AI may help cut legacy tech debt, but clients fear creating a new one

Conversations with IT industry clients indicate that enterprises are still not entirely sold on the idea of making AI bets, even as the broader IT services industry has begun shifting towards AI-led deal wins.

India’s IT services companies have entered FY27 with a measurable reduction in clients’ legacy technology debt. Firms such as Tech Mahindra have set ambitious targets of achieving a 30-35 per cent reduction in technical debt. Yet, clients, particularly from the manufacturing and BFSI sectors, told businessline that the industry remains cautious about inadvertently creating a new form of technical debt in the future.

“The worry is that enterprises may potentially end up creating a new tech debt. For example, in AI-assisted coding, there is the risk of accumulating tech debt because organisations may not have complete visibility into every line of code being generated. That is a little concerning,” said Vijay Balakrishnan, Chief Digital & Information Officer, Godrej Enterprises Group.

Despite modernising nearly 90 legacy applications and achieving around 60 per cent faster migration through AI-enabled software development, enterprises continue to view technical debt as an inevitable part of their technology investments.

“Technical debt follows a principle similar to the conservation law; it cannot be created or destroyed, only transformed from one form into another,” said Balakrishnan.

On the financial services side, BFSI clients continue to raise three key concerns: the probabilistic nature of AI, which limits large-scale production deployment; the lack of regulatory clarity, which restricts mass adoption; and the governance mechanisms required to safeguard against risks associated with the technology. Additionally, the token-based cost of AI usage remains a significant concern.

“A lot of banks and NBFCs are realising that while AI can accelerate digitisation, the associated compute costs, if not managed efficiently, can become a major financial burden,” said Sanjay Varma, President, Fintech Solutions Group, Aurionpro Solutions. The company advises its BFSI clients to optimise token consumption and manage AI costs carefully.

BFSI and manufacturing rank among the top four verticals for India’s large IT services companies. Amid broader macroeconomic pressures, the cautious approach adopted by these sectors is increasingly reflected in the performance outlook of IT services firms. This trend was evident in the recent earnings calls of companies such as Tata Consultancy Services (TCS) and Wipro, where management acknowledged a slowdown in discretionary technology spending.

“AI is reshaping spending priorities across organisations. From a client perspective, budgets for traditional IT services, BPO operations and support functions are increasingly coming under pressure,” said Srini Pallia, CEO and Managing Director, Wipro, during the company’s latest earnings call.

The concern is particularly relevant for BFSI and manufacturing, where a “mostly correct” answer may not be good enough. The real risk is the dependence on an AI system without clearly understanding its limitations or how easily it can be replaced, as per Sandeep Gogia, MD, Tech & Digital, Equirus Capital.

Acknowledging the tech debt concerns, Titus M, Practice Director at Everest Group, noted that AI introduces significantly more variables into technology decision-making.

“Every decision across the technology stack now has implications. Even committing to a particular AI model can create a form of technical debt. Poor data foundations, inefficient usage practices and low token efficiency can further add to the problem,” he said.