India’s largest IT services companies are beginning to report meaningful revenue from artificial intelligence. TCS says its annualised AI revenue crossed $3.1 billion in the September quarter, while Infosys reported AI revenue of about $539 million, more than 10% of its overall revenue. HCLTech has also reported strong growth in its advanced AI business.
On the surface, the message is straightforward: AI is becoming a commercial growth engine for India’s technology services industry. But a less comfortable metric tells a different story. Revenue per employee across the country’s largest IT services companies has remained broadly flat for five consecutive quarters, according to analysis reported by Moneycontrol on October 29.
This is the productivity paradox now confronting the industry. AI is generating new revenue and helping companies deliver work more efficiently, but those gains are not yet showing up as a decisive improvement in the amount of revenue generated by each employee.
The AI business is real, but the productivity story is incomplete
The latest quarterly disclosures show that AI is no longer confined to innovation labs and proof-of-concept projects. TCS reported annualised AI revenue of $3.1 billion in Q2 FY27, equivalent to more than 10% of revenue. The company said this business grew 19.2% sequentially and carried margins above its average. Infosys said AI revenue grew 29% sequentially to approximately $539 million, also exceeding 10% of revenue. HCLTech reported annualised advanced AI revenue of $876 million, up 72% year on year in constant currency.
More Read
These figures indicate growing customer demand for AI-led transformation, integration and implementation. Enterprises need help connecting models to data, modernising applications, redesigning workflows and deploying AI systems inside complex operating environments. Indian IT services firms are well placed to sell that work because they already manage large technology estates and long-standing enterprise relationships.
However, AI revenue is not the same as productivity improvement. A company can generate more revenue from AI projects while continuing to employ large teams across legacy services, application maintenance, infrastructure and business processes. The composition of revenue may change before the underlying delivery model does.
Why the gains are not flowing straight to the bottom line
One explanation is that enterprise customers are negotiating for the efficiency gains themselves. If AI helps a provider complete a task in fewer hours, the client may expect a lower price, faster delivery or a higher volume of work for the same budget. The supplier may therefore deliver more output without retaining all the economic benefit.
This is particularly important in a sector that has historically relied on billing models linked to people, time and large delivery teams. AI weakens the relationship between effort and output. A provider that charges by the hour may find that the very technology improving its delivery economics also reduces the number of billable hours.
More Read
The industry is trying to move towards outcome-based contracts, platform-led services and higher-value transformation work. But these models require different capabilities: reusable intellectual property, domain expertise, stronger solution architecture, risk-sharing and the ability to demonstrate business results. They cannot be adopted across an entire portfolio overnight.
There is also a measurement problem. Revenue per employee is a useful indicator, but it does not capture every change in cost structure, subcontracting, wage mix or business composition. Moneycontrol’s report cited staffing-industry analysis showing that the collective manpower cost-to-revenue ratio of the top five firms has also remained largely stagnant over several fiscal years. That suggests the efficiency challenge is not solved simply by reducing headcount or deploying automation.
The transition is harder than adding AI to existing services
For an IT services company, the most immediate opportunity is to use AI to accelerate existing work: code generation, testing, documentation, service desks, application modernisation and knowledge retrieval. These use cases can improve delivery speed, but their commercial value depends on how contracts are priced and how much of the savings the provider can retain.
The larger opportunity is to redesign the work itself. Instead of using AI to make a conventional software maintenance team marginally faster, a provider could create a more automated service with a smaller human team supervising exceptions, quality and business outcomes. That is a more fundamental change, but it requires clients to trust a new delivery model and providers to accept different revenue economics.
It also changes the skills equation. Demand shifts towards AI engineering, data architecture, security, domain specialists and people who can integrate AI into live enterprise processes. Reskilling large workforces while maintaining existing services is a complex operational task. The transition can create short-term costs before it produces durable productivity gains.
The market is rewarding AI revenue, but asking a harder question
Investors have reasons to welcome the growth in AI services. It creates new addressable work, supports premium offerings and gives Indian IT companies a role in enterprise AI adoption. But the market will eventually look beyond AI revenue disclosures and ask whether these businesses can grow profitably without relying on proportional increases in headcount.
TCS’s Q2 results, for example, reported annualised AI revenue above $3 billion while its total workforce stood at 598,056. Infosys reported strong sequential AI revenue growth, but also revised its FY27 revenue growth guidance to 1%–2%. These figures do not invalidate the AI opportunity. They show that a growing AI business can coexist with modest overall growth and a workforce model still in transition.
For customers, the question is whether AI projects produce measurable improvements in cost, speed, accuracy and customer experience. For service providers, it is whether they can capture a fair share of those gains. For employees, it is whether the industry can create enough higher-value work to offset changes in traditional delivery roles.
What to watch next
Three measures will reveal whether the productivity story is changing. The first is revenue per employee, tracked over several quarters rather than inferred from a single earnings call. The second is the share of AI work delivered through repeatable platforms and outcome-linked contracts, rather than bespoke implementation alone. The third is whether operating margins improve sustainably without depending mainly on wage restraint, lower hiring or one-off cost actions.
AI is already a meaningful business for India’s IT services leaders. The harder task is converting that business into a structurally different delivery model. Until productivity gains appear in unit economics, the industry’s AI transformation will remain commercially promising but operationally unfinished.
The next phase of competition will not be decided by who announces the largest AI revenue number. It will be decided by who learns to deliver more value with a fundamentally different relationship between people, software and revenue.
