India’s AI strategy is evolving from costly model development towards leveraging local, sector-specific data, offering new opportunities for enterprise-driven innovation and differentiation.
India’s artificial intelligence debate is moving away from an expensive race to build giant foundation models and towards a more practical advantage: proprietary data. That shift matters because the biggest gains may not come from trying to match US and Chinese technology groups model for model, but from using India’s linguistic breadth, sector knowledge and enterprise relationships to build AI that solves local problems.
The logic is simple. Large language models demand huge amounts of capital, electricity and specialist computing power, which makes them a poor fit for many Indian firms. Sridhar Vembu, the founder and chief scientist of Zoho, has argued that India should avoid costly large-model build-outs and focus instead on smaller models that are easier to train and deploy. A survey cited by Medianama suggests that many Indian AI deployments already reflect that reality: 74% rely on proprietary closed models accessed through APIs, while most GPU use is concentrated on inference rather than training.
That dependence on existing models does not necessarily weaken India’s position. Instead, it highlights where the commercial opportunity lies. Industry observers say the real value is often in turning messy, sector-specific information from areas such as healthcare, agriculture, manufacturing, logistics and banking into structured data that can be used to customise AI systems. TechRadar has reported that poor data quality is one of the main barriers to enterprise AI, while Microsoft chief executive Satya Nadella has warned that companies can end up paying for intelligence twice if they hand over valuable proprietary data to model providers.
For India’s IT services companies, that creates a more promising route than competing directly with global model makers. Firms with deep knowledge of compliance, workflows and customer operations can act as the bridge between powerful external models and practical business use. The risk, however, is that India becomes little more than a data supplier. To avoid that, companies will need to own the datasets, control the intellectual property created from them and focus on vertical AI products rather than low-end annotation work or generic infrastructure services. Naveen Tewari, chief executive of InMobi, has said the future lies in models trained on proprietary data, underscoring the view that India’s edge may come from specialisation rather than scale alone.
Disclaimer: This article is intended to inform and educate, not to recommend or endorse any financial product, investment or strategy. Please consider your own financial circumstances and seek professional advice where appropriate before making financial decisions.
