Chinese humanoid makers raised $13 billion in six months — now the race is for the brain
Chinese embodied AI funding jumped fivefold in early 2026 as the race shifts from hardware to the software that makes robots useful.
Chinese companies building the intelligence for humanoid robots 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. The surge marks a turn for the industry: after years spent perfecting robots that can run, dance and perform backflips, the money is now chasing the software that could make those machines useful across a wider range of jobs. Funding is increasingly reaching specialist developers of robot intelligence alongside the manufacturers themselves.
As competition drives hardware prices down, robot makers have concluded that the profit will not come from the body but from the brain, said Dermot McGrath, founder of Shanghai technology consultancy ZenGen Labs, in an interview with Reuters. Spirit AI co-founder Gao Yang called the brain “the weakest link in the complete robotics stack” in an interview with Reuters. Specialist developers are now raising at scale: X Square, Galaxea and Spirit AI have each reached valuations of roughly $3 billion, company disclosures show. Established manufacturers are investing too. Hong Kong-listed UBTech advertised a chief scientist role overseeing AI model development with pay of up to 124 million yuan ($18 million) including equity and benefits, and Unitree has about 30 researchers working on AI that predicts how objects behave. X Square co-founder and COO Yang Qian put the goal this way in an interview with Reuters: “We want robots to shift from being driven by code to being driven by common sense.”
The race has two lanes. American startups including Physical Intelligence, Skild AI and Generalist are building software meant to teach robots new skills without extensive retraining. Generalist says its GEN-1.5 model completed about 60% of simple tasks after watching a demonstration lasting three to 12 seconds. China’s X Square has published research showing robots picking up handling skills from a single human video without retraining the model for each task. American labs still lead in some advanced robot AI, robotics consultant Georg Stieler said in an interview with Reuters, but he called China’s strongest startups credible competitors. The United States holds an edge in advanced AI research and access to powerful computing chips; Chinese firms counter with close work alongside robot makers and industrial customers, and with access to real-world training data.
Unlike chatbots, which learn from internet text, robots need examples of physical interaction — gripping objects, handling flexible wires. Companies are collecting that data in different ways. Spirit AI has hired about 1,000 people to record everyday and industrial activities. ACE Robotics uses wearable sensors to capture workers’ movements and grip forces in stores and hotels. Ant Group’s robotics unit Robbyant says it trained a model on 160,000 hours of material, including 50,000 hours of human-guided robot movements, across 20 robot configurations from 17 brands. DexForce takes the opposite approach, teaching robots in computer-generated environments, and says it can prepare some industrial tasks for deployment within six to eight hours, with success rates above 99.9% in specified precision operations. Simulations cannot perfectly reproduce physical contact, cautioned Jerry Wang of spatial intelligence company Manycore Tech, which designs simulated environments for AI training. “Data quality is more important than data diversity, and more important than simply collecting data at massive scale,” Yang said.
The first real jobs are appearing. Battery maker CATL says Spirit AI robots connect flexible wires during battery-pack testing at its Zhongzhou plant, reporting insertion success above 99%. Robbyant says its robots handle orders in two Shanghai pharmacies, with staff checking medicines before dispatch. Unitree founder Wang Xingxing said this month the company had tested OpenAI’s GPT-6 on robots with encouraging results, and that GPT-7, if released next year, could bring robotics its own “ChatGPT moment.” His test for that moment: robots completing 80% of tasks in 80% of unfamiliar environments from spoken instructions. Stieler urged caution — a breakthrough in demonstrations would not necessarily mean robots were ready for broad commercial use. “If the question is when a customer can count on it, I would add years, not months,” he said. Chinese firms also face headwinds: shortages of advanced Nvidia chips under US export controls, and a squeeze on IPO funding amid regulatory concern over inflated valuations and overcrowding.
The question now is whether the same AI can work across different customers and workplaces — and whether Chinese firms can train increasingly powerful models with fewer advanced chips, using more efficient models, synthetic data and simulations, while domestic chipmakers including Huawei build alternatives to Nvidia hardware. How far that offsets the shortage remains unclear. What is clear is where the industry’s money and attention have moved: the body is largely built, and the contest is now for the brain.