For nearly three decades, the pitch that built Indian IT was simple enough to draw on a napkin. Take work that costs $150 an hour in New York. Route it to engineers in Bengaluru at a fraction of that. Keep the spread. Scale by hiring more people. Repeat. The model was so durable, so elegantly scalable, that it created a $250 billion export industry and made TCS, Infosys, Wipro, HCL, and Tech Mahindra among the most valuable companies India has ever produced. And FY26 was the year the people running those companies publicly admitted — on earnings calls, in CFO language, with numbers attached — that the model is changing shape permanently.
FRACTURE ONE: HEADCOUNT AND REVENUE COME APART
The most structurally significant thing that happened in Indian IT in FY26 was not a deal win or a margin improvement. It was a decoupling. For the first time in the industry's history, revenue grew at multiple large-cap firms while headcount did not.
HCL's CEO C. Vijayakumar put numbers to it directly: revenue has grown 4–5% over two years while headcount has not grown at all — with at least a 1–1.5 percentage point gap between the two in recent quarters. Mphasis called it "de-linkage between revenue growth and headcount growth." The industry has a word for this now: non-linearity. What it means in plain terms is that the old equation — one more engineer billed one more set of hours — no longer governs how revenue is made.
The uncomfortable corollary is what Vijayakumar called people being "released" due to productivity improvements — noting that not all of them are "readily redeployable" because the entry-level work they used to do is being absorbed by automation. The Indian IT majors employ a combined two million people. Small percentage shifts in redeployability translate to large absolute numbers of careers needing to find a new direction.
FRACTURE TWO: SAME TECHNOLOGY, TWO OPPOSITE MARGIN STORIES
Here is the strangest thing about AI's impact on Indian IT pricing: the same productivity gain shows up on one CFO's slide as a discount the company is forced to give clients, and on another's as a premium the company can now charge. Infosys's own CFO articulated both views on the same earnings call — first noting that pricing has actually firmed up and most of FY26's growth was pricing-led, then acknowledging that competitive intensity means productivity gains will largely be passed back to clients.
HCL went furthest, openly telling clients: let us use AI tools on your contracts and we will find savings — savings that will mean lower revenue for us. By Q4, Vijayakumar had put a rough number on the structural drag: a $100 million deal in the old model now closes at closer to $80 million. He estimated that 40% of the industry runs the risk of being disrupted by AI, potentially shrinking at 3–5% annually for several years, and that for HCL's specific portfolio this translates to a 2–3% deflation headwind.
To defend pricing elsewhere, both HCL and Tech Mahindra began doing something the industry almost never did previously: walking away from deals. HCL disclosed it had voluntarily foregone at least $1 billion in TCV during FY26 by declining contracts that did not make financial sense. Tech Mahindra called it "deal discipline." In an industry that measured quarterly momentum by TCV growth for two decades, this is a meaningful signal.
FRACTURE THREE: THE BILLING UNIT IS CHANGING
If hours are no longer the unit, what replaces them? Everyone is auditioning a candidate.
Tech Mahindra's CEO Mohit Joshi offered the sharpest reframe — borrowing language directly from AI pricing: "service tokens." A defined unit of business outcome, purchased regardless of whether it is delivered by a human, an AI agent, or some blend of both. The client buys the output. The billing rate disappears from the conversation entirely.
Mphasis has gone to market with a different but related model — selling the savings rather than the hours, and demonstrating value in live sandbox environments rather than through proposal documents. In Nitin Rakesh's words, "RFPs are turning into hackathons." The salesforce that wins is the one with working AI agents in the room, not slideware in the back office.
Infosys's CFO acknowledged the transition without committing to its destination — noting that outcome-based, pod-based, and studio-based pricing models are all emerging, but that the full shift will take more than a year.
FRACTURE FOUR: FROM LABOUR TO PLATFORMS
The strategic consensus that emerged across every large-cap management team in FY26 was the same: stop being a pure services firm, start owning intellectual property and platforms embedded in client environments. HCL framed it most clearly — the industry must evolve from labour-based to people-plus-IP-plus-platform. HCL's AI Force reached $155 million in quarterly Advanced AI revenue by year-end, up from a $100 million annual milestone just two quarters earlier.
Wipro launched a dedicated "AI-native business and platforms unit" with separate leadership and a distinct operating model — describing it as a "dual engine" running traditional services on one side and AI-native platforms on the other. Tech Mahindra renamed entire service lines from "application development and maintenance" to "agentic development and modernisation" — calling it not a name change but a fundamental shift in how value is delivered.
TCS made the boldest structural move: announcing a 1 GW data centre build through its HyperVault business. One gigawatt of capacity, for context, is roughly equal to India's entire existing colocation market. A services company is now building serious infrastructure — using it as the foundation for a full-stack AI services ambition that runs from data centre compute all the way up to model operations and fine-tuning.
FRACTURE FIVE: THE NEW GROWTH POCKET
The most underappreciated shift in FY26 was who the growth clients actually are. The fastest-moving new work is not enterprises adopting AI tools. It is the AI companies themselves — the hyperscalers, model developers, and GPU infrastructure builders — who are buying Indian IT services to build, run, and refine the systems that will eventually automate parts of the industry.
Wipro disclosed a flagship engagement running and improving the frontier AI models of a global technology company — end-to-end model operations, training, governance, and domain validation. Tech Mahindra's COO noted that close to one-tenth of its Business Process Services revenue now comes from helping technology companies train and fine-tune AI models. BPS — historically the lowest-margin, most automatable layer of the Indian IT stack — is now, paradoxically, the layer that most requires humans.
HCL has put the clearest framework on what the portfolio looks like going forward: three buckets, three growth trajectories. AI-disrupted services shrink. AI-amplified services grow in line with the business. AI-native services are targeted to grow at 25–30%. The central question for FY27 is whether the third bucket scales fast enough to offset the drag from the first.
THE VERDICT
The Indian IT industry spent FY26 publicly admitting that its core business is changing shape — and attaching numbers, language, and organisational decisions to that admission. None of the major players have completed the transition. All of them have started it. The companies that once sold engineering hours by the thousand are now building platforms, owning IP, and billing for outcomes.
The math of that transition becomes legible in FY27. When the new revenue mix is large enough to either validate the bet or expose the gap, investors will have something concrete to evaluate. For now, the most important thing to understand is that this is not a cyclical slowdown. It is a structural rewrite — and the companies doing it to themselves have a better chance of surviving it than those waiting to be done to.
