Introduction

The India IT services business has been running on a pretty neat, very predictable model for 30 years: hire from a large pool of engineers, take advantage of the wage differential, and increase revenues in tandem with the number of employees. But this traditional linear model is being challenged by a powerful, structurally disruptive force: generative artificial intelligence (AI).

The labor pyramid is falling apart as AI takes over repetitive coding, testing, and maintenance tasks. This change, along with the competition from in-house corporate tech hubs (Global Capability Centers, or GCCs) and global macroeconomic friction, is driving an unprecedented transition. The slowdown in hiring and in growth of large-cap stocks suggests short-term disruption, but the shift toward proprietary platforms, niche consulting and big bets on infrastructure suggests that AI can help grow the sector’s addressable market.

The Dual-Track Landscape

To appreciate this disruption, one needs to understand the size of India’s IT export engine. It is worth 26 lakh crore, accounts for 7% of GDP and has more than 50 lakh professionals working in it. The industry earned 20 lakh crore in export revenues in FY25, mainly for the US, UK and Europe. While the sector could reach 35 lakh crore by 2030, its growth path is no longer linear.

The market is now in a state of extreme bifurcation. Large-caps such as Tata Consultancy Services (TCS), Infosys and Wipro are experiencing slowing growth and conservative FY27 outlook. On the other hand, agile mid-cap companies such as Coforge and Persistent Systems are seeing strong double-digit revenue growth. Mid-caps are more agile and can quickly pivot to meet new AI spending opportunities on smaller bases. Clients are increasingly turning to mid-caps as Tier-1 giants have at times been slow to invest in specialization and transformation.

Deflation in Sourcing and the Strategic Response

Traditional profit pools are under threat from three formidable challenges. The first risk is deflationary risk, which is the risk of AI. The classic model was based on effort-based, per-hour billing, with a pyramid-shaped workforce of junior engineers at the bottom doing repetitive support, testing, and maintenance. Today, agentic AI automates these tasks in hours, breaking the headcount-revenue correlation.

Clients expect these productivity benefits to be passed through. During the Q1 FY27 earnings call, Infosys mentioned that clients were asking for discounts at the time of renewal and mid-project. TCS is passing 10% to 15% productivity gains back, which is reducing revenue realization. Moreover, multi-year contracts are being replaced by short-term, non-recurring, and volatile AI projects. For example, TCS’ AI revenue fell from $125 million in Q4 FY26 to $75 million in Q1 FY27.

To protect margins, major players cut headcounts. The aggregate headcount of the top five IT companies decreased net-net by 6,981 in FY26, after an increase of 12,078 in FY25. TCS reported the highest number of job cuts with 23,460, followed by Tech Mahindra with 118. Revenue per employee increased by 3% to 4% in FY26, but analysts are unsure if this is due to AI leverage or a short-term bump from high utilization and subcontractors.

Second, there is a threat from Global Capability Centers (GCCs). Foreign clients are setting up in-house Indian hubs instead of outsourcing. Over 1,700 GCCs employ 19 lakh people in India. Vanguard, for example, set up a Hyderabad hub in November 2025 to build full-stack teams, bypassing partners. Because GCCs transition entry-level tasks in-house, vendors experience revenue lags; LTIMindtree (LTM) blamed its recent decline on aerospace and automotive clients ramping up GCCs. On top of that, GCCs earn 20% to 40% more than their counterparts, further fueling the talent war.

Third, geopolitical factors such as the Middle East conflict. Regional conflict has led to a slowdown in spending by clients in consumer, airline and financial services, as companies seek to diversify away from the US, where 60% of revenue comes from. The instability has also caused delays in the delivery of essential hardware and memory chips needed for AI models, which has hindered deployments.

Indian IT vendors are taking a multi-pronged defense, however. First, they are aggressively developing their own AI platforms and IP. In the first year, 17 implementations have been made in the LTMs “Blueverse” ecosystem, and Persistent Systems is using “SASVA” and Coforge is using “Neuron”. The takeaway: It’s not about creating raw LLMs, it’s about embedding AI into complex legacy environments.

Second, companies are transitioning from a labor pyramid to a “diamond” model, with junior positions becoming automated and engineers being upskilled to use AI tools. Infosys has trained 90% of its developers in AI, and TCS has 2.7 lakh associates trained in AI/ML. Firms are also hiring AI-skilled graduates, including TCS (14,000), Infosys (4,000), and LTM (1,308) in Q1 FY27.

Third, companies are leaving low-margin work for high-value consulting. Coforge has abandoned a 15 million government project to focus on AI-native contracts, while Infosys and Tech Mahindra are steering clear of large contracts with low value.

Fourth, there is a growing trend of aggressive mergers and acquisitions (M&A) that are driving domain specialization. In Q3 FY26, Infosys acquired silicon design specialist InSemi and automotive engineering firm InTech. HCL Tech bought Jaspersoft and Bobbie for data governance, Wipro bought Harman DTS, and Tech Mahindra bought Avant Techno Solutions.

Last but not least, companies are spending a lot on physical infrastructure. HCL Tech is investing 3,500 crore in AI data centers, while TCS plans to build 1 GW of AI capacity by 2032, acquiring land in Pune and Visakhapatnam for hyperscale hubs.

The Future Ahead

The competitive field is growing. Large language model providers such as OpenAI and Anthropic are moving into enterprise implementation and consulting, directly challenging IT providers. OpenAI’s $4 billion investment in deployment and Anthropic’s Wall Street-backed consulting arm suggest a structural problem with traditional outsourcing.

By 2030, however, NASSCOM estimates that AI adoption will create an incremental $300 billion Total Addressable Market (TAM). This shift is similar to the cloud computing revolution of the 2010s. Next, critics said cloud would kill outsourcing, and it was outsourcing that fueled cloud. To achieve this, Indian IT needs to train its employees beyond coding and prompt engineering to system design, design thinking, and scalable system architectures.

Conclusion

Generative AI is an existential shift for Indian IT, moving away from the low-cost, effort-based billing model that helped it get where it is today. However, Indian IT is actively redefining its value proposition by reshaping workforces into a technology diamond, investing in proprietary platforms, and creating physical data infrastructure.

Short-term revenue fluctuations and job cuts should not be interpreted as terminal decay for global investors. Rather, they are the painful birth pains of a sector that is moving from selling human hours to selling sophisticated, platform-based AI integration. The final winners will be those who can integrate intelligence into global enterprise without any hassle.