In the 31st annual general of TCS meeting on June 9, N. Chandrasekaran, chairman of Tata Sons, described artificial intelligence as the biggest growth opportunity in the company’s history. He predicted that TCS will soon manage as many AI agents as it employs humans, and that the AI systems that govern them will become the next recurring revenue stream for the company.
The optimism is understandable, but the threat from AI agents is more immediate than the opportunity. The same modernization project that created the Indian IT services industry is now being automated by the hyperscalers that these companies are reselling. of Amazon AWS transformation develops AI agents that handle code analysis, dependency mapping, refactoring and test generation that TCS, Infosys, Wipro, HCLTech, Tech Mahindra and Cognizant have been charging by the hour for two decades. Most importantly, one of the largest cloud platforms offers basic upgradeability at no extra charge.
How Indian IT Built Its Income
To understand the threat, we need to look at the evolution of the Indian IT services model. It started with labor arbitrage, as Indian companies outsourced software development and maintenance at a lower cost than onshore teams. Custom application development followed as customers outsourced the construction of their systems. Then manufacturing hardened into application maintenance, where one-off projects became long-term support contracts billed year-over-year. Cloud migration and modernization programs opened up the next revenue pool, and managed services extended it. AI agents now pursue repetitive work in maintenance and modernization, where annuity, not construction, has always been the profitable half.
Inside AWS Transform
At its core, AWS Transform is an agent modernization platform designed to automate the repetitive steps of application migration. It runs specialized agents in three parts: .NET, mainframe and VMware. The platform starts by discovering the application’s assets, then analyzes the source code and dependencies, creates a migration plan, restores the application to a modern architecture, and produces the tests that validate functional parity with the original system. The output is not just conversion code, but a modernization workflow that removes much of the manual inspection, rewriting, and testing that large engineering teams once performed.
Enterprises can direct AWS Transform into a legacy estate and reduce the discovery, design, and refactoring effort that large modernization programs typically require. With AWS Transform, Amazon enables businesses to compress migrations that once took years to months, and reports that customers have pushed more than 1.1 billion lines of code through the service and saved more than 810,000 hours of manual effort. For .NET workloads, agents perform conversion up to four times faster, while eliminating Windows licenses that carried much of the operational cost. CIOs can now ask service providers to update prices based on results rather than headcount, and platform teams can hand over to agents the iterative analysis and conversion done by offshore delivery teams.
Market leaders publish the same discount. Infosys told investors it moved 3 million lines Hertz COBOL in a microservices environment using AI foundational models at 60% lower cost and 60% less timescale than a conventional migration. A seller who advertises a deliverable for 60% less is teaching any prospective buyer what that result will cost now.
AWS Transform vs The System Integrators
Indian IT majors are not standing still. TCS, Infosys and Wipro have built platforms such as MasterCraft, Topaz and Wipro Intelligence to automate parts of the modernization lifecycle and these elements continue to matter where domain knowledge, regulatory framework and customer architecture determine the outcome. The challenge is that these tools are tied to a service business model, while AWS Transform is tied to cloud consumption. Amazon can afford to make the core service free because it monetizes the destination, the AWS infrastructure where modernized workloads ultimately end up. With this, AWS turns modernization into a cloud acquisition drive rather than a service-based transformation program. The key difference lies in monetization, where hyperscalers monetize the destination while Indian IT companies monetize the journey to get there. Microsoft and Google Cloud build the same kind of agent-based modernization tools, which makes it a platform-level direction rather than a company bet.
Chandrasekaran’s strongest point is that context and trust are becoming scarce resources and that the AI systems that govern them will need constant oversight. He’s right that agent services like AWS Transform produce blueprints and refactored code, while a team still needs to handle the fraction they can’t handle, confirm parity, manage change, and connect the output to how the business works. This supervision is unlikely to become an annuity at the old scale and margin, because government AI works in hyperscalers, model labs and Microsoft within Azure AI, who are closer to it than any integrated and intend to sell it themselves.
What the change means for each side
What does this mean for corporate customers? How should CIOs view modernization contracts in the age of agents? For enterprise buyers, the shift moves modernization from effort-based pricing to outcome-based pricing and puts them in a stronger negotiating position. For Indian IT companies, the old annuity is shrinking and the defensible position lies in governance, domain specific agents, industry workflows and controlled AI functions rather than reselling agents built by hyperscalers. For hyperscalers, modernizing with resellers becomes a customer acquisition strategy that pulls workloads into their own infrastructure.
Where does the pressure occur?
The pressure is already being felt in hiring, where India’s five largest IT companies cut a combined 6,981 jobs in FY26, reversing a net addition of 12,718 a year earlier. Cognizant has flagged plans to remove 12,000 to 15,000 roles in an AI-led restructuring that is hitting its workforce hardest in India.
Revenues are also under pressure, with Wipro’s IT services annuity flat in constant currency for the year even as large deal bookings increased, and Infosys guiding FY27 growth between 1.5% and 3.5%, while naming AI’s productivity squeeze as a force working against it. Analysts model 2% to 3% annually deflation across the application service base as agents change their pricing model for application work.
The visa environment adds another layer, with the H-1B approvals for the six largest companies down about 40% year-on-year, raising the cost of putting Indian engineers in front of US customers at the same time as automation reduces the number they need.
Where this leaves Indian IT
Enterprise customers are likely to gain leverage from this transition. A vendor whose historical profit came from maintenance that is now being phased out has little reason to modernize at full speed. Businesses can use it to price these contracts based on results rather than hours, and ask what the vendor has at the level of governance that an overscaler can’t replicate. The replacement opportunity is genuine, but it is neither captive nor proven on the same scale as the alimony it purports to replace.
The way forward for Indian IT companies depends on whether they can build defensible IP governance, domain and operational layer around AI agents instead of reselling the agents sent by hyperscalers. If they can, AI will expand their market rather than erode it, and until then the safest reading is that agents shift profit to the companies providing the intelligence and away from those who have long incorporated it.
