Just now, Sweet Potato Robotics raised another 800 million yuan.

The robotics industry is heating up again.

Touzhongwang learned that Digua Robotics has completed a $120 million Series B1 financing. Following a $100 million Series A financing in 2025, the total financing of Series A and Series B for Digua Robotics has reached $220 million.

This round is a standard club deal, with more than 15 new investors. Their backgrounds are clearly divided into two major camps: industrial capital and financial capital. The industrial investors include leading industrial capital such as Synstellation Capital, Didi, and Meituan Longzhu. Strategic investment institutions include Borui Capital, Joyoung Family Office, Yongning Gaoxin, BAIC Capital, Jiukun Venture Capital, Xinlian Capital, and Yare Capital.

Financial investors include first - tier institutions such as Jinqiu Fund, Xingrui Capital, Chuxin Capital, Gengxin Capital, and Yunbai Capital. At the same time, old shareholders such as Hillhouse Ventures, Vertex Growth Fund under Singapore's Temasek, Linear Capital, Hexuan Capital, Huangpu River Capital, Wuyuan Capital, and Plum Ventures have all over - subscribed.

Digua Robotics was originally the robotics division of Horizon Robotics, and its predecessor was the AIoT division of the company. According to public interviews with the enterprise, the opportunity for independence came in 2023 when GPT emerged, which reconstructed the technical pain points in the robotics industry such as insufficient intelligence and scenario implementation. Realizing the trend of mutual penetration between robots and large - scale models and how to better reuse the technology and resources accumulated by Horizon in the field of robotics, Digua Robotics was officially spun off from Horizon in 2024.

Different from most robotics companies in the market, Digua Robotics positions itself as a "general base provider" for robot software and hardware. Generally speaking, its main business is not to develop robot bodies independently, but to be a "shovel seller" that provides various toolkits for all kinds of robot products from R & D, mass production to application, covering all aspects from chips, algorithms to software. The service scenarios cover almost all current types of robots, including humanoid robots, wheel - legged robots, quadruped robot dogs, service and companion robots, logistics AMRs, etc.

In a few interviews with the founder, Wang Cong, the CEO of Digua Robotics, described Digua's positioning as "the greatest common divisor for robot implementation", which is in line with Horizon's way of doing things - competing in areas where there is no competition.

At present, when embodied intelligence is booming and the technology has not yet converged, Digua Robotics is at a sufficiently scarce and certain ecological bottleneck. This may be the core reason why Digua Robotics is favored by so many investors.

A club deal with extremely high industrial concentration

Since its establishment, Digua Robotics has had a smooth financing process.

In May 2025, Digua Robotics completed a $100 million Series A financing, jointly invested by institutions such as Hillhouse Ventures, Wuyuan Capital, Linear Capital, Hexuan Capital, Jiuhe Ventures, Vertex Growth, Lisi Capital, Dunhong Assets, Feidian Capital, Plum Ventures, and Huangpu River Capital. Among them, Hillhouse, Wuyuan, Linear, and Huangpu River Capital were early or even the first - round investors of Horizon.

In December of the same year, Digua Robotics completed a financing of hundreds of millions of dollars, jointly participated by multiple US - dollar funds and strategic investors. The clearly disclosed investor is a Middle - Eastern capital, Prosperity7 - the fund under Saudi Aramco's venture capital.

This round continues the luxurious investor lineup. One of the characteristics is that the concentration of industrial capital is very high, accounting for a large proportion, and most of them are old players with experience in the field of embodiment.

Synstellation Capital has no official website yet. According to industry clues, it is very likely a hard - technology industrial investment platform established by Zheng Zhigang after stepping down. Since the beginning of the year, it has participated in the new $2 billion financing of Qianxun Intelligence and a new round of financing for the critical point of dexterous hand robots.

Borui Capital belongs to the "CATL - related" investors. It was founded in 2024 by Li Ping, the co - founder and vice - chairman of CATL. It invests at an early stage. In 2024, it invested in the Angel + round of Qianxun Intelligence and increased its investment in two more rounds. In 2025, it invested in the seed round of Vita Power.

BAIC Capital is the CVC platform under BAIC Group and has been quite active in the field of embodiment in recent years. Last month, it participated in the $1 billion Series B financing of Xinghaitu. In November 2025, it participated in the $1 billion Series A+ financing of Xingdong Jiyuan. Even earlier, BAIC Capital invested in Yinhe Tongyong, Zhiyuan Robotics, Pasini, Songyan Power, and Pudu Robotics.

Xinlian Capital is the CVC platform under the listed chip company Xinlian Integration and has invested in Yingshi Robotics with dexterous hands and Xingyuanzhi Robotics.

Yare Capital is a science and innovation platform initiated by Professor Wang Tianmiao of Beihang University. It focuses on the early - stage incubation and investment in high - end manufacturing, precision medicine, and intelligent services. It has invested in dozens of AI and robotics projects, including Yinhe Tongyong, Xiaoyu Zhizao, and earlier companies such as Ninebot, Tianzhihang, Zhixingzhe, and Shenzhilan, covering multiple scenarios such as services, medical care, unmanned driving, and underwater robots.

Joyoung Family Office is the investment platform under Joyoung Co., Ltd. and has been exploring and deploying robots in food service scenarios such as kitchens and households. Yongning Gaoxin is an industrial fund jointly established by Ningbo Tongshang Holding Group and the High - tech Zone Management Committee. The local area has a solid foundation in the semiconductor and intelligent manufacturing industries.

There is also a special one - Jiukun Venture Capital, an early - stage technology investment platform established by Jiukun Investment, a leading quantitative private equity firm. It participated in the angel round of Yuanli Lingji in February last year.

It is not difficult to see that in addition to funds, the above - mentioned investors also have a large number of embodied scenarios and ecosystems. This practical screening has long been a common practice in embodied financing. In addition, with a quota of $120 million and nearly 20 investors, it is probably the result of the company's restraint. The financing sequence from market - oriented US - dollar funds, international capital to industrial investors is almost the same as that of Horizon back then.

The "shovel seller" in the era of embodiment

In the just - past 2025, we have repeatedly reviewed the development of embodied intelligence. Among them, there are two key sets of data that are often cited:

One set is at the level of capital enthusiasm. According to the statistics of Touzhong Jiachuan CVSource, the entire embodied intelligence track completed more than 310 financings in 2025, among which more than 100 financings reached over 100 million yuan. If we refine the statistical data to "new start - up companies" established after 2024, as many as 47 embodied intelligence start - up companies completed financings of over 100 million yuan. Using the same criteria (established after 2024 and having completed a cumulative financing of over 100 million yuan) for a horizontal comparison with other current popular investment tracks, there are only 12 start - up companies in the low - altitude economy that meet the same criteria, 8 in commercial space, and only 3 in intelligent driving. Only artificial intelligence outperforms, with 107 such start - up companies emerging.

The other set is at the level of market participation. In 2025, the investors in embodied intelligence showed the characteristic of "diversification". Industrial capital, university funds, VC/PE, and government - guided funds, the main participants in the current capital market, all appeared in this round of investment wave. At the same time, these investors also showed a "high willingness to cooperate". There are many cases where investors with different capital attributes and investment strategies, such as large industrial players, VC/PE, and government - guided funds, appear in the same financing event.

The simultaneous appearance of these two sets of data largely confirms a core conclusion: Embodied intelligence is not just an emotional bubble driven by hot topics in the economic downturn cycle. As a cutting - edge field of deep integration of AI and advanced manufacturing, robot body manufacturing, as its core carrier, naturally has no shortage of entrepreneurial opportunities. At least because the technological evolution of "manufacturing" always follows the law of "alternating progress": the progress of the body drives the prosperity of upstream components, the prosperity of upstream components improves the performance of the body, and the high - performance body promotes the generation of incremental demand after entering specific scenarios, and the incremental demand further seeks the progress of the body.

The key problem is that the "entrepreneurial wave" of embodied intelligence in 2025 is also a continuation of the "large - scale model investment wave". A basic logic is that although artificial intelligence is an inevitable trend, when the model's capabilities can be converted into actual productivity and how much resources and capital investment are needed before forming actual productivity are all unknowns. As a horizontal comparison, embodied intelligence is a subset of AI's practical applications, a new productivity that will bring a new incremental market, and each actual application scenario is a rational decision of end - users.

In addition, the embodied intelligence track has a stronger "talent - intensive" nature. Compared with general large - language models with hundreds of billions of parameters, the parameter quantity of embodied intelligence models is usually only in the range of a few billion to dozens of billions, and the training and inference costs are significantly lower. The funds concentrated in the large - scale model and artificial intelligence tracks will naturally flow to all aspects of embodied intelligence, applying the overflowing model capabilities to interactive intelligence, motion intelligence, and operation intelligence.

Embodied intelligence has not yet grown into a manufacturing industry with a complete supply chain. Under this premise, the natural demand for "supply - chain collaboration" in the manufacturing industry has actually become a bottleneck for embodied intelligence. The most intuitive problem is that the lack of underlying infrastructure makes every step of progress extremely difficult. Most companies have to start from chip adaptation, optimize operators by themselves, debug drivers, and do system - level adaptation and optimization, consuming a large amount of R & D resources in "reinventing the wheel". The high proportion of underlying R & D investment directly slows down the overall iterative speed of the industry.

And this is an important reason why Digua Robotics has attracted much attention since its birth.

Digua Robotics' technical advantages can be broken down into three levels. The first is the accumulation of chip architecture. Horizon's Journey series uses the BPU (Brain Processing Unit) architecture, which is specially optimized for end - side neural network inference. In the embedded scenarios of real - time perception, planning, and control of robots, its performance in terms of power consumption, latency, and computing power utilization is significantly better than that of general - purpose GPU solutions at the same level. The computing power of Journey 6P reaches 560 TOPS, which can already support the parallel processing of multiple tasks of real - time perception, planning, and control of robots. More importantly, the strict requirements of automotive - grade chips in terms of functional safety, temperature adaptability, and service life give them natural advantages in the industrial robot scenario.

The second is the integrity of the toolchain. Horizon's largest hidden investment in the past decade is the full - stack software development platform built around the Journey chips, including the AI toolchain, simulation environment, and operating system adaptation layer. This toolchain enables automobile companies to quickly deploy algorithms on Journey chips, and the same capabilities are now being opened to robot developers. As an extension of Horizon's robotics business, the developer platform launched by Digua Robotics has built - in mainstream perception models, planning and control algorithms, and simulation test environments. Developers can directly call the underlying computing power without optimizing operators and debugging drivers from scratch.

The third is the transfer of mass - production experience. The real competitiveness of chips lies not in laboratory benchmarks, but in the reliability data and cost - control capabilities accumulated in tens of millions of shipments. Horizon's Journey chips have a cumulative shipment of over 10 million sets and have undergone adaptation and verification for multiple brands and models, which means that their hardware stability, software compatibility, and supply - chain maturity have all been tested by the market. This "mass - production gene" is difficult for most robot chip start - up companies to replicate in the short term.

In an interview in May 2025, Wang Cong, the CEO of Digua Robotics, clearly stated: "Digua Robotics positions itself as the software and hardware Infra (infrastructure) for robots... We have always had a judgment: In the future, the types of robots will be extremely rich... In the future, there will definitely be different types in each scenario, and there will definitely be a bunch of companies behind each type. There will be thousands of robot companies in the future. And such an industry form needs a company that does the underlying software and hardware Infra to support it, so that the industry can become better and bigger. Just like in the Internet industry, if there were no Amazon or Alibaba to do Infra, a team of twenty people would have to build their own servers. Digua Robotics positions itself as not making robot bodies, but serving all companies that make robot bodies."

Hexuan Capital, the first - round investor of Digua Robotics, also told Touzhongwang in a previous conversation: "The reason we finally decided to invest in Digua Robotics is that we believe that no matter how competitive the domestic robotics track is and how the industry pattern changes, all leading robot companies will eventually need domestic robot chips and development platforms for the sake of supply - chain security and cost. As the saying goes, 'In the gold rush, the real money - makers are the shovel sellers'."

This article is from the WeChat official account "Touzhongwang". Author: Pu Fan, Cao Weiyu. Republished by 36Kr with authorization.