India’s IT services companies are hiring even as their overall workforce shrinks. TCS ended 2025–26 with more than 23,000 fewer employees while reporting around 25,000 campus offers. These figures do not establish how much of the decline was caused by artificial intelligence. They do, however, raise a question that net employment numbers cannot answer: what happens to existing workers when the roles firms need begin to change?
Hiring for new skills can coexist with declining demand for older ones. Workers whose roles are reduced may need to retrain, accept different work or find another employer. The policy challenge is whether firms and public institutions support that transition, or leave workers and their households to finance and manage it themselves.
What Earlier Transitions Show
When Indian banks began introducing computers in the 1980s, employee unions challenged the terms of computerisation because of its implications for clerical employment. Negotiations concerned where machines would be introduced, the pace of adoption and the protections that would accompany the change.
For a period, manual and computerised systems operated alongside each other. This gave existing workers time to adapt, while making the terms of technological adoption a matter of workplace negotiation.
The experience does not establish that earlier technological transitions were uniformly gradual or protective. It illustrates something more specific: institutions capable of negotiating on workers’ behalf could shape the pace and consequences of adoption. Time to adjust was partly an outcome of those arrangements.
The Two Speeds of Technological Change
The speed of technological adoption and the speed of worker adjustment are different. A firm may introduce a new technology relatively quickly. A worker whose role is changing needs time to identify viable work, acquire the required skills and find an employer willing to hire them.
When adoption moves faster than this process, the costs appear in the interval between roles: time spent training, income forgone and uncertainty about whether the next job will offer comparable pay or security. New vacancies provide little reassurance if displaced workers cannot realistically qualify for them.
The OECD’s 2020 paper, Technology and the Future of Work in Emerging Economies: What Is Different?, emphasised that the pace and extent of adoption affect how automation potential translates into displacement. Its estimates predate the current wave of generative AI, but the distinction remains useful. Technical capability alone does not determine employment outcomes; firms’ choices and the institutions surrounding them also matter.
When Adjustment Costs Shift to Workers
Employment totals reveal only part of this process. A worker may find another job at lower pay, spend months without income or take work that makes little use of their previous experience. A household may draw down savings to cover the gap. These losses can persist even when the worker returns to employment.
A transition can consequently appear manageable at the firm level while imposing substantial costs on individuals. Slower recruitment, redeployment and selective exits may allow a company to change its workforce without a sudden collapse in headcount. They do not reveal whether affected workers have received adequate support.
Training is one part of the answer, but completing a course does not guarantee access to a suitable role. Support must connect new skills with actual opportunities, including those with other employers.
Making Technological Transitions Work Better
The first requirement is clearer information about workforce change. Where firms identify automation or changes in the organisation of work as reasons for reducing roles, they should explain which kinds of work are affected, what redeployment has been attempted and what support is available. Hiring and headcount figures alone cannot show whether existing workers are moving successfully into new roles.
Second, firms should retain responsibility for helping workers adapt when their adoption decisions change the skills required. This means identifying affected roles early, providing time and training for redeployment, and assessing whether workers can move into available positions. Where internal placement is not feasible, assistance with training and job search should form part of the transition.
Third, public support needs to cover the period between jobs. Access to essential benefits, employment services and affordable training should continue when employment ends. Workers also need financial support that allows them to pursue a viable transition without being forced immediately into whatever work is available.
Technological adoption can improve productivity while changing who has access to work. Its success should therefore be assessed partly by whether existing workers can reach the new opportunities it creates. Firms and public institutions need to build those routes before workers are left to bear the costs of their absence.