L&T Finance cuts cloud dependency, builds private infrastructure to control AI costs

L&T Finance is reworking its technology strategy by reducing dependence on public cloud providers and building its own private computing infrastructure, as rising costs linked to artificial intelligence begin to strain operations.The move comes after the company’s growing use of AI tools significantly improved business performance but also drove a sharp increase in expenses. Over the past few years, L&T Finance has embedded machine learning models across its core operations, particularly in credit underwriting and risk management.At the centre of this transformation are its proprietary AI systems, including Cyclops, which supports real-time credit decisioning, and Nostradamus, which predicts portfolio risk in advance. While these tools enhanced credit quality and forecasting accuracy, they also required heavy data processing and compute power, pushing cloud-related costs higher.Running these workloads on third-party cloud platforms led to a 50–60% increase in compute and data-processing expenses year-on-year, turning what was initially a flexible cost into a structural financial burden.To address this, the company, under Managing Director and CEO Sudipta Roy, has shifted strategy from renting computing capacity to owning it. Leveraging the engineering capabilities of the broader L&T Group, it is building a private, on-premises cloud environment, allowing it to take control of its infrastructure and reduce reliance on global hyperscalers.The transition has already delivered meaningful cost benefits. Internal processing on the company’s own systems is estimated to cost around one-third of what it previously paid to external cloud providers, significantly improving the economics of running AI at scale.The migration is expected to continue over the next 18–24 months, with L&T Finance planning to shift its core financial applications and critical workloads to the private infrastructure, further reducing dependency on external platforms.Alongside the infrastructure shift, the company is also preparing a targeted entry into the payments space. After securing a UPI licence, L&T Finance plans to launch a premium, niche payments offering focused on its base of approximately 28 million customers, rather than competing in the mass-market cashback-driven segment.This strategy will integrate its AI capabilities into front-end offerings, using insights and analytics generated from its internal systems to support personalised, high-value transaction experiences, including premium QR-based solutions.The broader shift highlights how rising AI adoption is reshaping technology strategies in financial services. As machine learning becomes core to operations, companies are increasingly focusing not just on algorithms but also on the cost and control of underlying infrastructure.L&T Finance’s pivot reflects a move toward greater ownership of computing resources, as firms look to balance innovation with long-term cost sustainability in an AI-driven environment.Also read: Meta scales practical AI ad tools, unifies creator ecosystem and expands business agents