---
title: US Is Not on Track to Lead in Physical AI as China Installs Nine Times as Many Factory Robots, Bipartisan Panel Concludes
description: China installed about 354,200 industrial robots in 2025 to 38,400 in the US. A bipartisan commission explains why America is behind in physical AI.
author: Darie Nani (Editor-in-Chief)
date: 2026-10-15T13:32:38.626Z
updated: 2026-10-15T13:32:38.644Z
canonical: https://www.sovereignmagazine.com/article/what-is-physical-ai
image: https://cdn.nanimediahouse.com/pexels-a-robotic-arm-welding-in-an-industrial-setting-emitting-brig-11951215.jpg
categories: Artificial Intelligence
content_type: News
region: United States
publication: Sovereign Magazine
schema_type: Article
---

The United States develops much of the software that runs modern robots but is not on track to lead in physical AI, meaning artificial intelligence built into machines that sense and act in the real world, a bipartisan commission co-chaired by Senators Ted Budd (R-NC) and Elissa Slotkin (D-MI) said in a preliminary report on October 13.

The National Security Commission on Robotics for Advanced Manufacturing is an initiative of the Special Competitive Studies Project (SCSP), and SCSP President Ylli Bajraktari is its third co-chair. It found that the US leads in artificial intelligence, computer vision, autonomy and software, which the commission calls the "brain" of modern robotics, but lacks the scale to deploy robots, produce robotic systems and make critical components at home. "Physical AI at scale will define the next industrial era, and the United States is not on a trajectory to lead it," [the report says](https://www.scsp.ai/wp-content/uploads/2026/10/NSCRAM_Interim-Report.pdf). The commission argues that leadership in physical AI "will depend on productive deployment and resilient production capacity, not model capability alone."

> "The United States invented much of the intelligence behind modern robotics, but intelligence that can't be built, deployed, and sustained at scale won't secure American industrial leadership."
> — Ylli Bajraktari, President, Special Competitive Studies Project

## China Installed About 354,200 Industrial Robots in 2025, Compared With 38,400 in the US

Chinese factories installed approximately 354,200 industrial robots in 2025, more than nine times the roughly 38,400 installed in the United States, according to International Federation of Robotics (IFR) figures cited in the report. China's total was 59% of all installations worldwide. By the end of 2024 China had [more than two million industrial robots at work](https://www.sovereignmagazine.com/article/the-new-tech-hierarchy-s-impact-on-global-manufacturing-as-china-is-winning-the-ai-race), over five times the US stock of 393,700.

US installations still grew 12% in 2025, and the US overtook Japan to become the world's second-largest market for industrial robots, behind China, [the IFR reported](https://ifr.org/ifr-press-releases/news/five-million-robots-now-operate-in-factories-globally) in September. Most robots installed in American plants are imported, the report notes: Japanese and European companies supply much of the hardware, and many of them keep large production, integration and service operations in the US. In China, domestic robot makers supplied 195,000 of the robots installed in 2025, 55% of their home market, according to the IFR.

IFR President Jane Heffner said in September: "China's success in modernising its industrial system is based on its national robotics strategy, which was initiated ten years ago." She said the update in China's 15th Five-Year Plan, covering 2026 to 2030, "aims to focus its AI research on physical applications, with robots being the main drivers of economic growth."

All of these figures cover industrial robots only, and the report says comparable data on mobile robots and [humanoids](https://www.sovereignmagazine.com/article/xpeng-iron-humanoid-robot-mass-production) "are needed before drawing broader conclusions about leadership in physical AI."

## The US Ranks Eighth in Robot Density, Ahead of China in 22nd Place

The US ranks eighth in the world for robot density, with 307 industrial robots per 10,000 manufacturing employees, according to IFR rankings the report cites. China ranks 22nd, with 166 robots per 10,000 workers based on updated labor data from its National Bureau of Statistics, so US factories average more robots per worker. South Korea leads with 1,220.

Only 12.1% of US manufacturing plants reported using robots in 2021, according to US Census Bureau data the commission cites. Small and medium-sized manufacturers make up about 95% of US manufacturing firms. Many of these smaller manufacturers remain unsure of what robots can do, where to introduce them into their processes, how to choose the right technology or how to handle integration, the commission found. Narrow return-on-investment measures make the upfront cost hard to justify, and these companies also face capital limits and uncertainty over cybersecurity and liability.

China's deployment scale gives its companies more chances to improve processes and test systems, the report says. "That does not guarantee a lead in frontier innovation. It does mean that a persistent U.S. deployment gap could become an innovation gap," it adds.

## Manufacturers Are Waiting for Cheaper Robots, and Robot Makers Need Sales to Bring Prices Down

Manufacturers wait for robots to become cheaper, more adaptable, easier to integrate and better proven, while robotics companies cannot lower costs or improve performance without deployment at scale. The commission calls this supply-demand paradox a market failure that is unlikely to resolve on its own. "The market is trapped in a first-mover problem: each firm has reason to wait for someone else to absorb the uncertainty," the report says.

Robotics startups also struggle to finance production, because venture capital timelines do not fit hardware and customers run paid pilots but rarely commit to multi-year, large-volume orders until a robot has proven reliable in the field. In the report's words, "startups cannot secure financing to scale manufacturing without binding orders, and buyers will not issue binding orders without proof of scalable manufacturing capability."

## Federal Robotics Programs Are Spread Across at Least Four Departments

No single authority is accountable for federal robotics programs, which are split across at least four executive departments, the commission found. A manufacturer may get technical help from one program, financing from another, cybersecurity guidance from a third and workforce support from state or local bodies, each with different rules.

## Company Data Stays Locked Away, and Key Robot Parts Come From Concentrated Suppliers

The operational data US companies generate stays locked inside each company, in incompatible formats and under contracts that bar sharing, so no one can pool it and learn from it, the report says. Integration costs in the US remain high because deployment is fragmented and engineering is one-off.

Robots depend on parts such as [magnets](https://www.sovereignmagazine.com/article/us-rare-earth-magnet-plant-south-carolina-china), motors, precision reducers, sensors and power electronics, and the report says concentrated supply of these parts raises cost and strategic risk. The report does not assume the US must make every robot or component itself; the commission's concern is where dependence on foreign suppliers creates vulnerability.

## Up to 1.9 Million Factory Jobs Could Go Unfilled by 2033

US manufacturers could need about 3.8 million workers by 2033, and up to 1.9 million of those jobs could go unfilled, according to a 2024 Deloitte and Manufacturing Institute study cited by the commission. Robots can help fill dangerous and chronically vacant roles, but the commission says their effect on employment is "not predetermined" and depends on how fast companies adopt them and whether training and job redesign keep up.

## The Commission Will Publish Its Recommendations in March 2027

The commission says the US needs a national mission organized around deployment, drawing on private companies, research, capital markets, regional manufacturing ecosystems and skilled workers, "without copying China's state-directed model or requiring domestic production of every component." Its commissioners include executives from Nvidia, FANUC America, Boston Dynamics, General Motors, AMD and Intrinsic, along with Anne Neuberger, a general partner at Andreessen Horowitz and a former deputy national security adviser.

The commission describes the preliminary report as "a diagnosis, not a set of prescriptions." Its final report, due in March 2027, will make recommendations for federal, state and regional leaders, manufacturers and technology companies, educators and workforce bodies, workers and labor representatives, investors and research institutions.

## FAQ

**Q: What is a physical AI example?**
Nvidia gives cameras, robots and self-driving cars that perceive, reason and act in the physical world as examples. In a self-driving car, physical AI processes sensor data in real time so the car can perceive and understand its surroundings, Nvidia says.

**Q: What is the difference between generative AI and physical AI?**
Generative AI learns mostly from text and images, while physical AI also has to understand the 3D world well enough to act in it. Nvidia notes that generative models such as GPT and Llama are trained largely on internet data and are "limited in their grasp of the physical world and its rules." Physical AI adds an understanding of spatial relationships and how objects behave, using inputs such as images, video, speech and sensor data.

**Q: How far are we from physical AI?**
It depends on the machine. Industrial robots are already common, with about 5 million operating worldwide in 2025 according to the IFR, and the commission says AI-assisted programming, low-code interfaces and collaborative robots are widening where automation is feasible. Humanoids are less mature: the report notes that a humanoid that balances dynamically needs power to stay upright, so cutting its power in an emergency can be more dangerous, one reason safety standards lag behind the technology.

**Q: What is the Special Competitive Studies Project?**
The Special Competitive Studies Project is a nonpartisan, not-for-profit project focused on America's long-term competitiveness as AI and other emerging technologies reshape national security, the economy and society. SCSP launched the robotics commission on March 17, 2026 as a year-long effort and said the commission would serve as "the primary architect for a national strategy to scale next-generation robotics."