150 Companies Crowd Humanoid Robotics, but Building Robots No Longer Valuable

For those still sharing humanoid robot funding news on social media,

this article may be worth a read.

Since the start of 2026, embodied AI funding has reached nearly 15 billion yuan, with 7 domestic companies now valued over 10 billion yuan—up from zero a year ago.

Yet over 77% of robot companies remain at seed, angel, or Series A stages, making this a game for a select few.

Money is concentrating: in 2021, China's humanoid robot sector raised 2.8 billion yuan; by 2025, that soared to 23.6 billion, an eightfold increase in five years. In 2024, there were 12 deals over 1 billion yuan; in 2025, that jumped to 28, including 8 over 2 billion.

Galaxy General raised 2.5 billion yuan in March 2026, setting a domestic record, led by the National Big Fund III with Sinopec, CITIC, and SAIC participating. The company has raised over 5.5 billion yuan in the past year, with a post-money valuation exceeding 20 billion.

Qianxun Intelligence founder Han Fengtao said bluntly: "2026 embodied AI will resemble 2023 large models. Without ample funding, model performance cannot reach the top, and there will be no seat at the table."

Orders do not equal productivity. UBTech's annual humanoid robot orders approach 1.4 billion yuan, and AgiBot delivered 5,000 units. But UBS analysts judge that over 50% of commercialization orders are essentially PR displays, not productivity replacement.

Efficiency data is equally harsh: UBTech's Walker S2 operates at only 45% of a human worker's efficiency, and AgiBot's bin transfer speed equals 0.7 people. Galaxy General's Galbot S1 can deploy autonomously at CATL's battery plant for 8 hours per charge, but this is top-tier, not the norm.

Factories need more than acrobatics; they need to keep pace with production rhythm. If the previous process delivers a part every 15 seconds, the robot cannot take 30 seconds. 9 out of 10 successes in the lab is excellent, but factories require thousands to tens of thousands of stable cycles, with automatic recovery from failures.

The technology divide is emerging. Unitree founder Wang Xingxing said at the 2026 World Robot Conference that the main reason robots have not scaled is insufficient efficiency and generality; new tasks require retraining. Motion can be built with hardware, but dexterous manipulation and generalization are the true Everest. AI models achieve near-100% success in fixed scenarios after specialized training, but when the environment changes, success rates plummet.

The government has stepped in. In June 2026, MIIT and SASAC launched a real-scenario training initiative, aiming to deploy over 10,000 humanoid robots in commercial settings by year-end. The policy is clear: rather than tweaking parameters in labs, throw robots into real factories.

The key to surviving the elimination round is not technology, but mass production capability and cash flow management. Tesla is converting Xpeng's production line for Optimus, targeting annual capacity of about 70,000 units at Fremont. Xpeng's robot completed a first funding round of over $900 million, with a post-money valuation of about 43 billion yuan. Capital backs this valuation because Xpeng provided a clear mass production timeline: IRON mass production by the end of 2026.

Companies still relying on PPT funding may meet their end in 2026. A Silicon Valley humanoid robot company had three successful funding rounds, with OpenAI on its client list, yet disbanded its team and refunded investors in November 2025, on the eve of mass production. It had orders, technology, and customers, but still died.

The reason: mass production requires systems engineering capability, not single-point breakthroughs.

One question remains: as policies establish entry and exit mechanisms, and capital only plays with companies valued at over 10 billion yuan, what happens to the 100-plus robot companies still lingering at Series A? Will they go bankrupt, or survive in a "low-speed operation" state—alive but with no chance? This question is left for the end of 2026.