Heiße Gelder strömen ein, und kleine und mittlere Kapitalbetriebe können sich nicht an den Tisch der Embodied Intelligence setzen.
Image source
In March this year, Zhu Xiaohu, a partner at GSR Ventures, made remarks that were seen as a “wholesale” bearish view on humanoid robots, sparking intense discussions in the market. The debate about the ebbing tide of investment in embodied intelligence has been heating up. Many believe that the commercialization cycle of this sector is long, and the cost - effectiveness of investment at the current stage is getting lower.
However, in reality, capital activities have been gradually heating up. Since 2025, leading companies in the field of embodied intelligence have continued to receive financing. Unitree Technology completed its Series C financing, with its valuation exceeding 10 billion yuan. Galaxy Universal received a new round of financing of 1.1 billion yuan led by CATL in June.
Meanwhile, tech giants such as JD, Meituan, Tencent, and Huawei have also turned their attention to this field. In addition to investing in startups, big - name companies and their former senior executives have also actively entered the arena.
Tech Planet found through the IT Juzi venture capital data service platform that as of July 3, 2025, there have been 114 investment events in the domestic embodied intelligence industry this year, exceeding the total of 91 in the whole year of 2024.
On one hand, the “ebbing tide theory” is widespread; on the other hand, leading companies are continuously getting financing. Investment institutions that have not yet entered the market are in a dilemma: they are worried that the commercialization cycle of embodied intelligence is too long and the money invested will not yield returns in the short term. At the same time, they are afraid of missing out on the opportunity of the industry's explosion and can only watch early entrants reap the benefits.
Therefore, from industry giants to startup teams, from investors to entrepreneurs, all are trying to find a balance between risks and opportunities in this “gold rush” full of uncertainties. An investment manager from an early - stage VC firm told Tech Planet frankly, “The speed of technological iteration and the path of commercialization of embodied intelligence projects are full of unknowns, but no one wants to completely withdraw before the industry explodes.”
Embodied Intelligence: A Re - evaluation of Value
Embodied intelligence usually refers to an “intelligent system” with the abilities of environmental perception, logical reasoning, and interaction with the physical world. Embodied intelligent robots can be regarded as the physical carriers of this system.
This concept can be traced back to 1950. However, due to long - standing technological bottlenecks such as sensor accuracy and computing power costs, its development has been sluggish. It was not until the breakthrough of large - model technology in recent years that opportunities for the iterative upgrade of embodied intelligence emerged.
A senior industry insider said that the capabilities of large - model technology are intuitively reflected in products, making robots more intelligent, more flexible in movement, and having stronger motion control abilities.
This technological upgrade first attracted wide attention during the Spring Festival Gala in the Year of the Snake. Unitree Technology's humanoid robots performed on stage with dancers. They not only smoothly completed complex movements such as waving handkerchiefs and doing yangge dances but also adjusted their postures according to the dance rhythm, showing human - like interaction abilities. This was also the first time that embodied intelligence appeared in the public eye in an “anthropomorphic” form.
This also became a critical moment in the investment circle. The above - mentioned VC investment manager revealed that after Unitree Technology attracted attention through its performance at the Spring Festival Gala, the valuations of some embodied intelligence projects soared.
An investor from another investment institution mentioned that the market trend triggered by Unitree Technology directly doubled the valuation of an embodied intelligence project he had been tracking. “However, in fact, the commercialization path of this project is not clear.”
Driven by this industry boom, the financing pace of embodied intelligence startups has significantly accelerated. For example, Tashi Zhihang received a $120 million angel - round financing just 50 days after its establishment. Yuanli Lingji, which was founded less than half a year ago, also completed a 200 million - yuan angel - round financing. Qianxun Intelligence, which has been established for about a year, successfully completed a 528 million - yuan Pre - A round financing.
The valuations of some star embodied intelligence companies have continued to rise during multiple rounds of financing. For instance, after Zhipu Robotics, founded in 2023, completed its Series B financing led by Tencent in March this year, a third - party research institution estimated its valuation at 15 billion yuan, directly doubling that of the previous round of financing.
The above - mentioned investor from the investment institution frankly said that even for embodied intelligence projects in the startup stage, the investment threshold has risen. Financing scales of hundreds of millions of yuan are far beyond the participation capabilities of small and medium - sized investment institutions.
With such a high capital threshold, leading capital has become the dominant force in the industry. According to incomplete statistics from Tech Planet, top VC/PE institutions such as Hillhouse Capital, Qiming Venture Partners, and Sequoia Capital, the strategic investment departments of Internet giants such as JD, Tencent, Meituan Longzhu, and Ant Group, as well as industrial capital such as CATL, Lenovo, and TCL Capital, are frequently involved in the embodied intelligence investment track.
Betting on the Technological Inflection Point
During an investor exchange at the end of March this year, Zhu Xiaohu bluntly pointed out that the commercialization path of humanoid robots is still unclear. He also revealed that when an embodied intelligence project exited at a valuation of $300 million in its Series B, the company's annual revenue was only $8 million, and the cash - flow gap was as high as $27 million.
These remarks sparked a lot of controversy in the venture capital circle. However, from the current commercialization progress of embodied intelligence, his view shows relative rationality. A senior industry insider frankly said that even leading companies have not yet explored a scalable and profitable business model.
According to Unitree Technology's official website, its current product portfolio includes the Go series of quadruped robots, the B series of industrial - version quadruped robots, the G/H biped humanoid robot prototypes, and modules and software (vision, controllers).
According to Tianyancha information, Unitree Technology's winning bids are mainly for universities, energy, medical, and fire - fighting fields. Among them, scientific research universities are the main source of orders. For example, Tongji University contributed the largest order amount in Unitree's publicly announced winning bids, reaching 8.2566 million yuan. The procurement content was 10 H1 2 general humanoid robots priced at 700,000 yuan each.
The above - mentioned investor said that Unitree Technology's current customer structure reflects the concentrated implementation of embodied intelligence technology in scientific research scenarios. It also shows that there is still a lag in its large - scale application in the consumer and industrial sectors. At the same time, he said that Unitree has achieved mass delivery, but most players in the track are still in the R & D stage, and only a few companies have the ability to achieve mass production and delivery.
This is closely related to the current bottlenecks in embodied intelligence technology. “There are application prospects in fields such as medical care, elderly care, retail, and industrial manufacturing. However, given the current hardware performance and software algorithm levels, the development of the industry is bound to be a gradual process,” an industry insider pointed out.
He also said that for investment institutions, this situation actually presents new opportunities. “As the valuations of whole - machine companies are getting higher, institutions can turn their attention to the upstream industrial chain of humanoid robots.”
Since this year, many component companies have indeed received financing. Lingxin Qiaoshou recently completed a Series A financing of hundreds of millions of yuan, led by CICC Capital's private equity and participated in by institutions such as Ant Group. Qianjue Robotics, an intelligent sensor company, received investments from Yuanhe Origin, Gobi Partners, and Xiaomiao Langcheng.
As the financing boom in embodied intelligence extends to the upstream industry, an industrial investor said that what he is actually investing in is not the current orders but the critical point of the integration of large models and robots. “Once the breakthrough is made, the value of the entire industrial chain will be re - evaluated.”
From Capital Frenzy to Commercialization Test
Before the financing storm in embodied intelligence, large - model entrepreneurship was the hottest track in the capital circle. From 2023 to 2024, a “battle of a hundred models” took place in the domestic large - model field. However, just one year later, this battle has subsided. Several of the “six little tigers” of domestic large models have begun to adjust their strategies and transform.
Embodied intelligence is indeed replicating the capital frenzy at the early stage of large - model entrepreneurship. They share similar technological imaginations, similar valuation bubbles, and unproven commercialization puzzles.
However, against the backdrop of continued pressure on investment activities in the equity investment market, investment institutions are becoming more cautious. The number and scale of investments have generally declined, and investors are also becoming more rational.
At a public event in mid - May, Ma Lin, the managing partner of SEEFUND Unlimited Fund, pointed out that in 2025, the most important thing for embodied intelligence startups is PMF (Product - Market Fit). They need to have cash flow and revenue. “Even if it's not profitable, it at least proves the market's acceptance of you.”
Recently, Songyan Power became the focus of the humanoid robot startup track due to its excellent performance in the humanoid robot half - marathon. According to Jiang Zheyuan, the founder and chairman of Songyan Power, “Within one month after the Beijing Robot Marathon, we received more than 1,000 orders. As of now, the cumulative order volume has exceeded 2,000, with a total contract value of over 100 million yuan.”
Meanwhile, Galaxy Universal's humanoid robots have entered seven unmanned pharmacies in Beijing. The robots pick up medicines 24/7 and hand them over to delivery riders.
Similar to the logic of building valuations for large models, the investment in embodied intelligence is shifting from teams and technology to delivery. “This actually shows that capital has begun to focus on the ability of technology to be implemented in specific scenarios,” said the above - mentioned industry insider.
In this regard, an investor also said that this year is the first year of robot delivery. After embodied robot companies reach a delivery volume of 1,000 units, the scene requirements will no longer be limited to simple operations such as exhibitions. “At that time, companies will need to gather signing parties and users to jointly discuss issues such as requirements and costs. In this context, those who can deliver quickly and stably, ensure quality, and reduce prices will have an advantage in the competition.”
An institutional investor who has not yet entered the embodied intelligence field said that the concept dividend period of embodied intelligence has passed, and investment will be more cautious. In his view, referring to the industry cycle of intelligent driving and the development path of large models from parameter competition to scenario - based applications, the next investment window may appear in the mass - production stage. “Only then can high - quality targets that can truly integrate technology and scenarios be screened out.”
This article is from the WeChat official account “Tech Planet” (ID: tech618). Author: Ren Xueyun. Republished by 36Kr with permission.