The global
robotics
market.

This document covers the whole of robotics, with an in-depth analysis of Physical AI (humanoids and embodied AI). The assessment is as of November 8, 2026.

Two further axes complete the taxonomy in the data pack: form factor (fixed arm, wheeled mobile, quadruped, biped, flying, aquatic) and degree of autonomy (programmed, teleoperated, supervised partial autonomy, autonomous).

Industry and logistics already buy in the hundreds of thousands of units. Defense is scaling up under the impact of the war in Ukraine. Companions are a nascent market, funded mainly by third parties. Our long-term hypothesis, dated November 8, 2026, is that these three markets will rank among the largest after 2030. No published data establishes this today, and consumer robots (34.2 million units in 2025) remain by far the largest market by volume.

Scope: humanoids, mobile manipulators and companion robots. Military drones and ground robots, already deployed in volume, are not included.

This order differs from the priority of buyer markets in the next chapter. That priority combines volume, ability to pay and speed of adoption across all robots.

A job is a set of tasks, and robots take over only some of them. In the United States, current technology could in theory automate 57% of hours worked: 44% through software agents and only 13% through robots. McKinsey specifies that this is a technical potential, not a forecast, and expects a narrower short-term effect on physical work, given limits in dexterity, safety and cost (McKinsey Global Institute, Nov. 2025). In France, fewer than 10% of jobs combine vulnerabilities that threaten their existence, but about half will see their content change significantly (Conseil d'orientation pour l'emploi, 2017).

Remote supervision and data collection jobs depend on the level of robot autonomy. Our assumption is that they will decline as autonomy improves.

The gains go to technical profiles; the low-skilled jobs created are poorly paid and can be offshored. In East Asia, robots displaced about 1.4 million low-skilled jobs between 2018 and 2022 and created about 2 million skilled jobs; the least skilled and oldest workers were pushed into informal employment (World Bank). Working conditions are changing in both directions: less load carrying in Douai, but cybersickness and repeated takeovers for the teleoperators in Manila. In China, nearly 60% of Longyu Big Data’s data collection team are part-time stay-at-home mothers (Yicai).

Humanoids are not yet weighing on employment; conventional robotics is. About 7,750 humanoids were put into productive use worldwide in H1 2026: this is our calculation based on IDC (25,000 shipments, 31% of which outside research, education, demonstration and data collection; confidence C). The United States alone has 6.9 million material movers (BLS).

Adjustment happens first through hiring. According to internal documents revealed by the New York Times, Amazon aims to avoid more than 600,000 hires in the United States by 2033; the company responds that they reflect the view of a single team (Allwork, citing the NYT, confidence B). In Germany, robot-related industrial job losses were fully offset by services: the adjustment fell on young entrants, while incumbent employees changed tasks within their plant (Dauth et al., 2021).

In France, 90% of the 760,000 positions to be filled each year through 2030 are replacements for departures (Dares and France Stratégie). Employers plan 2.28 million hires in 2026, 43.8% of which are considered difficult (France Travail, BMO 2026).

The physical task is being offshored: supervision from Manila for Tokyo convenience stores, data collection in India, Nigeria and Argentina. China officially recognizes the new occupations. France automates less than its neighbors: 4,500 industrial robots installed in 2025, versus 24,800 in Germany and 7,800 in Italy (IFR EU). The tasks targeted first belong to occupations facing labor shortages: 80.6% of planned hires of skilled machining workers are considered difficult in 2026 (BMO).

Out of every 100 workers worldwide, 59 will need training by 2030: 29 in their current role, 19 to be redeployed, and 11 are at risk of not receiving it (WEF, Future of Jobs 2025). In France, pathways already exist, such as the UIMM robotics integrator certification. Metalworking companies cite robotics and automation specialists, and maintenance technicians with updated skills, as growing needs (OPCO 2i).

None of these organizations discloses the effect of its robots on its headcount, role by role.

Companions sell poorly direct to consumers: isolated people are rarely the ones who buy technology. Public payers are stepping in. In Korea, local governments distribute Hyodol to seniors living alone: Seoul's Guro district has distributed 412 since 2019 (Rest of World). In the United States, ElliQ is provided through public programs in five states; in March 2026, Washington State added it to Medicaid, the first Medicaid coverage. In China, the MIIT and the Ministry of Civil Affairs are testing, from 2025 to 2027, senior-care robots at home, in community services and in care facilities, including for emotional companionship. The model taking hold is a subscription paid by a third party: local government, health insurer, pension fund.