Chinese embodied AI companies raised 93.5 billion yuan, or $13 billion, across 322 deals in the first half of 2026, five times the amount a year earlier, Reuters reported on 2 November, citing data research firm ITjuzi. By our arithmetic, that is about $40 million, or about 290 million yuan, per deal.
Taking Reuters’ “five times” at face value, the same companies raised roughly 18.7 billion yuan in the first half of 2025, by our arithmetic, so the year-on-year increase was about 74.8 billion yuan. Reuters writes that “Funding is increasingly reaching specialist developers of robot intelligence alongside manufacturers”.
Business Pill · Physical AI Explained
A short Business Engineer explainer on physical AI, models that perceive and act in the real world. It teaches the general idea only and does not discuss ITjuzi’s data or the companies in this piece.
The key insight: Chinese embodied AI companies raised 93.5 billion yuan ($13 billion) across 322 deals in the first half of 2026, according to ITjuzi data reported by Reuters, which is about $40 million per deal by our arithmetic, and Reuters reports more of the money reaching developers of robot intelligence alongside manufacturers.
Update, 4 Nov, 09:46Z: October fundraising slowed to about $1.6B, Bloomberg reports. Fundraising in China’s robotics sector fell to about 11 billion yuan ($1.6 billion) in October, Bloomberg reported, citing the same tracker, ITjuzi. That is less than half the nearly 24 billion yuan raised in August and down 67% from an all-time high of 32.8 billion yuan in June, according to the report.
For scale, the first-half total in this piece works out to about 15.6 billion yuan (about $2.2 billion) a month, so October came in roughly 30% below that average, by our arithmetic. The two counts may not cover the same companies: Reuters described the first-half figure as embodied AI funding, while Bloomberg says ITjuzi’s robotics count includes humanoid developers such as Unitree and AgiBot, makers of AI models for robots and component suppliers from sensors to dexterous hands. Treat the comparison as indicative only. The figures we cite from Bloomberg’s report cover June, August and October.
The Reuters report, by Laurie Chen and Eduardo Baptista, describes Chinese robot makers moving resources from hardware that can run, dance and perform backflips to the AI that controls the machines. It draws on interviews with founders and top executives at five Chinese robotics firms and with industry analysts.
As hardware prices fall, robot makers have “concluded that the margin won’t come from the hardware but from the brain”, Dermot McGrath, founder of Shanghai-based consultancy ZenGen Labs, told Reuters. Spirit AI co-founder Gao Yang said: “The brain is indeed the weakest link in the complete robotics stack”.
First-half funding raised by Chinese embodied AI companies, ITjuzi data as reported by Reuters on 2 November 2026: 93.5 billion yuan ($13 billion) across 322 deals in H1 2026. The H1 2025 bar is our arithmetic from Reuters’ ‘five times’ and is approximate; the per-deal average ($13 billion / 322) is our arithmetic. Valuations are company disclosures cited by Reuters.
What Was Posted on X
Xiaomi is peak Chinese tech firm.
Went from making cheap smartphones and household appliances to EVs and AI models.
Now Xiaomi is making humanoid robots that run autonomously on its own VLA model to help build its EVs.
Kyle Chan (@kyleichan) traces Xiaomi from cheap smartphones and household appliances to EVs and AI models, and writes that it is now making humanoid robots that “run autonomously on its own VLA model to help build its EVs”, quoting Xiaomi’s post below.
8 hours on the production line. ๐ค
Xiaomiโs full-size humanoid robot takes on an 8-hour task at the Center Console Side-Cover Sorting Workstation at Xiaomi Auto Factory.
Beyond sustained operation, it can autonomously detect and correct anomalies while coordinating with the production line system โ another step toward real-world industrial deployment.
Xiaomi (@XiaomiTech_) says its full-size humanoid robot took on an 8-hour task at the Center Console Side-Cover Sorting Workstation at Xiaomi Auto Factory, and that it can autonomously detect and correct anomalies while coordinating with the production line system; Xiaomi calls it “another step toward real-world industrial deployment”.
Where the Money Is Going
Specialist Chinese startups X Square, Galaxea and Spirit AI have each reached valuations of roughly $3 billion, according to company disclosures cited by Reuters. “We want robots to shift from being driven by code to being driven by common sense”, X Square co-founder and COO Yang Qian said.
Hardware makers are spending on AI 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, Reuters reported. By our arithmetic, that one package is about 43% of the average first-half deal. Unitree has about 30 researchers developing AI that predicts how objects behave, according to a person briefed by the company.
How Chinese Firms Gather Robot Data
Unlike chatbots, robots need examples of physical interactions. 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, Reuters reported.
Ant Group’s robotics unit Robbyant said it trained a model on 160,000 hours of material, including 50,000 hours of human-guided robot movements, and trained software on 20 robot configurations from 17 brands. DexForce uses simulated environments; its COO Nicole Yang said some industrial tasks can be prepared for deployment within 6 to 8 hours, with success rates above 99.9% in specified precision operations.
“Data quality is more important than data diversity, and more important than simply collecting data at massive scale”, X Square’s Yang said.
Named Deployments
Battery maker CATL said Spirit AI robots connect flexible wires during battery-pack testing at its Zhongzhou plant, with insertion success above 99%. Robbyant said its robots handle orders in two Shanghai pharmacies, with staff checking medicines before dispatch.
Chips and IPO Headwinds
Reuters lists shortages of advanced Nvidia chips under US export controls, and a squeeze on IPO funding amid regulatory concerns over inflated valuations and overcrowding. As robots become more powerful, the chip shortage will become a serious problem for the industry, said Zhu Mingxuan, who leads development of JoyIn Technology’s Aether robot AI.
Unitree founder Wang Xingxing said last month the company had tested OpenAI’s GPT-6 on robots with encouraging results, Reuters reported. Wang said he would consider robotics’ “ChatGPT moment” reached when robots can complete 80% of tasks in 80% of unfamiliar environments from spoken instructions.
The US Comparison in the Report
Reuters names Physical Intelligence, Skild AI and Generalist as US startups building software to teach robots skills without extensive retraining. Generalist says its GEN-1.5 model completed around 60% of simple tasks after viewing a demonstration lasting 3 to 12 seconds. X Square has published research showing robots acquiring handling skills from a single human video.
“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. On readiness he added: “If the question is when a customer can count on it, I would add years, not months”.
How Each Side Reads These Figures
For readers who see momentum, the figures include five times the funding of a year earlier, three specialists valued at roughly $3 billion each and named deployments at CATL and in Shanghai pharmacies.
For readers who see caution, the figures include an average deal of about $40 million spread across 322 deals, chip shortages under US export controls, a squeeze on IPO funding, and a consultant who adds “years, not months” before customers can rely on the systems. We cite no other analysts’ views here.
As we read the report, the money and the hiring point the same way: three AI specialists valued at roughly $3 billion each, and a UBTech chief-scientist package worth about 43% of an average first-half deal, by our arithmetic.
As we read the data section, the Chinese firms Reuters names are competing on how they collect physical training data, from Spirit AI’s roughly 1,000 recorders to Robbyant’s 160,000 hours and DexForce’s simulations.
Dermot McGrath, founder of ZenGen Labs, to Reuters
“concluded that the margin won’t come from the hardware but from the brain”
Three Implications
FOR INVESTORS ITjuzi’s data, as reported by Reuters, shows 322 first-half deals averaging about $40 million each, by our arithmetic, alongside a squeeze on IPO funding that Reuters also reports.
FOR ROBOT MAKERS UBTech advertised up to 124 million yuan for a chief scientist, and Unitree has about 30 researchers on AI that predicts how objects behave, Reuters reported.
FOR US LABS Reuters names Physical Intelligence, Skild AI and Generalist as the US comparison, while Chinese developers face shortages of advanced Nvidia chips under US export controls.
The framework opens: “The transformer is a commodity now; what changes between paradigms is where intelligence accumulates and who can capture it.” It names a fourth moat, “The Container Moat”, built by closed data loops inside a customer’s environment.
As we read it, the Reuters report maps onto that question: Chinese developers say they hope access to real-world training data will help them, and the deployments named, at a CATL plant and in Shanghai pharmacies, are inside customers’ sites. The framework does not discuss robotics firms or ITjuzi’s data.
Every funding figure here is ITjuzi’s data as reported by Reuters on 2 November 2026. We could not read ITjuzi’s own report, so its definition of an embodied AI company and of a deal is not set out here. The first-half 2025 figure is our arithmetic from Reuters’ “five times” and is approximate.
An average per deal says nothing about how the 93.5 billion yuan was split; a few large rounds can lift it. The valuations come from company disclosures cited by Reuters, and the success rates are the companies’ own. We did not contact ITjuzi or the companies.
This Week In The AI Supercycle
Anthropic’s Reality Check, The Capital Squeeze, & The New World Order
This week’s issue: for two years the AI build-out was priced as a growthstory. This week it started being priced as a credit story.
Chinese embodied AI companies raised 93.5 billion yuan ($13 billion) across 322 deals in the first half of 2026, five times a year earlier, according to ITjuzi data reported by Reuters. By our arithmetic that is about $40 million per deal, and Reuters reports that more of the money is reaching developers of robot intelligence alongside manufacturers.
A note on sourcing. Funding figures are ITjuzi’s data as reported by Reuters on 2 November 2026; we read the Reuters text through its syndications and could not read ITjuzi’s own report. Valuations and success rates are company statements as reported by Reuters. Per-deal averages and the implied 2025 figure are our arithmetic.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
