China has established a considerable lead over the USA and different main economies of their readiness to construct and deploy bodily AI techniques, in response to a brand new examine from robotics know-how firm OpenMind.
The OpenMind Physical AI Readiness Index 2026 assesses 58 international locations and 22 metropolitan clusters on their capability to construct, energy, provide and function superior embodied AI, together with humanoid and legged robots, general-purpose manipulators, autonomous cell machines and more and more autonomous industrial techniques.
China receives a rating of 82.9, in contrast with 69.6 for second-placed Japan and 69.0 for the USA. South Korea scores 66.9 and Germany 66.1.
The 13.3-point hole between China and Japan is the most important between any two adjoining international locations within the index.
The findings add one other dimension to the rising worldwide debate over competitors in synthetic intelligence, significantly between the USA and China.
US President Donald Trump has repeatedly emphasised American management over China in AI. In February, Trump stated: “AI is a giant deal. AI goes to be perhaps the most important factor – larger than the Web, larger than anything. We’re main China by quite a bit. We’re main everyone by quite a bit.”
The OpenMind examine doesn’t try to measure the identical factor. As an alternative, it argues that management in software program and frontier AI fashions doesn’t essentially translate into a bonus when AI has to maneuver from computer systems into bodily machines.
OpenMind says: “Bodily AI readiness shouldn’t be an AI query. Each nation that may plausibly compete has entry to a lot the identical fashions. They differ in electrical energy, actuators, magnets, machine instruments, integration labour, and permission.”
China leads within the bodily foundations of AI
Considerably, OpenMind doesn’t attribute China’s general lead primarily to AI analysis or software program.
The USA scores 78.7 for human capital and software program, in contrast with China’s 73.8, and the US leads the index’s compute and semiconductor class.
China’s benefit as an alternative comes from the bodily infrastructure and industrial provide chains required to fabricate robots at scale.
China scores 95.3 for supplies, in contrast with 48.5 for the USA, in addition to 83.2 for robotics and actuation and 82.1 for power.
The report says China put in 295,000 industrial robots in 2024, representing 54 % of worldwide installations and greater than all different international locations mixed.
The index makes use of six weighted classes: power and grid infrastructure; compute and semiconductors; robotics and actuation; supplies and provide chains; human capital and software program; and demand.
Robotics and actuation receives the most important weighting at 25 %, adopted by human capital and software program at 20 %. Vitality, compute and supplies every account for 15 %, whereas demand represents the remaining 10 %.
The report’s nation rankings place China nicely forward of a second group comprising Japan, the USA, South Korea and Germany. A broader group of European and East Asian economies follows, with the UK scoring 44.4.
Humanoid benefit comes all the way down to motors and magnets
One of many report’s central arguments is that the availability chain for humanoid robots could show extra essential than entry to superior AI chips.
In line with OpenMind, a humanoid robotic can comprise between 1,000 and a couple of,000 chips, however semiconductors account for less than round 10 % of its invoice of supplies.
Actuators, by comparability, symbolize an estimated 40 to 60 %. Gearboxes account for 30 to 50 % of the price of an actuator.
The vital elements subsequently embody strain-wave and cycloidal reducers, planetary curler screws, frameless torque motors, precision bearings and rare-earth magnets. Japan stays significantly sturdy in high-precision elements, whereas China has developed the manufacturing capability required for quantity manufacturing.
The report identifies everlasting magnets as maybe essentially the most concentrated a part of the availability chain.
China refines 91 % of the world’s uncommon earths and manufactures 94 % of sintered rare-earth everlasting magnets. OpenMind says trade estimates additionally put China’s share of processing for heavy uncommon earths corresponding to dysprosium and terbium, that are essential for robotic motors, at 99 % or greater.
The financial penalties could possibly be appreciable.
Citing a McKinsey estimate, OpenMind says manufacturing a Tesla Optimus Gen 2 with out Chinese language suppliers would improve its estimated invoice of supplies from roughly $46,000 to $131,000 – a rise by an element of two.85.
China dominates early humanoid shipments
The report additionally factors to China’s quickly increasing share of the rising humanoid robotics market.
OpenMind estimates international humanoid shipments at between 13,300 and 18,000 items throughout 2025, with Chinese language firms accounting for round 90 %.
Within the first half of 2026, shipments reached roughly 19,100 items, in response to figures cited by the researchers, with Chinese language distributors accounting for about 97 %.
China’s Ministry of Trade and Data Expertise is projecting output of greater than 100,000 humanoids in 2026 and has directed that greater than 10,000 must be in industrial use by the top of the 12 months, in response to the report.
The researchers warning, nevertheless, towards treating introduced manufacturing capability as precise manufacturing. Throughout humanoids, magnets and batteries, OpenMind says introduced capability can exceed delivered capability by elements starting from 5 to 50.
US leads in compute, however faces infrastructure constraints
The USA retains substantial benefits.
OpenMind scores the US at 91.6 for compute and 78.7 for human capital, giving it important strengths in AI computing, analysis, software program and funding.
However its profile is significantly weaker within the supplies and manufacturing infrastructure wanted for bodily AI. The report describes the US profile as “compute- and talent-heavy with a skinny supplies band”.
Vitality infrastructure is one other potential constraint.
OpenMind estimates China’s electrical energy system is roughly 2.3 instances the dimensions of America’s and says China added roughly 540 GW of producing capability throughout 2025, in contrast with about 53 GW of US utility-scale additions.
In the meantime, greater than 2,000 GW of era and storage tasks are sitting in US interconnection queues, with tasks accomplished in 2025 taking a median of greater than 5 years to maneuver from utility to industrial operation.
Capacity to construct robots is simply half the query
OpenMind individually examines what it calls “working latitude” – whether or not a rustic offers the authorized, labor and social surroundings essential to deploy more and more autonomous machines.
The researchers assess AI and equipment legislation, the power of labor establishments, public attitudes and demonstrated willingness to permit autonomous machines to function.
The outcomes reveal an uncommon sample: some international locations with sturdy technological and industrial capabilities could face better obstacles to deploying robots.
Three of the 5 international locations with the best bodily AI functionality scores fall beneath the median for working latitude. Germany, for instance, scores 66.1 for functionality however solely 28 for working latitude.
OpenMind says China is the one main economic system combining what it considers top-tier bodily AI functionality with excessive working latitude.
The report comes as governments are taking more and more totally different approaches to AI. UK Prime Minister Andy Burnham instructed the UN Normal Meeting this week that Britain needs to assist set up worldwide ideas and requirements for frontier AI, describing the know-how as each a significant financial alternative and a possible supply of threat.
That debate has largely centered on frontier fashions, security and regulation. OpenMind’s analysis means that the rising competitors over bodily AI could also be decided by a unique set of things: factories, elements, electrical energy, engineering expertise, provide chains and the power to deploy autonomous machines at scale.
Eight of prime 10 robotics clusters are in East Asia
The geographic focus turns into nonetheless extra pronounced when OpenMind seems to be at particular person industrial areas moderately than international locations.
Eight of the ten main metropolitan clusters in its evaluation are in East Asia.
The report identifies totally different roles inside China’s robotics ecosystem: Beijing focuses on algorithms, requirements and datasets; the Yangtze River Delta offers quantity manufacturing; the Pearl River Delta offers speedy {hardware} iteration; and Wuhu focuses on lower-cost industrial robotics.
By comparability, the San Francisco Bay Space receives the utmost rating of 100 for human capital and software program however solely 32 for supplies.
OpenMind argues that these localized manufacturing ecosystems matter as a result of {hardware} growth relies upon closely on suppliers and specialist engineering data that can’t essentially be reproduced just by offering extra capital or AI compute.
The corporate acknowledges important uncertainty in its rankings. Round 55 % of the weighted index relies on cited quantitative information, whereas roughly 45 % depends on analyst-assigned scores.
To check the outcome, OpenMind says it reran the index 2,000 instances whereas various analyst assessments and index weightings. The ensuing 90 % confidence intervals averaged 9.4 factors, making many intently spaced nation rankings unsure.
China, nevertheless, remained in first place in each simulation, in response to the researchers, together with a separate take a look at that diminished all Chinese language-sourced inputs by 20 %.
OpenMind says it plans to publish month-to-month updates to the index and has launched the underlying dataset and methodology so researchers can apply totally different assumptions and weightings.
