India’s AI opportunity massive, but legacy infrastructure holds back scale: Confluent's Rubal Sahni

India stands at a pivotal moment in its artificial intelligence journey, with the scale of its digital ecosystem offering a “once-in-a-lifetime opportunity” to shape the global data economy. However, unlocking this potential will depend on how quickly enterprises move beyond legacy systems and embrace real-time data infrastructure. 

According to Rubal Sahni, Area Vice President & Country Manager, India at Confluent, the biggest challenge lies in what he describes as a “modernisation paradox” — where the most valuable enterprise data remains trapped in outdated systems. 

“Legacy modernisation sits at the heart of India’s enterprise AI journey because it’s where all the ‘interesting' data actually lives, but it’s also where most of the friction is. Banks, insurers, telcos, PSUs and large conglomerates still run their most critical workloads on mainframes, monolithic databases and traditional messaging stacks. Modernising these environments is what unlocks this core data so it can feed real-time analytics and AI/GenAI use cases like instant payments, hyper-personalisation, fraud detection and predictive maintenance.” 

While many organisations have shifted to cloud, APIs, and microservices to enable AI experimentation, Sahni notes that true transformation requires rethinking how data flows across systems. AI, by design, cannot rely on static or outdated datasets, making real-time context critical for meaningful outcomes. 

As AI adoption becomes a boardroom priority, CIOs are increasingly walking a tightrope between speed and control. Sahni observes that successful leaders are not slowing innovation but embedding governance into the data layer itself. 

“The shift I am seeing is from ‘move fast and experiment’ to ‘move fast, but responsibly’. For example, in sectors like banking or healthcare, AI decisions cannot be opaque. Every output must be explainable, traceable, and auditable. That’s where governance is moving upstream into the data itself. 

Put simply, CIOs are balancing AI ambition with governance by moving fast on use cases, but only on top of a controlled, observable, and auditable AI foundation.” 

Despite heavy investments, nearly 90% of AI initiatives fail to move beyond the pilot stage — a phase Sahni terms “pilot purgatory.” This is largely because pilots are built on static datasets and controlled environments, which fail to replicate real-world complexity. Without unified data streams and clearly defined KPIs, these projects struggle to deliver tangible business impact. 

To address this gap, Confluent is partnering with firms such as Deloitte, Infosys, and Tata Consultancy Services to help enterprises “industrialize” AI deployment. These collaborations aim to combine domain expertise with real-time data infrastructure, enabling organisations to move from experimentation to production at scale. 

The shift to real-time data is already reshaping industries. In BFSI, it enables instant payments, fraud prevention, and real-time risk assessment. In retail, it powers dynamic pricing and hyper-personalised experiences, while in manufacturing, it drives predictive maintenance and efficient supply chains. 

Even India’s high-demand periods, such as the festive season, highlight the importance of real-time systems. Digital platforms like Swiggy and Zomato manage massive demand surges without disruption, underlining the role of real-time data orchestration. 

Sahni believes the next phase of India’s digital evolution will depend on moving from data ownership to data orchestration — where real-time intelligence drives productivity, financial inclusion, and public services. 

“The opportunity is massive, but channelizing the investments and right tech stack with clear objectives will be the differentiator for us as a nation. We are not just participating in the digital economy anymore. We are helping define how it will operate.”