_________

Published on Oct. 21, 2026

“My biggest mistake over the past decade or so was not believing enough in AI.” Wang Xingxing said this, founder and CEO of Unitree Robotics, arguably China's best-known humanoid robot maker.

His admission came during a roundtable at the China International Industry Fair in Shanghai on October 12, reported the Chinese financial news outlet Kechuangban Ribao (in Chinese).

This was also a tumultuous time for Unitree's shareholders. Since the company's listing on Shanghai's STAR Market on August 19, the company's market value fell by more than 281 billion Yuan (about US$42.0 billion). That was more than 60 percent from its post-IPO peak.

Such roller-coaster valuations of startups are a typical warning sign of a stock market bubble forming. Unitree was not the only case, just a very eye-catching one. All of them together set off alarm bells in Beijing.

Just three days before Wang's speech, the Communist Party's Central Committee and the State Council had released new guidelines on developing AI and other "new quality productive forces," cautioning against the kind of enthusiasm that inflated Unitree's stock.

"We must not abandon the real economy for the virtual one, blindly pursue unwarranted scale and blind expansion, or create economic bubbles," the document states. Officials at every level who cause major losses through blind investment will be held strictly accountable, the document warns. The original document has been published by the State Council Information Office (in Chinese).

Reading this only as a crackdown on an AI bubble would miss an important point. The guidelines seek to restrain speculative and wasteful investment while maintaining support for industrial AI. In the real economy, Beijing wants to continue the development of physical AI and its deployment in the manufacturing industry.

Point 10 of the same guidelines demands AI upgrades for traditional industries and more applications involving humanoids. It also asks for more core research, computing power, pilot bases and safeguards for reliable deployment.

Investment in general is also meant to continue. The guidelines encourage venture capital, government-backed funds, technology bonds and stock listings. Beijing wants “disciplined capital” to support productive uses of AI.
The Chinese government, in other words, has just reaffirmed its commitment to AI Plus.

While quite effective in pursuing its industrial policies in the long run, the Chinese system also has distinct shortcomings. One of them is that local governments compete with each other, often with little or no coordination, in the process temporarily creating needless overcapacities.

As the Chinese saying goes, “The mountains are high, and the emperor is far away.” In a country the size of a continent, asserting central power from the capital is no small feat. It has been like that for millennia. We are watching the embodied AI version of this age-old friction unfold in real time.

Local governments have been chasing robots and AI with the same enthusiasm as retail investors. In its accompanying explanation of the guidelines on October 9, the National Development and Reform Commission (NDRC) said some regions have pursued similar emerging-technology projects without regard for their industrial strengths, and it warned against “blind investment” and “reckless spending” for “duplicative construction.”

In traditional industries such as machinery, chemicals, textiles, and shipbuilding, Beijing still wants fast deployment of AI upgrades. For robotics and other emerging industries, it wants to see quality over quantity.
Most of these policy tools existed before October. What we observe is a recalibration.

The central government now wants to play a bigger role in selecting sectors, building testbeds, and providing computing power. This is state-directed AI adoption in practice.

Beijing still wants all that money to flow into industrial AI. Point 8 of the same guidelines names embodied intelligence among China's future industries and calls for “mechanisms to increase investment in future industries and share risks.” (in Chinese)

Overall, with this fine-tuning of prior regulations, Beijing is seeking to curb speculative excess and wasteful industrial investment while maintaining strong support for AI deployment and other new technologies.

Unitree illustrates why that combination matters. Expectations for its humanoids had run ahead of their ability to do what pundits like to call “real work.”

Part of Unitree's valuation problem was the gap between robots that could dance and box and their limited usefulness in factories, the Chinese financial commentary portal Dachang Caijing She wrote.(in Chinese)

It described a company that had excelled at cutting hardware costs while expecting to adopt outside AI models as they matured. Its robots “lacked a true AI soul,” it wrote. The spectacle had helped build public and investor expectations that their practical capabilities could not yet meet.

Wang sees AI itself as part of the solution. He said Unitree had, a few months earlier, proposed a development process in which AI writes code, trains models and helps test them on physical robots, with human feedback still required. “My faith in AI grows daily,” he recently told the roundtable in Shanghai. This suggests Unitree might from now on put more emphasis on the robot's brain.

Tests with GPT-6 have also made the robot maker more optimistic. In August, he had put a robotics breakthrough at two to three years away at the earliest; now he sees it as possible within one or two. He even raised the prospect of a breakthrough in 2027. “If GPT-7 comes out next year, it could directly crack the problems of embodied intelligence,” he said.

For Beijing and for Wang, the case for industrial AI remains intact. A better investment focus has to bring the machines closer to useful work in the factory.