Autonomous Production Market Outlook 2026–2038: Market Size, Growth Drivers, Key Players, Strategic Developments & Adoption Forecast for Humanoid and Mobile Robots, Physical AI, Autonomous Cell Control and Lights-Out Factories — A Meticulous Next™ Foresight Brief

This Meticulous Next™ brief examines how autonomous production, meaning factories in which robots, vehicles and control software perceive, decide and act with little human direction, advances from programmed automation to physical AI and eventually to lights-out operation over the next twelve years. The change runs from the single part on the line to the whole plant, and the first steps are now in production.

The base is already large. Global installations of industrial robots reached 542,000 units in 2024, more than double the level of ten years earlier, and the operational stock stood at about 4.66 million units. China alone installed 295,000 units, or 54% of the world total. Yet conventional robots repeat fixed motions in fenced cells. The next investment is not more of the same machine but machines that adapt, which is why the market for autonomy is distinct from the market for robots.

The proof is coming from automotive plants. At Spartanburg in the United States, two Figure 02 humanoid robots worked in the body shop of BMW for about ten months, five days a week in ten-hour shifts. They supported the production of more than 30,000 BMW X3 vehicles, moved over 90,000 sheet metal components and logged about 1,250 operating hours, positioning parts with millimetre accuracy. The pilot also forced changes to safety barriers and to wireless coverage in the hall. A humanoid has been shown to work a real shift, and the lessons are about integration as much as about the robot.

Europe has followed. BMW announced in February 2026 the first humanoid pilot in its European network at Leipzig, where the AEON robot of Hexagon Robotics, a wheeled machine 1.65 m tall that weighs 60 kg, first entered the shop floor in December 2025 and is being introduced to high-voltage battery assembly and exterior component production. BMW also created a Centre of Competence for Physical AI in Production to carry pilots from the laboratory to the plant. The company reports no layoffs linked to these programmes and describes the work as ergonomic relief. Adoption is starting with the hardest jobs for people, which is where acceptance is easiest.

The pace of the next steps is set by safety, cost and software. Figure announced a third model in June 2026 that sorts loose parts into sequencing trolleys, a task still done by hand. New standards for robot safety and the European Machinery Regulation, which applies from January 2027, will shape what can run without fences. A plant becomes lights-out only when perception, planning, maintenance and logistics are all autonomous, so the road is long and each step has its own business case.

The brief maps the technology, its indicative market size and forecast, the factors behind its growth, the developments of the last 24 months, the key players operating in the space and the adoption trajectory to 2038.

It is a focused 116-page decision brief for manufacturers, system integrators, robot and automation suppliers, software and compute providers, investors and policy makers. It presents an indicative trajectory rather than a segmented market model. Its purpose is to identify which tasks and plants adopt autonomy first, how safety rules and unit economics sequence demand, and who captures value as production moves from programmed to autonomous.

Autonomy is built in layers. The lowest layer is the machine: arms, wheeled bases, legged or humanoid forms with grippers and sensors. The middle layer is perception and skill: models that see parts, plan motions and handle variation. The top layer is the plant: software that schedules work, moves material, monitors quality and maintains equipment without waiting for an engineer. Each layer can be sold separately, but value is created when they work together.

The robot base explains why this matters now. Conventional industrial robots are programmed for one task and fenced for safety, so they suit long runs of identical work. Asia took 74% of new installations in 2024, and the 700,000 unit mark is expected to be passed by 2028. Growth in this base gives suppliers scale in motors, sensors and controllers, which are the same components that humanoid and mobile machines need. The market for autonomy therefore grows on top of an existing supply chain.

Humanoids are the visible edge, but their role is narrow at first. At Spartanburg the Figure 02 robot loaded sheet metal for welding, a physically demanding and ergonomically poor task. At Leipzig, AEON handles battery component work and delivers materials to the line while navigating around obstacles, and it moves at up to 2.5 m a second on wheels rather than feet. The pattern is that the early humanoid is a mobile manipulator doing a few defined jobs next to people. The business case is ergonomic relief, flexibility and availability, not replacement of whole lines.

Physical AI is the technology that widens the range of jobs. Instead of coding every motion, a model learns from demonstration and simulation to pick parts that arrive unsorted, as the third Figure model is designed to do for sequencing trolleys. BMW now evaluates technology partners through a central pathway from concept to plant pilot, which shows how large manufacturers intend to industrialise it. The skill is only as useful as its reliability, so the next years are about hours of operation, failure rates and recovery without a human.

The limits are safety, integration cost and trust. The pilots at BMW required new barriers and partitions and better wireless coverage, which shows that the surrounding plant has to change. Standards for industrial robot safety and the European Machinery Regulation, applying from January 2027, will define when robots may work without fences and with AI in safety functions. Lights-out operation in the strict sense is therefore an end point, reached cell by cell. The decade turns on whether reliability, safety certification and unit cost reach the point where autonomy is the default choice for new lines.

The autonomous production market, covering humanoid and mobile manipulation robots, autonomous mobile robots, physical AI platforms, and autonomous cell and line control software, together with integration services, and excluding conventional programmed industrial robots, standard programmable logic control and non-autonomous conveyors, is valued at USD 9–12 billion in 2026 and is projected to reach USD 100–140 billion by 2038, at a CAGR of 22–23%.

The 2026 base rests on autonomous mobile robots in factories and warehouses, mobile manipulators, early humanoid pilots, physical AI software and integration services in the automotive, electronics and logistics sectors. From that base, growth follows the pilot pathway. Spartanburg and Leipzig show how a manufacturer moves from a pilot to a plant, and orders from other carmakers and electronics producers, together with the expected rise in robot installations to 700,000 units by 2028, build the order book.

The forecast is sensitive to the reliability of physical AI, to safety certification and to unit cost of humanoid and mobile manipulation robots. Regionally, China leads on volume and humanoid policy, the United States on humanoid start-ups and physical AI software, Germany and Japan on industrial automation suppliers, and Korea on carmaker-led robotics programmes.

The base case assumes steady gains in physical AI reliability, safety standards that allow fence-free work from the early 2030s and falling unit costs, giving the inflection around 2032. An accelerated case adds rapid humanoid cost reduction, strong Chinese and Korean programmes and labour shortages in automotive and electronics, bringing the inflection to 2030 and lifting the 2038 value to the top of the range. A delayed case assumes slower reliability gains, strict safety certification and weak investment in new plants, holding growth at the bottom of the range.

Growth drivers

• Large installed base: 542,000 robot installations in 2024, 4.66 million in operation and 700,000 installations a year expected by 2028.
• Proof in production: more than 30,000 BMW X3 vehicles supported and 90,000 components moved by humanoids at Spartanburg.
• Physical AI that adapts to variation, such as sorting unsorted parts and handling battery components.
• Labour, ergonomics and flexibility pressures in automotive, electronics and logistics.

Enablers

• Supply chains for motors, sensors and controllers scaled by the industrial robot base.
• Central pathways for pilots, such as the Centre of Competence for Physical AI in Production at BMW.
• Industrial wireless coverage and digital twins that connect machines and plants.
• Capital for humanoid and physical AI developers from carmakers, technology groups and investors.

Restraints and barriers

• Reliability, uptime and recovery of humanoid and physical AI systems in long shifts.
• Safety certification and rules for robots working without fences, including AI in safety functions.
• Integration cost: pilots required new barriers, partitions and wireless coverage.

Workforce and acceptance concerns, although BMW reports no layoffs linked to its programmes.

Five converging forces will determine how fast autonomous production scales.

(1) Physical AI reliability and skill range. (2) Safety standards and regulation. (3) Unit cost of humanoid and mobile robots. (4) Plant integration and software. (5) Labour pressure and manufacturer investment. The brief assesses each force for direction, speed and confidence.

The key players operating in autonomous production include Bayerische Motoren Werke AG, Mercedes-Benz Group AG, Volkswagen AG, Toyota Motor Corporation, Hyundai Motor Company, Tesla, Inc., Amazon.com, Inc., Hon Hai Precision Industry Co., Ltd., Hexagon AB, Figure AI, Inc., Boston Dynamics, Inc., Agility Robotics, Inc., Apptronik, Inc., UBTECH Robotics Corp Ltd, Unitree Robotics Co., Ltd., Neura Robotics GmbH, Physical Intelligence, Inc., FANUC Corporation, ABB Ltd, KUKA AG, Yaskawa Electric Corporation, Universal Robots A/S, Siemens AG, Rockwell Automation, Inc., Schneider Electric SE, Omron Corporation, Robert Bosch GmbH, Hitachi, Ltd., Mitsubishi Electric Corporation, Symbotic Inc., Locus Robotics Corp., Zebra Technologies Corporation, Teradyne, Inc., NVIDIA Corporation, Microsoft Corporation, Telefonaktiebolaget LM Ericsson, the International Federation of Robotics, the Verband Deutscher Maschinen- und Anlagenbau, the Fraunhofer-Gesellschaft, the Massachusetts Institute of Technology, Carnegie Mellon University, ETH Zurich, the European Commission and the International Organization for Standardization.

The brief profiles representative players in each archetype and assesses which are positioned to make autonomy the default in new factories.

The competitive landscape is forming around six archetypes. Manufacturers as lead adopters. Humanoid and physical AI developers. Industrial automation and robot groups. Mobile and logistics autonomy providers. Compute, software and connectivity providers. Research bodies, standard-setters and regulators. Competitive intensity is moderate in 2026 and is expected to rise as pilots become programmes and automation groups bring physical AI into their lines through 2033.

Where value migrates.

In 2026 value sits in robot hardware, autonomous mobile robots and the first pilots of physical AI. By 2032 it moves to physical AI platforms, plant-level control software, integration services and safety systems that let robots work without fences. By 2038 it settles in platform owners with the largest fleets, robot makers at scale and integrators with plant-level software. Fixed automation for short runs and one-off pilots lose standing.

The signals converge on production, not demonstration. The base is large, with 542,000 robot installations in 2024. Humanoids have worked real shifts, with more than 30,000 vehicles supported at Spartanburg. Programmes are forming, with a central physical AI centre at BMW and a second plant at Leipzig. Tasks are widening, from loading sheet metal to sorting parts. Each signal moves autonomy from a laboratory topic toward a plant budget line.

Research and patent activity reflects where the chain is weakest: reliability of manipulation, recovery from failure, safety of AI-driven motion and the cost of the machine. The brief therefore tracks four indicators: operating hours per robot between interventions, the number of humanoid and mobile manipulation robots ordered per manufacturer, the share of new lines specified with physical AI and the certification of fence-free work.

Regionally, China leads on volume and humanoid policy, the United States on humanoid start-ups and physical AI software, Germany and Japan on industrial automation suppliers, and Korea on carmaker-led robotics programmes. Automotive plants, electronics producers and logistics operators are the first demand pools.