SHANGHAI — After years of building humanoid robots that can run, dance and perform backflips, Chinese companies are now pouring resources into a harder challenge: developing the artificial intelligence that could make those machines useful across a much wider range of jobs, according to Reuters reporting published on November 1, 2026.
The shift is fuelling an investment boom in what the industry calls embodied AI — software that allows robots to understand their surroundings, learn unfamiliar tasks and adapt without having to be programmed for every single job. Chinese embodied AI companies raised 93.5 billion yuan, about $13 billion, across 322 deals in the first half of 2026, five times the amount raised a year earlier, according to data research firm ITjuzi, as cited by Reuters.
Funding is increasingly reaching specialist developers of robot intelligence alongside the manufacturers themselves. As competition drives down hardware prices, robot makers have, in the words of Dermot McGrath, founder of Shanghai-based technology consultancy ZenGen Labs, "concluded that the margin won't come from the hardware but from the brain."
Illustration: A humanoid robot at work on a factory line, learning to handle parts under the watch of engineers.
What Happened
Reuters, drawing on interviews with founders and senior executives at five Chinese robotics firms as well as industry analysts, reports that the race for robot intelligence has become a new front in the technology rivalry between the United States and China. American companies lead in some advanced robot AI systems, Reuters reports, while Chinese developers hope that their manufacturing strength and access to real-world training data will help them challenge their American competitors.
The stakes, as framed by the reporting, go beyond novelty demonstrations. The country that best combines AI with machines could gain an advantage in automating factories and logistics, potentially reshaping industrial productivity and the global market for robotics technology.
"The brain is indeed the weakest link in the complete robotics stack," Spirit AI co-founder Gao Yang told Reuters. That assessment captures the industry's central problem: the bodies have advanced faster than the minds that must direct them.
On the American side, startups including Physical Intelligence, Skild AI and Generalist are developing software intended to teach robots different skills without extensive retraining. Generalist says its GEN-1.5 model completed around 60 percent of simple tasks after viewing a demonstration lasting between 3 and 12 seconds, according to Reuters. On the Chinese side, X Square has published research showing robots acquiring handling skills from a single human video without retraining the model for each task.
"The leading American labs are setting a demanding benchmark, particularly in how quickly robots can acquire and reuse skills," robotics consultant Georg Stieler told Reuters, adding that he considers China's strongest startups credible competitors.
Specialist Chinese startups including X Square, Galaxea and Spirit AI have each reached valuations of roughly $3 billion, according to company disclosures cited by Reuters. Established manufacturers are investing heavily too: Hong Kong-listed UBTech advertised a chief scientist position overseeing AI model development with compensation of up to 124 million yuan, about $18 million, including equity and benefits. Hangzhou-based Unitree, one of the world's largest humanoid robot makers by shipments, has about 30 researchers developing AI that predicts how objects behave, according to a person briefed by the company.
Background
Unlike chatbots, which learn from vast quantities of internet text, robots need examples of physical interaction — from gripping objects to handling flexible wires. Chinese companies are pursuing different approaches to collecting that information, with some employing people to demonstrate tasks, others using computer simulations, and some combining both. The field remains crowded, Reuters reports, with no company yet establishing itself as the clear leader.
Spirit AI has hired around 1,000 people to record everyday and industrial activities, Gao said. ACE Robotics uses wearable sensors to capture workers' movements and grip forces in stores and hotels. Ant Group's robotics unit Robbyant said it trained a model using 160,000 hours of material, including 50,000 hours of human-guided robot movements, and trained software on 20 robot configurations from 17 brands, aiming to use the same AI across different machines.
DexForce is taking another approach, using computer-generated environments to teach robots. Its chief operating officer, Nicole Yang, said the company can prepare some industrial tasks for deployment within 6 to 8 hours and reported success rates above 99.9 percent in specified precision operations. But simulations cannot perfectly reproduce physical interactions, Jerry Wang of spatial intelligence company Manycore Tech, which designs simulated environments for AI training, told Reuters.
"Data quality is more important than data diversity, and more important than simply collecting data at massive scale," X Square co-founder and chief operating officer Yang Qian told Reuters. "We want robots to shift from being driven by code to being driven by common sense," Yang added.
Illustration: Embodied AI gives a robot a working understanding of its surroundings, rather than a script for every task.
Why It Matters
The outcome of this race will shape who captures the value in robotics. If hardware keeps getting cheaper, the money — and the strategic leverage — moves to the software layer: the models, the training data and the chips that run them. That is why investors are paying billions for specialist intelligence startups, and why established manufacturers are hiring chief scientists at eye-watering salaries.
It also matters for factories, warehouses and logistics networks, where robots that can adapt to unfamiliar tasks could be deployed far more widely than machines that must be reprogrammed for each job. Early real-world deployments in China already hint at that direction, though they remain narrow. Battery maker CATL said Spirit AI robots connect flexible wires during battery-pack testing at its Zhongzhou plant, reporting insertion success above 99 percent. Robbyant said its robots handle orders in two Shanghai pharmacies, with staff checking medicines before dispatch.
Yet the industry faces serious headwinds. Chinese firms face shortages of advanced Nvidia chips due to United States export controls, and a squeeze on IPO funding amid regulatory concerns over inflated valuations and overcrowding, Reuters reports. "As robots become more powerful, the shortage of advanced chips will become a serious problem for China's embodied AI industry," said Zhu Mingxuan, who leads development of JoyIn Technology's Aether robot AI.
Chinese companies are trying to reduce their computing requirements through more efficient models, synthetic data and simulations, while domestic chipmakers, including Huawei, are developing alternatives to Nvidia hardware. Reuters notes that it remains unclear how far those efforts can offset the shortage.
What Happens Next
Watch for two different kinds of progress. The first is the demonstration breakthrough: robots that learn new jobs from a single video or a spoken instruction. JoyIn and other developers are already testing whether their software can enable robots to tackle unfamiliar problems. Zhu described an experiment in which a robot repeatedly failed to place headphones in a cupboard beyond its reach. Then engineers showed it videos of how people managed the task, and the robot pushed a box towards the cupboard and tried stepping onto it. "Every time it was exploring a different behaviour," Zhu said.
Unitree founder Wang Xingxing said this month that the company had tested OpenAI's GPT-6 on robots with encouraging results. He said GPT-7, if released next year, could bring robotics its own "ChatGPT moment" — a step-change in capability akin to ChatGPT's 2022 launch. Wang said he would consider that threshold reached when robots were able to complete 80 percent of tasks in 80 percent of unfamiliar environments from spoken instructions.
The second kind of progress is slower and harder to film: robots that customers can actually count on, day after day, in workplaces that were not designed for them. Consultant Stieler cautioned against confusing the two. "If the question is when a model first handles a broad range of unfamiliar tasks in demonstrations, the optimists may well be right," he said. "If the question is when a customer can count on it, I would add years, not months."
Success, Reuters concludes, will depend on how the same AI performs across different customers and workplaces — and that will require companies to train increasingly powerful models with advanced chips. The race for robot brains, in other words, is only just beginning, and its winners may not be the companies that built the flashiest bodies.
Illustration: The US-China rivalry in robotics is shifting from hardware to the intelligence that drives the machines.
Sponsored: explore today's featured sponsored offers
FAQ
What is embodied AI?
Embodied AI is software that allows a robot to understand its surroundings, learn unfamiliar tasks and adapt without being programmed for every job, rather than simply following a fixed script.
Why are Chinese robot companies suddenly focused on AI?
Because hardware prices are falling as competition grows, so companies believe future profit will come from the "brain" — the models and data that make robots generally useful — rather than from the body alone.
How much money has flowed into Chinese embodied AI?
Chinese embodied AI companies raised 93.5 billion yuan, about $13 billion, across 322 deals in the first half of 2026, five times the amount a year earlier, according to ITjuzi data cited by Reuters.
What is holding the industry back?
Shortages of advanced chips linked to United States export controls, the difficulty of collecting high-quality physical training data, and questions about whether demonstration breakthroughs will translate into reliable everyday use.
When will robots have their "ChatGPT moment"?
Unitree's founder suggests a future model such as GPT-7 could trigger one, defined as robots completing 80 percent of tasks in 80 percent of unfamiliar environments from spoken instructions — but analysts caution that dependable commercial use could still be years away.
Sources
Reuters, "China's robot makers have mastered the moves. Now comes the race for brains," November 1, 2026. All figures, quotations and company claims in this article are attributed to that reporting and to the sources it cites, including ITjuzi and company disclosures.
Comments
Post a Comment