Физический ИИ в Индии: спрос на роботов и технических специалистов вырос на 500%

A few months ago, a Bengaluru engineer’s home-built “Thindy Dosa Robot” went viral for doing something unbelievable: making dosas. It poured the batter, spread it across the tawa with a practiced swirl, waited for the edges to crisp, and lifted it off — not just perfectly round, but golden, crisp, and unmistakably dosa. It was a hobby project, built because its creator was watching his family cook and wondering, “what if a robot could do this while we relax?” (Sign up for THEdge, The Hindu’s weekly education newsletter.) Now imagine Thindy’s more advanced version, a few years down the line, in a busy Bengaluru restaurant. A customer walks in, orders a masala dosa on the app – extra crispy, no onions, paneer topping instead of the usual potato filling, chutney on the side. The robot arm reads the order, adjusts the batter thickness for extra crispiness, spreads it thinner at the edges, times the flip to the second, folds in the paneer masala exactly where the customer wants it, and plates it - all while three other orders, each customised differently, are being cooked on adjacent stations. This is physical AI in action. In July 2026, researchers at the University of California San Diego demonstrated something that would have sounded like science fiction a decade ago: they had two five-foot-tall humanoid robots work side-by-side to remove the gallbladder of a pig, controlled remotely by a human surgeon. The objective of this pre-clinical experiment was not to replace a surgeon — a human was very much in control throughout, guiding every motion — but to ask whether a general-purpose low-cost humanoid could extend a skilled surgeon’s hands into an operation theatre a thousand kilometers away. The purpose of the experiment is to help patients who would otherwise not have access to a critical surgery, particularly in small towns or villages of India, facing a shortage of trained surgeons. These are examples of the same underlying shift: Artificial Intelligence (AI) has learnt to leave the computer screen and act, carefully and adaptively, in the physical world. What is physical AI? Traditional industrial robots — such as fixed mechanical arms on automotive assembly lines or automated guided vehicles in fulfilment centers — operate strictly on deterministic, hard-coded rules written by human programmers. They possess high physical precision, repeatability, and speed, but they lack adaptive intelligence. A humanoid mimics human anatomy, typically featuring a head, torso, two arms, and either two legs or a mobile wheeled base. It can operate seamlessly within human-centric environments—climbing stairs, opening door handles, operating standard hand tools, and navigating narrow factory aisles without requiring facilities to be redesigned. However, a humanoid is merely the physical container; without advanced cognitive systems, and remains a mechanical puppet executing fixed motion scripts. Physical AI, also known as embodied AI, represents the integration of foundation machine learning models, computer vision, optical tactile feedback, and physics-informed neural networks directly into physical bodies. While traditional AI processes digital data to generate digital output, physical AI continuously senses the physical world, predicts physical forces, reasons through dynamic disruptions, and generates motor actions for movement in real time. Physical AI’s adaptability makes it a general-purpose technology that can be used across diverse mission-critical sectors such as healthcare, defence and security, space exploration, supply chain and logistics, agriculture and, manufacturing. Tata Consultancy Services (TCS) physical AI readiness report (July 2026) found that 77% of manufacturers anticipate a transformational impact on warehouse operations, 75% on assembly lines, and 72% on logistics and material movement. 71% view Physical AI as a people-first initiative designed to assist employees and enhance workplace safety, while 42% anticipate major workforce augmentation in hazardous environments. However, 44% admit that they have no clear idea who is accountable when a system fails, a warning India’s regulators should note early. The classroom-to-career bridge: higher education imperatives The transition from screen-based software to embodied agents is redrawing the engineering skills landscape. Demand is surging for cross-disciplinary specialists capable of blending software engineering with physical mechanics. Physical AI is creating an entirely new job market, with roles like sensor calibration, simulation and digital-twin engineering, human-robot interaction design, and AI-fleet supervision. Now, the obvious question is: what should a fresh graduate actually learn to walk into it? This is where higher education institutions (HEIs) have their tasks cut out. Universities must dismantle rigid departmental walls. Future graduates need interdisciplinary training that weaves together mechanical engineering, electronics, computer vision, and machine learning. Most Indian engineering colleges still teach robotics as a standalone elective as part of a mechanical degree. Physical AI careers need something more integrated: courses that combine embedded systems, computer vision, reinforcement learning, and basic mechanical design in the same programme, so graduates can seamlessly understand the “brain” and the “body” of these machines. IIT Delhi set up Collaborative Robotics (Cobotics) centre, which acts as the technology innovation hub with focus on strategic application areas like medical surgical systems, agriculture and disaster management, defence, and industry automation. It is equally important to provide hands-on experience to students. A student who has watched a robotic arm crush an egg because it misjudged the grip force learns something a teacher cannot teach in a class. Universities should partner directly with embodied AI start-up firms for structured internships and sponsored capstone projects. There is also a wide-open lane for management and design institutions, not just engineering ones. The TCS report makes clear that companies need people who can supervise fleets of physical AI systems and make judgment calls when something goes wrong — roles that call for operations and change-management training layered on top of basic technical literacy, not a pure engineering degree. B-schools that build short courses in “robotics operations management” now will be producing exactly the graduates TCS’s surveyed manufacturers say they cannot currently find. Transformation in jobs For decades, popular narrative suggested that the tasks of manual and industrial workers would be automated first, leaving desk jobs untouched. The reality of the physical AI economy is proving to be precisely the opposite. In the last 5 years, the number of app-based gig workers – mostly ride-hailing cab drivers and last- mile food and retail delivery workers – are estimated to have grown by over 55% from about 7.7 million to 13 million and each one of them earning about ₹25,000 to ₹30,000 of take-home per month. Report from the recruitment firm Randstad (July 2026) , which analyzed more than 50 million Indian job postings between early 2022 and 2026, reveals a striking pattern: hiring volume of industrial workers in India surged by 93%, significantly outpacing desk job creation. As industrial facilities, fulfillment warehouses, and smart infrastructure deploy thousands of physical AI machines, the demand for human technical trade skills has experienced unprecedented growth. Job postings for certified electricians jumped by 242% over the four-year tracking period, while demand for industrial HVAC technicians rose by 200% to support cooling requirements inside high-density edge compute centers. Most dramatically, job openings for specialised robotics technicians skyrocketed by over 500%. Every physical AI agent deployed on a factory floor, inside a medical ward, or along a logistics hub requires human technicians to handle physical wiring, replace worn mechanical joints, align optical sensors, and conduct routine calibration. Physical AI is creating a new category of highly rewarding industrial jobs even as it automates pieces of the old one. However, workers in desk jobs face a quieter kind of pressure — chipping away at the routine, process-driven slices of knowledge work: entry-level coding, basic financial analysis, routine customer support. Technologists tracking India’s IT sector have warned of a possible squeeze in desk jobs as AI agents take over the linear, well-defined tasks fresh engineering graduates traditionally start their careers on. NASSCOM’s research suggests India faces a shortfall of over a million skilled tech workers even as graduate unemployment runs above 40% - clear proof that there is not really a shortage of jobs, but a mismatch between what colleges produce and what a physical AI economy needs. The uncomfortable truth is that both workers are being asked to move up the same ladder, just from different starting points. For workers in desk jobs, whose current skills are concentrated in routine, process-driven work, the window for proactive reskilling is narrow. Employers, universities, and professional bodies need to treat mid-career upskilling as seriously as fresh graduate training by building short, stackable credentials through AI supervision, data judgement, and human-machine collaboration. For industrial workers, the priority is fast, practical, vernacular-language training in the trades that physical AI is creating — robotics maintenance, sensor calibration, industrial electrical work, HVAC for data centres — delivered through Skill India Mission and Industrial Training Institutes, not just engineering colleges. The factory welder becomes the robotics technician who keeps the welding robot precisely calibrated. The radiology assistant in the hospital becomes the person who validates what an AI diagnostic tool flags. The entry-level analyst becomes the person who asks better questions of the AI system and takes responsibility for its answers. In every case, the human role shifts from doing the routine task to supervising, judging, and being accountable for a machine that does it — and that shift is precisely where the human being remains irreplaceable. Physical AI extends human capability; it does not, at least not yet, replace human judgement, empathy, or accountability. Global market for embodied AI The commercial market for physical AI is experiencing explosive growth, expanding from an experimental research domain into a foundational global industry. Global demand in the next 5 years is estimated to grow at 35%-40% per year, with Asia-Pacific, powered by China, Japan, South Korea and India, as the fastest-growing region. Japan alone has committed roughly $6 billion to develop a sovereign AI model and deploy 10 million AI-equipped robots across 18 sectors by 2040 — a plan driven by its shrinking, ageing workforce, and a useful contrast for India, whose challenge is the opposite: not too few workers, but too few of the right skills for the jobs physical AI will create. The U.S. Federal Communications Commission has recently banned imports of new foreign-made humanoid and quadruped robots, targeting China over national-security concerns. Indian startups are quietly entering this frontier, building physical AI-powered industrial robots for the global market. Bengaluru-based Ati Robotics and its peers are racing to set up manufacturing in the US while their software and engineering base stays rooted in India. Reliance Retail Ventures, which holds a majority stake in Addverb Technologies, is already testing physical AI robots for industrial and defence use, while Gurugram-based Novus High-Tech Robotic is building humanoids and quadrupeds with the software entirely “Made in India.” A major bottleneck restricting the integration of emerging technologies within HEIs is the shortage of academic faculty trained in advanced industry applications. Institutions must actively design and mandate curated learning paths for faculty upskilling. India’s advantage here is real but is time-limited. Unlike Japan, which is deploying robots because it is running out of workers, India has the opposite problem - plenty of young workers, but a widening gap between their skills and what a physical-AI economy demands. A strategic roadmap: policy recommendations for the physical AI era To capture the multi-billion-dollar physical AI dividend while safeguarding national interests, industry leaders and policymakers should execute a three-pronged strategy: Unified safety certification and accountability frameworks: national standards organizations (such as the Bureau of Indian Standards - BIS) should develop standardised certification benchmarks for autonomous physical agents, establishing clear audit trails, mandatory hardware kill-switches, and liability frameworks for human-agent handoffs. Pre-competitive research and development and shared foundational infrastructure: governments should subsidise pre-competitive research consortia between academia, startups, and IT integrators to develop open-source physical foundation models and frugal edge-computing architectures tailored for emerging markets. Further, establishing shared ‘Living Labs’ in alignment with national initiatives such as the IndiaAI Mission, National Quantum Mission, and Innovations for Defence Excellence (iDEX) will provide students and researchers access to expensive technology testbeds and GPU clusters. Dedicated regulatory sandboxes across critical sectors: Establish bounded, real-world sandboxes—including agricultural testing corridors, autonomous logistics zones, surgical robotics testing wards, and aerospace simulation ranges—to safely iterate and de-risk physical AI hardware prior to commercial deployment. As we stand on the threshold of the physical AI revolution, redefining human-AI capability, the boundary between computation and the physical world is dissolving. By navigating geopolitical realignments, nurturing agile domestic startups, establishing robust governance protocols, providing interdisciplinary training to workforce and, cross-disciplinary courses to fresh graduates, India can harness the immense promise of embodied AI to build a more resilient and productive economy, along with employment for the youth. (Professor O.R.S Rao is the Chancellor of the ICFAI University, Sikkim. Views are personal) Published - August 26, 2026 08:30 am IST