those are two different things, and the distinction matters. TCS will not fire 300,000 people. It will simply stop replacing them when they leave, stop hiring the thousands of new graduates it has historically absorbed each year, and let the workforce contract naturally while AI handles the work those roles previously did.

The jobs most directly in the firing line are the ones that built the outsourcing model in the first place. TCS's restructuring has primarily affected middle and senior management employees, reflecting the reality that AI can now handle the coordination and oversight work that populated those layers. Software engineers handling routine code generation, testing, and debugging face the sharpest exposure. "A lot of the code and software will get auto-generated, so those roles will diminish over time," TCS CTO Harrick Vin said.

Data entry and document processing roles have already been hit hard across the sector. Major outsourcing companies including Cognizant, Infosys, and Wipro have reduced data entry headcount by 30 to 40 percent since 2024. AI-powered document processing tools have made manual data entry obsolete for most standard document types, and 65 to 80 percent of current data entry roles are expected to be eliminated by 2028. Customer service, compliance documentation, and routine legal work face equivalent exposure.

In February 2026, venture capitalist Vinod Khosla warned that India's IT services and BPO sectors could "almost completely disappear" within five years as AI systems outperform human expertise. Khosla's prediction is more extreme than Chandrasekaran's, but it is directionally consistent. The question is not whether the labour arbitrage model is under threat. It is how much of it survives, and in what form.

What TCS Is Doing About It — and Whether It Is Enough

TCS is not standing still. It has announced one of the largest corporate reskilling programs in the world. TCS is retraining about 100,000 employees each year as AI reshapes the global technology landscape and redefines traditional outsourcing models. The company will also invest in new learning, certification, and career development opportunities. TCS's AI revenue crossed an annualised $2.3 billion in the quarter through March 2026, and Chandrasekaran expects all of TCS's revenue to have an AI component by 2028 to 2030.

The reskilling emphasis is on AI fluency: training engineers to build, supervise, maintain, and extend AI systems rather than perform the tasks those systems now automate. The theory is sound. The execution challenge is immense. TCS has 600,000 employees. Retraining 100,000 per year is 16 percent of the workforce annually — a significant pace, but one that leaves a large portion of the company in transition for years while the market moves faster than any retraining program can track.

The broader reskilling gap is even starker when viewed at industry scale. IBM projects that 1.4 billion workers globally need reskilling by 2028. Only 58 million completed AI training in 2025 — representing just 4.1 percent of the required volume. The gap between what needs to happen and what is actually happening is not a rounding error. It is structural, and no single company's internal training program closes it.

Where This Fits in the Broader 2026 Tech Layoff Wave

Chandrasekaran's prediction did not arrive in isolation. It landed inside a year that was already delivering one of the worst job loss periods in technology employment history. Over 142,000 US tech workers lost jobs in the first five months of 2026 alone — a 33 percent increase over the same period in 2025 — with AI cited as the direct cause in a growing proportion of those cuts.

TCS itself planned to cut 12,000 jobs — about 2 percent of its workforce — as part of a transition the company described as building a "future-ready organisation." That cut, announced before the shareholder meeting, affected primarily middle and senior management. The 50 percent prediction at the shareholder meeting is the long-horizon version of the same force that produced those 12,000 immediate cuts. The first number is the adjustment that is already underway. The second is where the adjustment ends up.

Anthropic CEO Dario Amodei has predicted that AI could eliminate half of all entry-level white-collar jobs within one to five years, potentially pushing unemployment to 20 percent. McKinsey estimates AI could replace up to 375 million jobs globally by 2030. A SignalFire report found AI automation led to a 25 percent decline in entry-level tech hires at Meta, Microsoft, and Google from 2023 to 2024 alone. Chandrasekaran's prediction at TCS is not an outlier. It is consistent with what the most credible industry analysts and company leaders are saying across the sector. The difference is that he said it about his own 600,000-person workforce, which makes it concrete in a way that industry projections are not.

What Workers in Tech Services Should Realistically Do Right Now

The practical question for anyone working in IT outsourcing, software services, or any adjacent career is not whether Chandrasekaran is right. He is largely right, and the pace of change in 2026 makes it harder to argue otherwise. The practical question is what you do with that information before the next cycle of cuts narrows your options.

The first and most important move is an honest assessment of your current role. The jobs at lowest risk in IT services are those that require judgment, client relationships, complex system architecture, and the ability to direct AI tools toward goals that the AI cannot define on its own. The jobs at highest risk are those that consist primarily of executing well-defined, repeatable tasks — writing standard code to spec, processing documents, testing predefined scenarios, answering tier-one support queries. If your role is primarily in the second category, the question is not whether it will change. It is whether you change first.

TCS's CTO has been direct about the direction of travel: auto-generated code and software will reduce the need for engineers performing those functions over time. The engineers who remain will be those who can supervise the systems doing the generation, catch what the AI gets wrong, architect the systems the AI builds within, and translate business requirements into specifications that AI agents can execute. Those skills are learnable. They are not the same skills that got most people hired at TCS five years ago, which is precisely the problem.

The reskilling programs TCS and its peers are offering are a real resource, but waiting for your employer to retrain you is a slower strategy than taking the initiative yourself. AI engineering courses, prompt engineering certifications, cloud architecture credentials, and practical experience building agentic workflows are all available through platforms including Coursera, DeepLearning.AI, and AWS's own training ecosystem. The engineers who are thriving inside IT services firms in 2026 are not the ones waiting to be redeployed. They are the ones who have already repositioned themselves as the people who make AI systems work rather than the people AI systems replace.

Conclusion: Is He Right?

Yes. The specific number — 50 percent — may prove optimistic or pessimistic depending on how fast agentic AI matures and how effectively the industry adapts. But the direction is not in question. The business model shift forces an overdue evolution from cost arbitrage to genuine innovation. A mismanaged transition could devastate urban economies, real estate, and ancillary services.

What Chandrasekaran did at the June 9 shareholder meeting was unusual and important. Most executives soften these predictions, bury them in qualifications, or make them about "transformation" rather than displacement. He stated plainly that AI agents will equal the human workforce at TCS and that hiring will permanently decline. That honesty is more useful than reassurance, even if it is harder to hear.

"Will it definitely lead to a decrease in hiring — absolutely. That does not mean there are no future opportunities," he said. Both halves of that sentence are true. The decrease in hiring is already happening. The future opportunities are real but require active pursuit rather than passive waiting.

The thirty-year career path that turned Indian IT engineering into a dream for millions of middle-class families was built on a model that is now under fundamental pressure. The next thirty years will be built on something different. The workers who will navigate that transition successfully are the ones deciding right now which side of the AI divide they want to stand on — the people who build and direct these systems, or the people the systems are built to replace.

That decision does not wait for the next shareholder meeting.