S05.2 · Industrials & Advanced Manufacturing
Robotics ($40-50B) and the broader automation stack ($220-230B) as AI automates the robot programming that has justified integrator margins.
Industrial automation and robotics is the hardware and integration work behind automated manufacturing and material flow. Industrial robotics alone runs roughly $40-50B in 2025 (Grand View Research / Mordor Intelligence); the broader automation stack of PLCs, motion control and machine vision is far larger, at $220-230B (2025E, MarketsandMarkets). The economics to watch are not in the robot arm. They are in the integrator's labor: AI is automating the robot programming that has historically been the system integrator's core value-add, while physical-AI research blurs the line between robotics hardware makers and AI model providers.
Historical growth of 7-9% is forecast to step up to 10-13%, on reshoring, labor shortages and EV and battery capex. Over half of global robot installations happen in China (IFR); the arms themselves are still built mainly in Japan, Germany and the US — demand and supply have different addresses.
Concentration depends on the layer. Fanuc, ABB, Yaskawa and KUKA hold roughly 45-50% of robot-arm supply, a genuinely consolidated hardware market. Automation software and PLCs fragment across Siemens, Rockwell and Schneider among others. Upstream sit motors, sensors and precision reducers, where Harmonic Drive and Nabtesco run a near-monopoly on the reducer component specifically — the chokepoint in any robot-arm supply-risk assessment. Downstream, product moves through system integrators and machine builders into plant operators.
Hardware earns 20-30% margins; software and services earn 40%+. That spread, not sentiment, is what pulls strategic attention toward the recurring software and integration layer. The business is capital-intensive and capex-cyclical, with recurring integration and software revenue a growing share of mix, and ISO 10218 safety certification gating deployment into most industrial settings.
Adjacent ground runs through test and measurement — machine vision specifically — industrial software and MES, and precision machining, where integrators already share the shop floor with contract manufacturers. Recurring software and service revenue is the hedge against hardware commoditization; mobile robotics, AMRs and warehouse automation are where new physical-automation dollars get chased.
The consequential number is 3-5x: what programming and integrating a robot into a production line has historically cost relative to the price of the robot hardware itself. The integrator's labor, not the machine, has been the dominant cost of automation. Agentic AI that handles path-planning and programming compresses that labor line directly — and with it the manual programming expertise that was the integrator's moat. A separate effect runs through vision-based quality control, where AI-driven machine vision automates defect detection and takes out the manual inspection labor that has traditionally run alongside automated lines.
Both effects converge on one structural question: what the integrator's business model is worth once the programming skill that justified its margin is substantially automated. The label forming around the shift is "physical AI" — robotics OEMs building their own AI models while AI model providers move into physical control, a boundary that did not meaningfully exist five years ago. Vision-AI quality control is deployed at scale today. Autonomous robot programming that removes most manual integration work sits on a 2-5 year horizon.