Infosys, KPIT Tech, Mphasis, Coforge, LTIM: Target prices for 17 IT stocks

Equirus Securities on Tuesday said the recent 12-18 per cent decline in the NSE IT index and top six large-cap IT stocks since February 3, 2026 reflected investor concerns over AI-led disruption, but maintained that the role of IT services and systems integrators is unlikely to vanish despite rapid advancements by LLM and born-AI platform vendors such as Anthropic and Palantir. 

The domestic brokerage assessed the impact of AI adoption by enterprises, explained why it believed the correction was driven by extreme pessimism, and outlined its preferred stock picks.

Equirus Securities said the recent correction reflected extreme pessimism driven largely by cannibalisation worries and concerns over terminal growth rates and price-to-earnings multiples, while overlooking potential new opportunities for IT services that have historically acted as integrators in prior technology cycles.

Equirus said it preferred Infosys Ltd, Tata Consultancy Services (TCS), HCL Technologies Ltd  and Tech Mahindra Ltd among large caps. It likes Mphasis Ltd and Zensar Tech among midcaps, and KPIT within ER&D on a relative basis. It cautioned that a higher-than-expected impact from AI remains a key downside risk.

Equirus acknowledged that the pace and scale of AI-driven technological evolution is materially different and larger than in previous cycles, and added that disruption is likely to evolve rapidly, with concerns potentially resurfacing at each stage.

However, the brokerage said it believed the role of IT service providers and systems integrators will still persist, albeit in a different form. 

It argued that legacy technologies typically co-exist with new technologies, particularly given the high cost of ownership of existing systems. It also noted that systems integrators understand clients’ industry-specific governance frameworks, applications, databases, contextual knowledge, business logic and workflows, much of which resides in systems of record built on legacy and mainframe platforms. 

These systems require modernisation and tight integration with AI layers, and Equirus said IT services and systems integrators are currently best placed to execute this transition.

Equirus expects incremental demand for cloud adoption, application modernisation, creation of clean structured and unstructured data layers, and cybersecurity frameworks to enable trusted AI deployment. It added that the impact of cannibalisation is likely to remain limited in the current phase, which it expects to last at least 1-2 years, compared with the post-AI adoption phase.

The brokerage said legacy technology is likely to co-exist with AI, given the massive investments enterprises have made in IT systems and applications, particularly systems of record. 

It noted that AI layers cannot generate reliable outputs unless LLMs and Agentic platforms have deep access to enterprise data housed in complex architectures. It cited examples such as Salesforce’s tie-up with OpenAI and Anthropic to allow clients to swap LLMs and build autonomous agents through its Agentforce platform, and ServiceNow’s partnerships with Anthropic and OpenAI via its AI Control Tower platform under a multi-model strategy. 

These developments, Equirus said, suggest that large enterprises can develop AI agents faster without rebuilding data infrastructure from scratch and that bypassing SaaS providers and IT service vendors remains difficult for LLM and AI platform vendors at this stage.

Citing IDC data, Equirus noted that coding accounted for just 16 per cent of developers’ time in 2024. It said the role of ISPs and SIs is evolving towards becoming developers, managers and orchestrators of AI models, as well as providers of data and application integration and modernisation services. They are increasingly expected to architect AgenticAI ecosystems by integrating agents with systems of record and applications, managing governance, orchestrating intelligent workflows and enabling agents to operate, adapt and self-heal across legacy and AI-native environments.