# Small Indian Manufacturers Hit Data, Legacy-System Barriers to Scaling AI > Source: > Published: 2026-09-14 05:53:04+00:00 BENGALURU, India — AI in manufacturing is moving beyond isolated proofs of concept, but Indian medium- and small-scale manufacturers face a more fundamental hurdle in the form of inconsistent data, legacy equipment, and the absence of standard operating models. At last week’s [Electronics City Industries Association (ELCIA) Tech Summit 3.0](https://elciatechsummit.in/), industry experts argued that scaling AI across manufacturing will depend less on sophisticated models and more on plant-level data, contextual knowledge, and collaboration between manufacturers and technology providers. In her opening address, Nandini B, chairperson of ELCIA and director of operations at TESCOM Electronics, said the broader manufacturing ecosystem needs to participate in the technology transition. “Manufacturing today is becoming more sophisticated with AI, automation, advanced electronics, sensors, and whatnot,” she said. “So, small and micro enterprises cannot afford to watch this transformation from the sidelines. They must be a part of it.” [View All](https://www.eetimes.com/category/sponsored-content/) For manufacturers, however, the immediate challenge is not simply adopting AI but ensuring that the systems, data, and processes needed to support it are in place. Aditi Sharma, president and chief manufacturing excellence officer at Uno Minda, said AI has already moved beyond a nice-to-have capability in manufacturing operations. Uno Minda operates 72 plants making 175 different product types, with some facilities dating back 50 years. “I am sitting in a space where I am questioning whether I should incorporate AI today or should I do it tomorrow,” Sharma said. The question, she said, is whether manufacturers should continue with conventional plants because of the availability of human labor in India, or whether human-intensive manufacturing could eventually become a constraint on agility and the adoption of new technology. Sharma pointed to repetitive operations and the number of physical touches involved in manufacturing as areas where AI can be applied. She added that its deployment also depends on the availability of data and clear ownership. “We need data, a large amount of data,” she said. “We need the right decision-making, and then we need the correct ownership.” Ravi Ramarao, founder of Mexo Technologies, said the data challenge began at the machine level. Manufacturing equipment, including modern machines, does not necessarily generate data in a standardized format, making it difficult to directly apply AI across different systems. Manufacturers need to acquire machine data, combine it with production data, understand production patterns and cycles, and then begin generating recommendations, he said. “This needs time,” Ramarao said. “Sometimes it varies from three months to six months to understand all the things that are happening in the company and move to an autonomous system, where the AI can make decisions.” He also cautioned against treating AI models as transferable solutions. “Each organization has its own operating context, which means that data from one organization cannot simply be used to train a model for another… That makes contextual data more useful than a sophisticated model that lacks the appropriate data,” he said. The challenge becomes more pronounced when manufacturers try to replicate a successful proof of concept across multiple plants. Sharma said proofs of concept are often conducted in controlled environments with the right data, machines, and people. The conditions can be very different at other plants, where data may still be collected manually, and operators may not have experience with digital systems. “All manufacturing plants in an organization are not at the same level. They have different levels of data,” she said. “Some of them have manual data while others work with digital data, making it difficult to achieve the same results everywhere.” She said manufacturers should consider replication before beginning a proof of concept, rather than treating it as a separate exercise after the initial project has succeeded. The experts also distinguished between existing factories and new manufacturing facilities. K. Bhavani Shankar, head of operations, D&M&C, Yaskawa India, said AI could be introduced into existing facilities without completely redesigning them, as much of the equipment installed in recent decades is already capable of connecting to external systems. “Maybe the equipment, which has been used from 2000 onwards, almost all is ready to connect to the external environment. So there, only small changes may be required,” Shankar said. New factories, however, offer an opportunity to incorporate AI from the outset rather than adapting older equipment and processes later. Shankar also said the potential of AI extends beyond productivity to quality, resilience, and changes in manufacturing processes. “If you look at it, the real opportunity of AI, I personally see, is bigger than productivity itself, extending to quality, resilience, and a possible redesign of manufacturing operations and processes,” he said. Ramarao said the technology should ultimately remain tied to the manufacturing problem it is intended to address. Technology teams can provide the technical foundation, but the relevant business function needs to define the problem. For Indian medium and small-scale manufacturers, the discussion goes beyond choosing an AI model. The more immediate task is to make different plants, machines, data systems, and operating practices capable of supporting AI technology. Without that foundation, a successful proof of concept may remain just a proof of concept, rather than a solution that can be scaled across the factory network. ##### See also: [India Budget 2026-2027: Semiconductors, Manufacturing, and Tax Reforms](https://www.eetimes.com/india-budget-2026-2027-semiconductors-manufacturing-and-tax-reforms/) [India Needs Component Depth and Skills to Support Manufacturing Momentum](https://www.eetimes.com/india-needs-component-depth-and-skills-to-support-manufacturing-momentum/) [India’s 2035 Chip Ambitions Focus on Targeted Design, Manufacturing Leadership](https://www.eetimes.com/indias-2035-chip-ambitions-focus-on-targeted-design-manufacturing-leadership/)