The Billable Hour Breaks: AI Rewrites India's $315 Billion IT Outsourcing Contracts

For three decades, India’s IT services industry ran on a beautifully simple formula: headcount times billable hours. TCS, Infosys, Wipro, HCLTech and Cognizant grew into global giants by hiring armies of engineers and renting them out by the hour. A Reuters investigation published August 21, 2026 confirms that AI has now broken that formula — not in some hypothetical future, but in the contracts being signed this quarter. Clients are demanding identical work for 25% to 30% less, fees are shifting from hours worked to performance outcomes, and work is disappearing entirely as customers use AI to pull tasks in-house.

“Artificial intelligence promised to disrupt India’s IT industry and it is delivering,” the report opens. The numbers backing that sentence are stark: India’s IT industry generates roughly $315 billion in annual revenue, and the Nifty IT index has tumbled by a fifth this year, with its ten constituents losing a combined $73 billion in market value.

It’s a desperate market — for the vendors

The power balance in outsourcing negotiations has inverted. “It’s a desperate market for the service providers. The odds are very much in favour of clients,” Jimit Arora, CEO of research and advisory firm Everest Group, told Reuters.

Persistent Systems CEO Sandeep Kalra quantified the squeeze: the IT provider’s clients are demanding the same work for 25% to 30% less while expecting faster delivery and higher productivity. That is not a negotiation over next year’s rate card — it is a repricing of what a unit of software work is worth, driven by the client-side assumption that AI coding tools have made each engineer dramatically more productive.

The repricing is rational from the buyer’s perspective. If an AI-assisted team can deliver in three weeks what used to take eight, the vendor’s hourly model suddenly looks like price-gouging. Clients know the productivity gains exist; they are simply refusing to let vendors keep them.

The outcome-based shift, in concrete numbers

The most consequential structural change is how contracts are priced. Instead of paying for 500 developers over three years, clients increasingly pay for defined outcomes: a process automated, a cost reduction achieved, a service-level target hit.

TCS Chief Executive K. Krithivasan told Reuters that about 80% of the company’s contracts within its finance, human resources and other business services segment are now based on outcome performance measures. A person with knowledge of the matter told Reuters that figure represents a doubling since AI went mainstream in late 2023. TCS declined further comment.

Two other deals, revealed by sources to Reuters, show how aggressive the new structures have become:

• Cognizant and Daimler Truck struck an AI and automation deal in February in which AI-related cost savings are split between the vendor and the client. “With AI, the fundamentals are shifting,” Cognizant said in a statement. “Clients now expect more value and measurable outcomes, and we are re-forging our model for that reality.”
• HCLTech and German utility E.ON structured a multiyear cloud management deal, forged in June 2025, so that HCLTech will not be paid for the first year at all — payments from the second year onward are tied to efficiency gains and specific business outcomes. E.ON declined to comment and HCLTech did not respond.

An HCLTech deal where the vendor works free for twelve months and gets paid only if it delivers measurable efficiency is a different industry than the one that minted India’s IT billionaires. It transfers execution risk from client to vendor, and it makes AI productivity the direct currency of revenue.

The headcount moat has evaporated

Perhaps the most quietly devastating line in the Reuters report concerns the industry’s oldest selling point. Where once the big IT companies won contracts because they could point to their huge employee bases, that advantage has shrunk as AI automates more tasks — leveling the playing field for smaller rivals that have jumped at the opportunity to snatch business.

A million-person bench was a moat when work scaled with people. When work scales with models and compute, the moat becomes a cost center — and mid-tier firms like Persistent are winning larger deals than before because clients no longer discount them for being small. Industry executives also told Reuters they are losing some work entirely as customers use AI to shift tasks in-house, and that AI-era uncertainty is pushing clients toward shorter contracts.

The report also contains a warning about vendor behavior in a buyer’s market: some firms are said to be making irrational decisions to please clients — presumably accepting outcome terms they cannot profitably meet, just to keep revenue flowing. In a transition period where every vendor is afraid of being the one who lost the account, underpricing risk is the classic failure mode.

Why this matters beyond India

India’s outsourcing sector is the world’s largest natural experiment in what happens to a labor-arbitrage business when the arbitrage collapses. Several dynamics deserve attention from anyone building or investing in AI-era services:

Productivity gains are being competed away — to buyers. The great question of the AI boom has been who captures the value of AI-driven productivity. For commoditized services, the Reuters reporting gives a clear answer: the client does, through 25-30% price cuts, gain-sharing clauses, and first-year-free structures. Vendors only keep value where they hold genuine differentiation.

Outcome pricing converts services into a product risk business. Getting paid per outcome means the vendor now underwrites the risk that AI performance improves, plateaus, or regresses. Firms with strong internal AI tooling and measurement can price this rationally; firms signing outcome terms out of desperation — the “irrational decisions” Reuters cites — are writing options they don’t understand.

Concentration risk cuts both ways. S&P Global noted in July that the disruption will hit unevenly, forecasting just 2-4% revenue growth for Infosys, HCLTech and Wipro through FY28, while better-positioned firms and AI-specialist mid-caps grow faster. The market has already rendered its verdict on the old model: $73 billion in market value erased from ten companies in under a year.

In-housing is the silent killer. Losing a price negotiation hurts; losing the work entirely because the client’s own engineering org, augmented by AI agents, can now do it — that is existential. Reuters reported in May that global firms are already moving more work into their own India-based captives for exactly this reason.

What to watch

The pace of the outcome-based shift is the leading indicator. If 80% of TCS’s business-services contracts are outcome-based today after doubling in two years, the legacy time-and-materials book is being repriced on a two-to-three-year renewal cycle. Watch the next two quarters of TCS, Infosys and HCLTech margins: outcome contracts signed in 2025-2026 under “irrational” terms will show up in profitability as they scale.

Watch also whether the mid-tier winners — Persistent, Coforge, KPIT and peers — can sustain their momentum, or whether they have simply front-run the same margin compression. Persistent’s experience is the test case: clients demanding 30% less on one side, larger deal sizes on the other.

The billable hour served India’s IT industry for thirty years. Its replacement — payment only for measured outcomes, at prices set by the assumption of AI productivity — is now the industry’s defining reality. The companies that survive the transition will be those that treat AI not as a feature of their pitch decks, but as the engine that makes outcome-priced contracts profitable at all.