Nirmal Bang Institutional Equities has cut its view on the Indian IT sector, saying generative AI is starting to weaken the labour-arbitrage model that built the industry and bring revenue and margin pressure forward. The brokerage has retained an 'underweight' rating on the sector for FY27, saying the business model transition is arriving sooner than it had expected.

Nirmal Bang said the productivity gains from GenAI are largely bypassing IT services firms and moving to model makers instead. It said developer throughput is rising 5.8-8 times, while clients are already building about 20 per cent compression into budgets. That, it said, is turning billable engineer-hour margins into compute spend and could reduce IT services' share of global IT spend to 27.2 per cent in CY28 from 30.9 per cent in CY25.

Why Nirmal Bang sees pressure
Nirmal Bang said Indian IT's core strength, people-led labour arbitrage, is now becoming its biggest liability as GenAI cuts the cost of the software development engineer hour itself. The brokerage said the industry is increasingly accepting compression in software development life-cycle timelines, effort and resources.

It added that renewals are likely to be the first point of pressure as contract sizes shrink, small-ticket competition squeezes margins, and large caps have more ground to defend, leaving IT services companies as a low-margin orchestration layer. Nirmal Bang also said this capex cycle is unlike 2019. While the earlier cycle expanded the software development life cycle, the current one is rewriting it.

As a result, assumptions of a volume-led recovery may not hold, and incremental demand from legacy modernisation is more likely to come through recycled savings than fresh TCV. It added that terminal value may still lie in a leaner, AI-native services model, citing NVIDIA's framing of intelligence at about 50 per cent of SDE cost.

Two cases for 2026
Nirmal Bang outlined two risks for 2026. In the first case, faster AI adoption could compress effort-based billing as generative coding and testing reduce incremental manpower needs, especially in software engineering. Enterprises, it said, are likely to push harder on pricing and outcomes, while productivity gains get passed through to clients. The report also said AI is unlikely to replace SaaS altogether in the near term, but uncertainty could lead to more furloughs and a steady shift in spending towards intelligence, estimated at 4.2 per cent of global IT spend in a CY28 bull case.

In the second case, if AI adoption slows or fails to deliver measurable returns in CY26, global technology budgets could tighten sharply, according to Nirmal Bang. It flagged the capital being deployed by hyperscalers into AI infrastructure, estimated at$609 billion in FY27E and increasingly supported by debt, as an area already facing scrutiny that could trigger a broader correction in technology sentiment.

Where the gains are going
The report said the GenAI productivity windfall is being redirected to frontier labs such as OpenAI, Anthropic and Google. Each unit of code that once generated billable engineer hours now also creates a token bill payable to model companies, it said. With token consumption per workflow rising and frontier output pricing climbing to $25-30 per million output tokens, the profit pool is shifting away from the systems integrator layer to the model layer.

Nirmal Bang added that agentic AI implementation could raise demand for Forward Deployed Engineers, but adoption may lag because of hallucinations, higher per-token costs, evolving legal frameworks and fragmented data infrastructure.

New target prices for IT stocks
Overall, Nirmal Bang said GenAI is creating two-sided pressure for Indian IT: faster adoption can compress services demand, while weaker adoption can hurt technology budgets. Either way, it sees a more front-loaded transition, with pressure on topline growth and margins arriving earlier than previously expected. It sees up to 25 per cent correction in the IT counters.

Nirmal Bang has downgraded TCS (Target Price: 1,693), Infosys Ltd (Target Price: 1,051), HCL Technologies Ltd (Target Price: 871), Wipro Ltd (Target Price: 152), Tech Mahindra Ltd (Target Price: 1,120) and LTM (Target Price: 3,185) to 'Sell' rating. It also downgraded Coforge Ltd (Target Price: Rs 1,395) and Persistent Systems (Target Price: 5,315) but maintained 'Hold' on Mphasis Ltd (Target Price: 2,347).