A humanoid robot that costs about $46,000 to build today will cost roughly $131,000 if you take Chinese suppliers out of the bill of materials, according to a report released today by OpenMind that ranks the top 25 countries by their ability to build and operate advanced robots.

"It’s not very useful for the US to have robots that cost 3x more than anywhere else," OpenMind founder and CEO Jan Liphardt told me on my Humanoid Daily podcast . "That immediately puts us at a disadvantage."

Liphardt is a Stanford bioengineering professor building OpenMind, which he describes as an open, hardware-agnostic software layer for robots. (Think Android for robotics.) His team’s new Physical AI Readiness Index scores 58 countries and 22 metro clusters on whether they can actually build, power, supply and operate advanced robots at scale.

Top countries for advanced robotics

The key factors in “physical AI readiness,” according to the report, include robotics and actuators, human capital and software, energy and grid capacity, compute and semiconductors, materials and supply chain, plus the amount of local demand for robots and advanced automation.

Here are the top 25 countries in the ranking, with their scores:

  1. China 82.9
  2. Japan 69.6
  3. United States 69.0
  4. South Korea 66.9
  5. Germany 66.1
  6. Taiwan 55.4
  7. France 51.8
  8. Switzerland 51.3
  9. Sweden 50.6
  10. Italy 50.5
  11. Netherlands 49.6
  12. Singapore 48.2
  13. Canada 45.3
  14. Austria 44.4
  15. United Kingdom 44.4
  16. Spain 43.1
  17. Denmark 42.2
  18. Finland 40.5
  19. Israel 40.5
  20. Norway 38.6
  21. Belgium 37.8
  22. India 37.2
  23. Poland 37.0
  24. Australia 35.4
  25. Czechia 34.1

Both China and Japan rank ahead of the United States in “physical AI readiness.” What that translates into is the ability to build and ship advanced new robots. China’s 13.3-point margin over second place is wider than the gap between second and fourteenth.

Interestingly, when you zoom in for more details, the American problem becomes much more apparent. The US leads compute at 91.6 and human capital at 78.7, both by wide margins, but America scores just 48.5 on materials and supply chain. In other words, the build blocks and the manufacturing components.

There, the US score just fifth overall, at barely half China's 95.

"Many people think when they think of physical AI is that physical AI is all about AI," Liphardt said. "There’s this notion that if you have the best models, then you're also going to win physical AI. But based on what we can tell, that's not true."

In part to rebuild local manufacturing the FCC recently added foreign-produced advanced robotic devices to its Covered List . That includes sidewalk delivery bots, inspection quadrupeds and humanoids and essentially any ground-mobile machine over two kilograms with sensors, autonomy and a real network connection. Under these new rules, foreign-made units can no longer be authorized for the US market. That’s good news according to some robotics experts , but the cost impacts, at least in the short term, are significant.

Adding tens of thousands of dollars to the cost of domestically-produced robots will not help the industry grow locally, which is why the IDC suggests robot adoption speed will take a serious hit if the rules stick around.

"The entire US robotics industry will now need to figure out how to make advanced robots in the US," Liphardt told me.

That is not and will not be easy, for a variety of reasons: manufacturing capability, raw materials availability, and ecosystem density … essentially the reasons why the U.S. scored fairly low on the Index.

It will not, however, be impossible either.

The U.S. does have the best AI, and is continually reinvesting in it – like Figure’s billion-dollar bet on video training data. But AI model advantages keep eroding as open source AI advances, just months behind the big proprietary models, seemingly. "Countries, especially China, are open sourcing advanced models also for physical AI," Liphardt adds. "That, of course, re-levels the playing field."

What’s harder than new AI models is the actual metal and rare earths robots require, along with advanced mechanical parts.

Actuators, a robot’s muscles, are 40% to 60% of a humanoid’s bill of materials. Semiconductors are about 10%, and most of those are mature 200mm and 28-to-90nm parts almost nobody is short of. The scarce parts are strain-wave and cycloidal reducers, planetary roller screws, frameless torque motors, precision bearings and rare-earth magnets. Magnets are a particular problem, because China makes 94% of the world’s sintered neodymium permanent magnets and refines 91% of all global rare earths. ( There are American alternatives , but not all of them are fully up to complete production speed yet.)

OpenMind calls magnets the tightest single constraint in the entire supply chain: tighter than chips, and far tighter than batteries, which turn out not to be a constraint at all.

A core challenge is that the FCC edict is global.

It targets China, because China makes 97% of today’s shipping humanoid robots , but it makes no exception for allies. The FCC defines "foreign-produced" via the Buy American Act’s domestic end product standard: 65% domestic component cost through 2028, rising to 75% in 2029. Which means Japan, the second-place finisher in this index, doesn’t solve the problem for American robot makers either.

Harmonic Drive Systems of Japan holds an estimated 71% to 85% of the global strain-wave reducer market, according to the report. Nabtesco holds about 60% of large industrial robot joints. A meaningful fraction of all of the humanoid robot joints on earth, therefore, depends on a Japanese company. But a Japanese part doesn't count toward 65% domestic content any more than a Chinese one does.

That said, there are domestic manufacturers who are massively vertically integrating , like 1X, which makes the Neo humanoid robot. 1X makes almost everything in-house, including actuators. That’s going to lessen the impact of the FCC decision.

Other manufacturers, like Figure and Apptronik, are also vertically integrating where possible.

Regional ecosystem density is a challenge

But there’s another problem for most robot makers in America – at least the ones who have raised billions of dollars and can’t vertically integrate everything. And that’s just sheer lack of ecosystem density.

"One of the things that China has done very well is spatially concentrate everything that you might need," Liphardt said. "Within about a 10 kilometer radial distance, you have everything you could possibly need from carbon fiber to CNC to bearings to sensors to PCBs."

The payoff for concentration, of course, is iteration speed. "If you're building something, you can go across the street, you can talk to someone who builds gears, and then you go in the other direction, and you're able to just move much more quickly than you can if everything is scattered across a big country."

Eight of the ten leading robot-building metro clusters are in East Asia. Shenzhen–Dongguan–Guangzhou is first, with an ecosystem score close to 90. The Bay Area is top in the United States but only ninth globally, with a materials score of 32.

Here are the top 20 regional clusters according to the report:

  1. Shenzhen–Dongguan–Guangzhou 87
  2. Shanghai–Suzhou–Hangzhou 86
  3. Tokyo–Kanagawa 80
  4. Seoul–Gyeonggi–Ulsan 78
  5. Beijing (Haidian + Yizhuang) 74
  6. Hefei–Wuhu 72
  7. Nagoya–Toyota 71
  8. Hsinchu–Taichung–Taipei 70
  9. Bay Area 70
  10. Stuttgart–Karlsruhe 70
  11. Munich–Augsburg 68
  12. Boston–Cambridge 67
  13. Zurich–Basel 65
  14. Eindhoven–Brainport 63
  15. Singapore 61
  16. Västerås–Gothenburg 60
  17. Austin 59
  18. Milan–Bologna–Modena 57
  19. Pittsburgh 57
  20. Odense 56

America’s problem isn’t the total lack of robotics companies, suppliers, or even materials producers. The problem is relatively extreme dispersion. "We certainly have a lot of the pieces potentially here in the U.S., but they’re typically in very different places," Liphardt told me. "And those pieces that we do have are typically very small and there aren't a lot of choices."

There are some bright spots. The Bay Area, as mentioned above. Detroit–Ann Arbor ranks 21st on cluster readiness globally. Austin is 16th. Boston is 12th.

All of these are significant starting points that could serve to galvanize more ecosystem density, but that might take government nudging, and maybe incentives.

"You can imagine, for example, different states making it much easier to field test physical AI in maybe hospitals or to build roads or bridges," Liphardt says. In other words, government can participate on both the demand and the regulatory side, and actively push robots into real work to generate the engineering feedback and training data that America can't otherwise buy.

That’s not unlike China, where most humanoid robots are bought by training facilities set up by various levels of government. They buy the robots, and the robot makers buy the data back from them to improve their robots. It’s circular, sure, but it’s an industry kickstarter too.

Ultimately, there is light at the end of the tunnel. There is some hope for local manufacturers.

"The gap is definitely surmountable," Liphardt said. "And so we shouldn't be afraid. This is, if anything, it's information that helps us prioritize."

OpenMind is releasing the full dataset and the code behind the index. "You're welcome to change assumptions, to play out various scenarios," Liphardt said. "Our goal is simply to make it a lot easier for people to have a better sense of what's going on."