India lagged on hi-tech factories, but AI tools can boost competitiveness

AI-driven "physical intelligence" is reshaping India's factories, pushing firms beyond labour arbitrage to precision, automation and higher local value addition

A combination of artificial intelligence (AI)-led upgrade and regulatory push is forcing manufacturing companies to think differently. The Ministry of Electronics and Information Technology (Meity) has emphasised that the benefits of the production-linked incentive (PLI) scheme are directly linked to meeting the domestic value-addition targets. The goal is to make India more than an assembly hub for electronics and establish it as a high-value manufacturing destination. This requires increasing local value addition in smartphones from the current 20 per cent to 35-40 per cent, and in electronic components from about 25 per cent to 50 per cent. 

To achieve this, manufacturing companies will have to think differently — and beyond human skills. Increasing local value add means making more of the components — the parts that go into electronics gear — locally. This includes capacitors, camera modules, complex printed circuit boards (PCBs), resistors, control panels, display screens, and so on. Manufacturing units will also have to lean on technologies which, for instance, detect and eliminate errors that are beyond the capability of manual human inspection. In other words, companies will have to do a massive upgrade driven by physical AI to ensure global quality at scale to eventually expand the local components ecosystem. 

Ideas like Waymo (formerly the Google self-driving car project) were early movers in the physical AI journey — embedding intelligence in cars to drive in the real, unpredictable world. Waymo’s primary product is its autonomous driving system, called the Waymo Driver. This system combines advanced hardware and software to navigate complex urban environments without a human driver. Now, in manufacturing units and elsewhere, something similar is starting to make machines intelligent.

  Mover over labour arbitrage 

For decades, India’s manufacturing pitch rested on labour arbitrage — cheap, abundant manpower that could assemble products at scale. But the next phase is being shaped by a very different proposition: Intelligence, precision, and speed. In this new world, factories are not just sites of production. They are data-rich ecosystems. 

“India is moving from low-value assembly to high-value precision engineering, with a robust chip ecosystem and development of fabs etc. Also, there’s a push to localise the production of hardware, including computers, servers, and an emphasis on defence electronics, aerospace components and so on,” said Vinod Sharma, managing director of Deki Electronics and chairman of the Confederation of Indian Industry (CII) National Committee on Electronics Manufacturing. “But to compete globally, companies need to invest in automation, process control, and advanced manufacturing systems.” That means rethinking the manufacturing shop floor itself, with AI. 

The most visible impact of AI on manufacturing is deceptively simple: Fewer mistakes, said Nakul Kumar, cofounder, Cashify, a recommerce platform that buys, repairs, and sells used smartphones and other electronics. 

In electronics manufacturing, where components can be smaller than a grain of rice — such as capacitors used in medical implants, wearables, or smartphones — even microscopic defects can derail an entire production batch. 

Earlier, quality checks relied heavily on human inspection. “Today computer vision systems can spot hairline cracks, alignment errors or soldering defects (in seconds),” Kumar added. “AI is also helping machines tell when they are about to fail. So, you can fix processes earlier instead of firefighting later.” 

The result is not just higher efficiency, but greater predictability. Production becomes smoother, less stressful, and more accountable. 

Varun Gupta, cofounder of GOBOULT, a hearables and wearables brand, sees a broader shift underway: “AI is fundamentally shifting the manufacturing conversation from labour cost to using machine intelligence.” 

Predictive maintenance reduces factory downtime, intelligent robotics ensures consistency at scale, and AI-led forecasting makes supply chains more responsive. The winners, Gupta added, “will not just be those who produce at scale, but those who embed intelligence across the value chain”. 

The rise of connected machines 

In next-generation factories, machines will no longer be isolated units performing discrete tasks. They will be part of a connected system. “Future hi-tech manufacturing factories are moving toward fully connected, intelligent production environments where robotics, internet of things (IoT) systems, and AI work together in real time,” said Raviteja Chivukula, CEO, Perceptyne Robots. Hyderabad based Perceptyne Robots develops AI-driven industrial humanoid robots designed to automate dexterous tasks in manufacturing sectors such as electronics and automotive. 

This convergence is giving rise to what experts like to call Chivukula “physical intelligence” — machines that can perceive, decide and act in dynamic environments. Unlike traditional automation, which is rigid and task-specific, these systems are adaptive. 

The goal is to remove dull, repetitive and physically demanding tasks from humans, allowing machines to take over such work. Robots are being designed not just for controlled environments but for real-world variability, where precision and flexibility are equally critical. 

IoT sensors feed data into AI systems, enabling real-time monitoring of equipment health, production quality, and operational efficiency. Predictive algorithms can flag potential failures before they occur, reducing downtime and waste. The factory, in effect, becomes a living system — learning, adjusting, and optimising itself. 

At the extreme end of this automation lies a fully automated facility that can operate without human presence, often with only robots. Globally, a few companies in Japan, the Netherlands, and China are experimenting with such facilities. These units rely on industrial robotics, IoT systems and AI-driven control mechanisms to enable 24/7 production with minimal human intervention. 

In India, however, the picture is nuanced. The capital costs are high, and given India’s labour advantage, fully automated factories may not always make economic sense. Instead, the more realistic model is hybrid. Tasks that pose safety risks or where fatigue may impact outcomes, are likely to be automated first. Optical inspection is a case in point. “After 30 minutes to one hour, the human eye gets strained and defect detection becomes unreliable. AI-powered inspection systems can perform this task more consistently,” said JS Gujral, managing director, Syrma SGS, a Gurugram-based electronics manufacturing services company. 

Sharma of CII is also of the view that hybrid factories are the near-term reality. “There will be areas where robot-led units make sense — like hazardous env­ironments or extreme precision — but in most manufacturing setups, we will see a mix of human and robotic systems.” 

Precision & miniaturisation challenge 

One of the biggest challenges for Indian manufacturing is in moving up the value chain — from assembling imported components to producing high-precision parts domestically. 

Consider earbuds, hearables, and wearables. While India has emerged as a major assembler of these products, much of the high-value manufacturing still happens overseas. The reason is simple: Miniaturisation demands extreme precision. 

Components such as capacitors, sensors, and microchips that go into these coin-sized — or even smaller — devices require manufacturing environments that can deliver consistent, defect-free output at microscopic scales. 

This is where AI and advanced automation become critical. By enabling real-time quality control, predictive maintenance and adaptive production processes, AI can help Indian factories meet global benchmarks. It can also reduce reliance on imports by making local production more viable. 

Ashok Gupta, director at Optiemus Infracom, a Noida-based telecommunications and electronics manufacturing firm, said the industry has already begun this transition. “We have moved from basic assembly to higher local value addition and vertical integration,” he said. Capabilities such as PCB fabrication, in-house testing and software integration are expanding. But the journey is far from complete — since around 70-75 per cent of the components used in electronics are imported. 

This can change if manufacturing companies start implementing AI tools. 

For investors, this convergence of AI, robotics and manufacturing represents a new frontier. “India has the engineering depth and manufacturing demand to be a genuine originator in physical AI, not just an adopter,” said Vishesh Rajaram, founding partner at Speciale Invest, a deep-tech focused venture investor. 

Startups are beginning to build the intelligence, sensing, and autonomy layers that will define this next phase. Adoption is also accelerating. AI use in Indian manufacturing has increased from 8 per cent to 22 per cent in FY24, according to TeamLease Digital. Meity and IT industry body Nasscom project that the domestic AI-in-manufacturing market will grow from $1.2 billion in FY25 to $8 billion by 2030. 

For all the benefits of physical AI, there are a few challenges as well. Integrating AI into existing manufacturing systems can be complex and expensive. 

For example, moving from manual, to AI powered vision to identify faults in products requires both software and hardware upgrade. Workforce reskilling is another critical issue. As machines take over routine tasks, workers will need to move into higher-value roles that require technical and analytical skills. For instance, if a shop floor worker is testing gadgets and if that task is now done by AI, can the worker transition into monitoring and analysing product testing data that comes on the dashboard? 

Factory shop floors will evolve from just men and machines into hybrid ecosystems — highly automated, deeply connected, and continuously learning. Robotics, IoT and AI will handle precision and repetition, while humans will oversee, interpret and innovate. In that sense, the transformation is not about replacing labour, but about redefining it.

The next phase of manufacturing, as Chivukula said, “will be a highly automated, yet human-supervised ecosys­t­em that combines robotics, IoT and AI to improve efficiency, safety and productivity while allowing humans to focus on higher-value decision making roles.” 

When that happens, it will overhaul the role of human workers in shop floors — much like knowledge workers now use ChatGPT, Anthropic, Gemini et al to assist in their daily grind.

Pros

• Improved quality control:

AI-powered computer vision can detect defects in microchips, circuit boards, or smartphone components with far greater accuracy than human inspectors. This reduces waste and ensures higher reliability

• Predictive maintenance:

AI systems analyse sensor data from machines to predict failures before they occur. This minimises downtime and extends equipment life

• Supply chain optimisation:

AI can forecast demand, optimise inventory, and streamline logistics, ensuring raw materials and

components arrive just in time for production

 

• Customisation and flexibility:

AI enables factories to quickly adapt production lines for different product variants, supporting mass customisation without sacrificing efficiency

 

• Cost efficiency in the long run:

While upfront investment is high, AI reduces operational costs over time by lowering defect rates, minimising downtime, and improving throughput

 

Cons

• High implementation cost:

Upgrading legacy systems with AI requires new sensors, computing infrastructure, and integration software — often an expensive transition

• Integration complexity:

Many factories run on older enterprise resource planning, or ERP, systems that don’t easily connect with AI platforms, requiring custom middleware and long transition periods

• Data requirements:

AI models need massive amounts of labelled data (eg, images of defects, sensor logs). Collecting and curating this data is

time-consuming and expensive

• Workforce upskilling:

Routine shop floor jobs may decline, but workers need retraining for technical roles like AI system monitoring, data analysis, and robotics programming

• Cybersecurity risks:

Connected machines relying on cloud platforms to perform will be vulnerable to cyberattacks that could disrupt production

---

The writer is a New Delhi-based independent journalist