The era when a 10-page PPT and a star-studded team could help embodied intelligence startups secure three rounds of financing within half a year is gone.
"We still review embodied intelligence projects, but we will not make new investments in this track anymore."
In October 2026, when asked about the investment heat of embodied intelligence, Chen Yuan, a well-known domestic investor, told Tencent Tech. He immediately gave a more pessimistic judgment: "Now AI investment in the primary market has entered a 'junk time'."
The institution where Chen Yuan works is one of the most aggressive early-stage funds in China that invest in embodied intelligence. At the end of 2022, he judged that the capabilities of large models would extend to the physical world, and since then he has continuously placed bets on robot bodies, embodied brains and world models. In the past, his team might discuss several embodied intelligence projects intensively in a week, but now there is not even one on average per week, and sometimes only one every few weeks. More energy is spent on supporting the enterprises that have already been invested in.
According to statistics from NE Times, in September 2026, 46 financing deals were closed in China's primary market for embodied intelligence, with a reference amount of about 116.24 billion yuan, down 4.2% and 40.1% respectively from August. This decline does not fully reflect the changes in the market — some projects that completed investment decisions as early as May and June often did not announce the closing until September. Chen Yuan estimates that if only the amount of investment that was truly newly approved by the investment committee in the current month is counted, it may only be 20% to 30% of the hot period in the first half of the year.
The changes in the old share market are more direct. Some investors told Tencent Tech that before August, the old shares of some leading robot companies "could be sold in minutes", and buyers could be found at about a 10% discount; recently, 50% discount quotations have even appeared in some transactions he contacted.
According to a report by Securities Times on October 29, in the past two months, the transfer quotations of old shares of some embodied intelligence companies have been concentrated at a 65% to 80% discount to the valuation of the previous round, and the sold targets include leading enterprises. Even if sold at a discount, some early shareholders can still get several times of returns, but buyers are not as active as before. Some investors said that only institutions that believe that star companies will be listed sooner or later are willing to take over the shares.
A sharp contrast is Agibot. A number of investors recalled to Tencent Tech that at the beginning of this year, Agibot did not accept any investment, and a large number of investment institutions could not get in even with money and connections. In July this year, Agibot announced that it had started preparations for listing on the Hong Kong Stock Exchange. But by October, Agibot began to push for a new financing of 10 billion to 15 billion yuan at a valuation of about 500 billion yuan. People close to the transaction told Tencent Tech that Agibot required investors to complete the closing of this round of financing as soon as possible.
A pessimistic view began to spread in the industry: there may be no new star embodied intelligence enterprises to be listed this year.
The market suddenly cooled down — as everyone had expected, but it was so unexpected. A group of investors who once scrambled to enter the market can't wait to exit.
But it's easy to stop placing bets, but it's not so easy to get back the money invested in the past few years.
01 Invested too much, exited too slowly
The bullets in the primary market have not been exhausted, but the originally smooth capital relay is failing. First of all, investment institutions have been struggling to find new projects with sufficient differentiation.
Zhou Heng is in charge of AI investment at a well-known industrial capital. From his personal perception, the investment in embodied intelligence has cooled down in May this year, mainly because the institutions at the forefront have almost finished their investment. Robot startups are still emerging, but many mainstream funds already hold three or four robot enterprises in their portfolios.
He told Tencent Tech, some people in the industry have even invested in 30 robot companies. "This means you have to convince the investment committee why this new project is a must-invest."
The price of projects is also getting higher and higher. Information obtained by Tencent Tech shows that both Zhisquare and Zibianliang Robotics have seen their valuations double from 100 billion yuan to 200 billion yuan in about four months. Xinghaitu's valuation rose from 100 billion yuan to 200 billion yuan in less than two months. An investor told Tencent Tech that in 2024, he looked at a project with an angel round quotation of 300 million yuan, but he did not invest. By 2026, this project claims to have a valuation of 400 billion yuan.
The speed of valuation growth has made it increasingly difficult for subsequent rounds of capital to calculate returns.
In the past few years, the high valuations of a large number of robot startups have been built on the expectation of entering the public market in the future and obtaining higher pricing. The listing of Unitree has provided a new pricing reference for the primary market. On August 19, Unitree Technology officially landed on the Sci-Tech Innovation Board with an issue price of 150.80 yuan. On the first day of listing, its intraday price once reached 1100 yuan, and the closing price was 845 yuan. However, after listing, Unitree's stock price soon experienced sharp fluctuations. By October 15, its intraday lowest point touched 404.03 yuan, a drop of more than 60% from the intraday high on the first day of listing; after that, the stock price rebounded rapidly, closing at 647.05 yuan on October 30, still about 41% lower than the intraday high on the first day of listing, with a total market value of about 2.617 trillion yuan.
"The best time to exit was before Unitree went public," said Xu Zheng. But he himself failed to seize this opportunity.
Xu Zheng is in charge of exit at a well-known industrial capital. In the past year, the institution he works for has invested in two robot companies, both with valuations of more than 2 billion yuan at the time of entry, and now both have exceeded 15 billion yuan. Calculated based on the book valuation, these are two investments with very considerable returns, but a few months ago, Xu Zheng was persuading his colleagues to consider selling some shares, but his colleagues thought they could wait a little longer, hoping that the company's valuation would continue to rise.
What Xu Zheng is worried about is that when everyone wants to sell, there may be no buyers left. "Everyone wants to withdraw when you want to withdraw."
What makes some institutions more worried than the fluctuations in the secondary market is whether robot enterprises can be listed as scheduled.
On October 13, Caijing reported that Agibot Innovation is pushing for a new financing of about 10 billion to 15 billion yuan, with a latest valuation of about 500 billion yuan. Agibot previously announced in July that it had started preparations for listing on the Hong Kong Stock Exchange. However, according to the rules of the Hong Kong Stock Exchange, if the pre-listing investment is not completed 28 full days before the initial submission of the listing application, and the investor still holds shares at the time of submission, or is completed on the day of the initial submission and before listing, the Hong Kong Stock Exchange will generally require the first trading day of the applicant's shares to be postponed to 120 full days after the completion of the last pre-listing investment, except in very special circumstances. This means that after Agibot continues to raise financing, it is very difficult to complete the IPO this year.
A person engaged in Hong Kong stock listing services told Tencent Tech that in the first quarter of this year, among the three enterprises with the highest valuations in the embodied intelligence industry at that time, except for Unitree, the other two were already pushing for the submission of listing applications to the Hong Kong Stock Exchange. But by October, the listing processes of the two companies had not met previous expectations. "Both are stuck."
Another Hong Kong stock listing transaction lawyer told Tencent Tech that Chapter 18C of the Hong Kong Stock Exchange provides a listing path for qualified special technology enterprises, where enterprises that have not yet commercialized do not need to meet the revenue threshold of HK$250 million for the latest audited year of commercialized enterprises, but need to disclose a credible commercialization path, including how to reach this revenue level in the future. She told Tencent Tech that if robot companies have relatively little existing revenue, the Hong Kong Stock Exchange will pay further attention to the number of customers, orders on hand, commercialization plans and future profit paths. The high-growth forecasts previously shown by enterprises to investors must be supported by more actual operating data at the IPO stage.
What is more sensitive is the quality of revenue. The above lawyer told Tencent Tech that a project was found to have circular contracts during the due diligence stage, and the sponsor finally stopped pushing it forward. For robot enterprises, if the companies that submit applications first are found to have major problems in customer authenticity, related party transactions or revenue recognition, subsequent enterprises adopting similar business models may also face more rigorous verification. In particular, for procurement and data transactions formed around data collection centers and local industrial platforms, the sustainability of revenue and the independence of customers will become issues that need to be explained.
She said that some robot companies are eager to go public, in addition to the founding team's own desire to speed up the progress, it also comes from the pressure of existing investors. Some investment agreements will stipulate the listing period, for example, requiring the enterprise to complete the IPO within 18 months. Well-known investment institutions behind star enterprises often have stronger bargaining power and put forward stricter requirements on the company's listing time.
02 The once-consensus view
In the past three years, embodied intelligence can be regarded as the track with the strongest wealth concentration effect in the primary market.
Lin Chuan did not initially favor embodied intelligence. The RMB fund he works for used to focus mainly on semiconductors and AI, and he himself came from semiconductor investment. In 2024, he thought the popularity of robots was "completely incomprehensible".
But by 2025, after more and more institutions began to lay out, he found that if his fund had never invested in robots, he would instead need to explain the reason to LPs. "Everyone around me is investing in embodied intelligence. If you don't invest in a single embodied project, the difficulty is too great."
He finally found a reason that could convince himself: if robots really become a huge market, what kind of chips, computing power and controllers will they need? For a semiconductor investment institution, studying the development of downstream robots is also studying the future demand for semiconductors.
The FOMO sentiment swept the whole industry. The confusion of investors is that no matter whether they understand it or not, if this is the next-generation AI platform and they do not lay out in advance, they may be abandoned by the times.
The annual statistics released by IT Juzi show that in 2023, there were 42 domestic embodied intelligence financing deals with an amount of about 86.76 billion yuan; in 2024, the number increased to 103 deals with an amount of about 93.55 billion yuan. In 2025, the number of financing events reached 325, with an amount of 398.32 billion yuan, an increase of about 216% and 326% respectively compared with 2024. In this statistic, early-stage investment still accounts for 62%. A large number of startups have obtained financing of hundreds of millions of yuan or even larger scale while they are still in the stage of R&D and technical verification.
Leading enterprises began to attract large industrial capital. Robot investment has gradually expanded from the technical bets of early institutions to large-scale capital competition involving Internet giants, manufacturing leaders, national and local industrial funds.
Capital did not flow evenly to all startups. Statistics from Inno-Tide Jiachuan show that among 168 invested embodied intelligence enterprises in 2025, the top ten companies including Zibianliang, Tushi ZhiHang, Leju Robotics, Xingdong Jiyuan, Xinghaitu and Zhongqing Robotics obtained a total of 134.72 billion yuan in financing, accounting for 40.95% of the total annual financing in this statistical caliber.
By 2026, some enterprises have grown large enough. Statistics from Zhidx on June 29 this year show that there are at least 26 enterprises in China's embodied intelligence field with a valuation of more than 100 billion yuan, 16 of which newly entered this echelon in the first half of 2026. At present, the valuations of at least seven leading enterprises including Agibot, Galaxy General, Xinghaitu, Qianxun Intelligence, Zibianliang, Zhisquare and Lingxin Qiaoshou have reached more than 200 billion yuan.
A project may have a valuation of 5 billion yuan one month, become 10 billion yuan the next month, and reach 15 billion yuan the month after.
An early-stage investor told Tencent Tech that she once contacted a robot startup. The founding team initially gave a financing valuation, and after evaluation, the investment institution thought that the project was worth investing in, but the valuation could be higher, so they took the initiative to suggest the founder to raise the quotation and helped the enterprise push the valuation up.
In the usual equity investment transactions, investors will try their best to get a lower price to obtain more shares. But at the hottest time of embodied intelligence, the financing price itself became part of the project's value. A higher quotation may make the enterprise look more attractive among a large number of star projects, and the price of the next round of financing becomes the proof of the rationality of the previous round of valuation.
At that time, for many people, the primary market was full of abundant capital, and the listing performance of some AI model and GPU companies gave robot startups an imaginable exit price. Even though some people had begun to question the valuation, new buyers kept appearing in the market.
03 100-billion-yuan valuation needs new proof
The enthusiasm of capital is also changing entrepreneurship itself in turn.
An entrepreneur with a background in the autonomous driving industry told Tencent Tech that in the first half of this year, he intensively contacted investment institutions and tried to prepare for the establishment of a robot company. At that time, VC friends generally told him that if you couldn't get 100 million yuan in the first round, you didn't need to start this business.
At that time, he had not even determined the specific entrepreneurial direction. In order to get the "entry ticket", he spent several months looking for partners, hoping to invite a technical guru to start a business together, but in the end he failed to form an ideal team.
During this period, a listed company took the initiative to propose cooperation with him. He recalled that out of the need for market value management, the other party was willing to support him to set up a robot company, and even had products ready for display. "Their company has already purchased products from a certain robot company. If I am willing, I can change the shell and logo, and use them directly for my new company."
An industry consensus is that the capital market's pursuit of the robot concept has spawned some operations that have nothing to do with technology R&D and real orders.
In September 2026, Shao Tianlan, founder of Mech-Mind, publicly questioned some data collection centers and related transaction models in the industry, triggering a debate on the authenticity of revenue in the embodied intelligence industry. This revenue quality dispute has also entered the vision of investment institutions. After the financing speed drops, investment institutions begin to re-examine the commercialization capabilities of robot companies. Enterprises that used to be able to obtain high valuations relying on the founder's background, technical demonstrations and financing progress now have to answer questions about order delivery, customer payment collection and sustainable revenue.
More and more people are beginning to feel the trend of industry reshuffle. On October 23, National Business Daily reported that Nanjing-based embodied intelligence enterprise Wujing Power was caught in a wave of wage arrears and shutdown. The company was established in February 2026. Local media previously reported that about four months after its establishment, Wujing Power had announced that 110 million yuan of financing was in place, signed framework orders of nearly 500 million yuan with a number of central SOEs, and had another nearly 500 million yuan of reserved orders in the local market.
But by October, some employees reported to the media that the company had been in arrears with salaries since August, and later stopped paying provident fund and social security, involving about 30