Indian IT Firms Plan to Cut Bench Strength to 8-10% by FY27

For decades, India's top IT outsourcers kept a "bench": a pool of engineers not currently assigned to a billable client project, kept warm in case a customer called with urgent work. The bench was also where fresh graduates learned the job. TeamLease now projects that buffer will shrink to 8-10% of staff in FY27, down from the 12-15% range that has held for years, and the driver is not layoffs. Generative AI is absorbing the demand volatility the bench was built to absorb, and the same companies that cut the headcount are also losing the training ground the bench used to be.

A March 2026 readout on the top five Indian IT outsourcers puts the arithmetic plainly. Pooled bench across TCS, Infosys, Wipro, HCLTech, and Tech Mahindra fell by roughly 75,000 over two years to about 225,000, shrinking the bench ratio from above 20% to a band of 8-15% depending on the firm. Utilization rates are now closing in on 85%, and the average time a fresher or mid-career engineer spends waiting for a project has nearly halved. Times of India and a parallel industry synthesis both track the same direction of travel.

Read that way, the 8-10% target describes the end of a real-options hedge. The bench was never a cost centre the firms wanted; it was insurance against a client demanding 200 Java developers on a Tuesday. Carrying that insurance had a second benefit nobody put on a slide: those engineers were also learning the company's tooling, its client verticals, and its quality bar while they waited. The same person who absorbed a demand spike also onboarded the next wave of freshers. TeamLease's FY27 outlook describes this rebalancing: utilization is the lever, generative AI is the substitute, and the buffer is the cost that comes out.

The substitution is concrete. Code generation tools now handle a meaningful slice of the boilerplate that used to land on a junior developer's plate. Test automation, particularly for regression suites, has compressed what used to be a multi-engineer effort into a smaller one. L1 support, the tier that triages tickets before a human reads them, has moved into AI workflows first, because the work is the most rules-bound and the most repeatable. The fresher cohort that historically got its first six months of project exposure on the bench is now meeting those tasks in a tighter, billable assignment from week one, with less slack to learn on someone else's ticket queue.

The macro frame is not benign. IANS reports that India IT services growth is forecast at 5-6% on revenue and 8-10% on volume for FY27, and the quarter that just ended is tracking slow on macro, AI, and deal delays. A leaner bench is the lever firms are pulling to defend margins in a year when revenue is not doing the work. TeamLease's Q1 FY27 earnings call calls the IT hiring environment steady and pushes the recovery to H2, framing the AI and digital skill demand as the thing that will pull hiring back, not the broad-based volume of the prior cycle.

The falsifier is mechanical. If a demand surge hits in Q2 or Q3 and the firms cannot stand up a project team inside two weeks because the buffer is gone, the 8-10% target will look like an under-bench rather than an efficient one. The market will know inside two earnings cycles. Until then, the industry synthesis on the same data is unambiguous: the buffer that used to be both insurance and on-ramp is being cut, and AI is doing the substituting. The new hires that arrive in FY27 will meet a tighter version of the job, with less margin to learn in someone else's slack.