The breakneck pace of artificial intelligence-led productivity gains is unsettling India’s traditional IT service industry, but market observers say fears of a structural collapse may be overstated.

Stocks of IT behemoths followed a global rout in tech stocks, with the key catalysts being US firm Anthropic pushing AI into executing structured professional tasks and Palantir highlighting in its earnings call how it is upending pay-per-seat software as well as third-party software with its own AI offerings.

While automation and agentic AI are reducing billing hours, analysts argue the shift is more of a transition than a tectonic break — at least in the near term.

The immediate concern of the software industry, long pilloried for under-investing in research and development in AI, is the potential compression of billable hours, the bedrock of the IT services model. As AI tools take on coding, testing and debugging tasks, the manpower intensity of projects is declining.

“In the past, if a software testing project required 100 engineers for six months, the company billed for those hours. Today, using agentic AI (AI that can autonomously code and debug), that project might only require 10 engineers and be finished in two months,” a senior executive of an IT services firm explained.

Even so, analysts say AI-led business modernisation is likely to play out as an evolutionary cycle. Drawing parallels with the cloud migration wave of 2016-2018, they contend that while parts of legacy revenue may face pressure, a broader and more durable opportunity could emerge once the current infrastructure-heavy phase stabilises. In an attempt to play catch-up, firms such as Infosys are now tying up with Anthropic to offer AI-led solutions.

Sandeep Gogia, sector lead – tech and digital at Equirus Capital, said the role of IT service providers (ISPs) and system integrators (SIs) such as TCS, Infosys and Wipro will remain central in the AI-led enterprise pivot, albeit in a different form.

He said enterprises will continue to run mission-critical applications on mainframes and other platforms that must be modernised and tightly integrated with AI layers. SIs are uniquely positioned to drive this integration. Their understanding of client guardrails, databases, applications and sector-specific compliance gives them an edge in linking legacy estates with AI and agentic AI.

“Impact of (revenue) cannibalisation could be considerably low in this phase, which is likely to last at least more than 1-2 years,” he said.

Abhishek Pathak, research analyst at Motilal Oswal Financial Services, compared the current AI cycle with the early cloud build-out period, when cloud adoption initially hurt traditional service lines before spawning new revenue pools.

“But as deployment scales, entirely new categories of spending will emerge,” he said, adding that the long-term effect, much like the cloud era, is likely to be positive.