Picture this: a company announces AI-driven layoffs and frames it as a bold move. Nine months later, it quietly calls the same people back. Sometimes at better salaries. Sometimes with better titles. Almost always at greater cost than before.
If you work in HR or have followed the 2025 and 2026 layoff cycle, you have probably seen this pattern more than once. It is not a fringe story. It has become one of the defining trends of this AI adoption era. The data behind it is significant enough that anyone making workforce decisions should pay attention.
This is the AI boomerang effect. And in 2026, it has moved well past the category of anecdote.
The Numbers Behind the Reversal
The data here is uncomfortable for anyone who approved AI-driven headcount reductions without doing the deeper work first.
Forrester's global Predictions 2026: Future of Work report estimates that 55% of employers who made AI-driven workforce cuts now regret those decisions. That is a majority, and it comes from research covering organisations across multiple markets globally. Separately, a US-focused survey by Careerminds in February 2026 covered 600 American HR professionals who had laid off employees in the prior year to make way for AI. It found that two-thirds of those organisations had already rehired some of the people they let go. Roughly 36% had brought back more than half the roles they eliminated.
When those US-based HR leaders were asked what went wrong, the answers were consistent. Forty percent said AI could not replace the institutional knowledge those employees carried. Thirty-eight percent said they underestimated how much human quality control their operations needed. And 35% said productivity gains from AI were simply disappointing: lower than projected, slower to arrive, and in some cases negative once hidden costs were factored in.
Forrester has since gone further. The firm predicts that half of all AI-driven layoffs globally will be reversed in some form before the end of 2026. That is a systemic miscalculation playing out at scale.
What Is Happening in India
India's technology sector gives us the clearest window into how this plays out on the ground.
The top five Indian IT services companies, TCS, Infosys, Wipro, HCLTech, and Tech Mahindra, cut a combined 6,981 jobs in FY26. TCS alone reduced its headcount by 23,460 as it repositioned itself around an AI-first delivery model. The framing was strategic: deploying more AI capacity into service workflows meant fewer people were needed for the same work.
What followed was more complicated. Infosys reduced headcount in Q4 FY26, yet still targeted 20,000 fresher hires for the full year. The broader Indian IT industry grew its total workforce during the same period, adding 1.4 lakh employees to reach 59 lakh, according to NASSCOM data.
The headline story of AI replacing humans is too simple. AI replaced certain kinds of workers while creating urgent demand for different kinds. Many of the individuals who were let go had exactly the skills and institutional knowledge those new roles required. Companies are learning this the expensive way.
What Companies Actually Got Wrong
The AI boomerang effect is a story about what makes human work harder to replace than it looks on a spreadsheet. Three things keep showing up in the research.
Institutional knowledge does not live in any system. The 40% of US employers who cited this as their primary regret found out the hard way. Experienced employees carry knowledge that is relational, contextual, and historical. The account manager who knows a client gets difficult every Q3 because of their own budget cycles adjusts without being asked. The operations manager who knows which supplier relationships need hands-on attention handles that without a prompt. The HR business partner who understands unspoken team dynamics acts on it daily. None of that lives in a database. When those people leave, that knowledge leaves with them.
Human quality control does not scale the way task automation does. Many organisations found that AI outputs still require human review. The volume of review required actually increases as AI takes on more work, because the consequences of unreviewed errors compound. Customer support teams, content reviewers, and compliance checkers who were part of the first wave of AI layoffs are now being rehired to sit downstream of the very systems that replaced them.
Connective tissue roles cannot be automated. Mid-level managers, customer success directors, and other connective tissue roles, meaning positions built on emotional intelligence and cross-departmental negotiation, are showing the highest boomerang rehire rates. These roles require nuanced situational judgment that current AI models cannot reliably replicate.
The Real Cost of Getting This Wrong
The financial case for slowing down before you cut is clear. Research from this period finds that one-third of employers globally who made AI-driven cuts spent more on restaffing than they saved through the layoffs. That figure alone should give any CHRO pause.
The cost escalation happens through three mechanisms. First, recruitment: sourcing, screening, and hiring the same roles again carries search fees, recruiter time, and extended vacancy periods. Second, salary premiums: employees who secured other jobs since leaving negotiate harder the second time. In India's current market, the rehire salary premium commonly runs between 15% and 30% above pre-layoff compensation. Third, the productivity ramp: even a returning employee takes time to reactivate institutional knowledge and rebuild working relationships.
SHRM research estimates the average cost of replacing an employee in the United States is approximately $4,700. For senior technical or leadership roles, that number rises significantly. In India's competitive market for experienced professionals, the lower end of that range is increasingly hard to find.
Put all of that together, and one-third of organisations spending more on restaffing than they saved stops being surprising. It becomes predictable.
The Employer Brand Damage That Is Harder to Measure
Beyond direct financial cost, there is an employer brand consequence that gets underestimated until it is already doing damage.
In India's professional networks, which are dense and increasingly vocal on platforms like LinkedIn, the story of a company that laid off staff for AI and then had to rehire spreads fast. For current employees, the message is clear: the organisation made a major decision that affected their colleagues, got it wrong, and had to walk it back. That does not build loyalty or trust.
For external candidates, it signals strategic instability. Companies with a reputation for impulsive workforce decisions pay a recruitment penalty in exactly the hiring cycles where the boomerang data shows they most need to hire urgently.
In the United States, where candidates routinely check Glassdoor before accepting offers, a poorly handled layoff can increase recruiting costs for years. The reputational impact in India tends to be felt faster and more personally, given how quickly information moves through professional alumni communities.
What HR Should Actually Do Differently
The lesson here is not that AI-driven workforce change is wrong. The lesson is that organisations which treat workforce restructuring as a pure financial exercise, without understanding what human judgment contributes and how AI actually performs in production, will consistently underestimate what their decisions cost.
Do the role analysis before the announcement, not after. Before eliminating a role on the basis of AI substitution, map the full scope of what that role actually does. Not just the tasks in the job description, but the judgment calls, relationships, contextual knowledge, and informal coordination that experienced incumbents handle daily. A process map of visible tasks will almost always understate what the role actually contributes.
Pilot before you scale. The organisations least likely to experience boomerang regret are those that piloted AI in a limited context first. They measured real-world performance against human baselines and made workforce decisions based on observed outcomes rather than vendor projections. India's BFSI and IT sectors, where AI deployment has been most aggressive, would benefit most from formalising this approach. It is slower up front. It is far cheaper overall.
Build for the Long Term
Build a proper alumni network and rehire path. Given how common boomerang hiring has become, organisations that treat departures professionally, with clear offboarding and maintained alumni relationships, are far better positioned to access that talent again when the calculus shifts. An ad hoc "would you consider coming back?" call is a much weaker offer than a proactive alumni engagement programme.
Change the workforce planning question entirely. For every role in scope for AI displacement, the question should not be "will AI do this?" It should be: what remains after AI handles the automatable portions, what will the oversight requirements look like, and who has the skills and judgment to manage that residual work? That question leads to different answers than the binary replacement logic that has driven much of the 2025 and 2026 AI layoff wave.
Conclusion
The AI boomerang effect is an expensive lesson in the difference between automating tasks and replacing judgment. India's organisations, many of which are at an inflection point in their AI integration strategies, have a real opportunity to learn from the first wave of AI-driven layoff reversals without having to live through them directly.
The companies that will look back on 2026 as a moment of strategic clarity are not the ones that moved fastest on AI-driven headcount reduction. They are the ones that moved most carefully, asked the harder questions before the cuts rather than after, and preserved the human infrastructure that proved irreplaceable once it was gone.
Fourteen years in HR has taught me that the most expensive workforce decisions are almost never the ones that seem risky at the time. They are the ones that seemed obvious.