The most luxurious industrial wheel of the year has been born.
Rarely do we see such a luxurious industrial profile —
Investment community sources have learned that Jiangxing Intelligence, a physical AI model company, has announced the completion of hundreds of millions of yuan in a Series B++ round of financing. The round was led by Chaoxi Capital, with participation from Chendao Capital, an industrial investment institution affiliated with CATL, JinkoSolar, Yankuang Capital of Shandong Energy Group, and Xiamen C&D Emerging Investment. Existing shareholder Zhuoyuan Asia continued to increase its stake.
In the past, the venture capital circle was mostly dominated by financial investments. But this time, industrial players have collectively entered the game.
Everyone has jointly placed their bets on a Tsinghua - affiliated team. Academician Liu Jiangchuan, the founder, graduated from the Department of Computer Science at Tsinghua University in 1999. Pang Haitian, the chairman and CEO, graduated from Tsinghua University as an undergraduate and later obtained a doctorate from the Department of Computer Science at Tsinghua. As early as 2018, the master - apprentice duo joined hands to start a business, and Jiangxing Intelligence was born.
Eight years have passed in a flash. Jiangxing Intelligence has witnessed the trend of AI entering the physical world, and more and more investors are flocking to it.
A post - 90s doctor leads the way to build the "brain" of the physical AI world
The story of Jiangxing Intelligence begins with Dr. Pang Haitian, who was born in 1993. He graduated from the Department of Automation at Tsinghua University as an undergraduate and obtained a doctorate from the Department of Computer Science at Tsinghua University at the age of 25. With both hardware foundation and software capabilities, such a composite background laid the groundwork for his future physical AI entrepreneurship.
Back in 2018, during his doctoral studies, Pang Haitian participated in an exchange program in Canada. Academician Liu Jiangchuan was his supervisor at that time. The outside world may be a bit unfamiliar with the low - key Professor Liu Jiangchuan. He graduated from the Department of Computer Science at Tsinghua University in 1999, obtained a doctorate from the Hong Kong University of Science and Technology, and was successively elected an academician of the Canadian Academy of Engineering and the Royal Society of Canada. He is an internationally renowned expert in the fields of multimedia networks and edge computing.
It was this experience that gave the two of them the idea of industrializing their scientific research achievements. "At that time, we saw that there was a huge amount of data in industrial scenarios, but its value had far from been fully realized. And the booming development of edge computing made everything possible," Pang Haitian recalled to the investment community. Subsequently, the two hit it off and, together with their laboratory classmates, founded Jiangxing Intelligence.
The team is young, but from the very beginning, they chose to immerse themselves in industrial scenarios. "At that time, we took our R & D colleagues to the site of a substation and stayed there for several months," Pang Haitian still remembers clearly. Due to the requirements of high precision, high intensity, strong anti - interference, and continuous operation in industrial sites, any slight change in factors could potentially lead to serious consequences. Thus, the Jiangxing team developed a gene of "growing from the site."
It can be said that it was in specific scenarios such as operating substation equipment at minus 20 degrees Celsius and inspecting high - risk coal mine tunnels that the company's team got close to the real needs of customers, and customers were able to truly see the improvement in safety and efficiency brought by AI technology. A milestone was in 2022 when the company's products passed the tests of the China Electric Power Research Institute, and since then, orders have been pouring in.
During this period, AI has evolved at an unimaginable speed. As everyone generally feels, AI has begun to move from the digital world to the physical world.
Pang Haitian has long noticed this. "Edge computing is like giving AI a brain and eyes in the digital world. Entering the era of physical AI is like giving it hands and feet." The significance behind this is that only when the capabilities of AI are gradually completed can it truly understand the physical world, and the era of interaction will come.
For Jiangxing Intelligence, based on its past industrial accumulation and the multi - source data fusion and industrial control capabilities in the edge computing era, the team has built a set of physical AI engines: Multi - modal spatial intelligent large model, Embodied intelligent operation model, and One - brain multi - body.
Specifically, this is a complete physical AI system from perception to execution. First is the Multi - modal spatial intelligent large model. Jiangxing Intelligence has achieved one - stop output from any multi - modal spatial information input to a three - dimensional spatial model. In the Embodied intelligent operation model stage, it endows AI with "flexible hands and feet" in the physical world. Finally, there is the One - brain multi - body — with the AI Agent Fleet as the core carrier, it can "manage multiple types of devices with one brain and complete large - scale operations collaboratively."
"The ultimate goal is to complete the business closed - loop." So far, Jiangxing Intelligence has upgraded from an edge computing solution provider to a global leader in physical AI technology. Moving from edge computing to physical AI is not a sudden change for Jiangxing Intelligence, but a natural technological extension. When data is collected and understood, the next step is bound to be execution.
Sequoia invests in multiple rounds, and the industrial round emerges: The CATL - affiliated investors are also here
Investors' real - money bets are the most indicative of the trend.
Looking back, Jiangxing Intelligence caught the attention of investors very early. In 2018, shortly after the company was founded, it received tens of millions of yuan in angel - round investment from Sequoia China. "At that time, we had just returned to China, and within just a few days, we received investment intentions from multiple institutions." The Tsinghua - affiliated team configuration made Jiangxing Intelligence a focus as soon as it was founded. "The investors made their decisions very quickly," Pang Haitian recalled.
It is worth mentioning that Sequoia has since become an important shareholder of Jiangxing Intelligence and has followed up with investments all the way. In 2019, Jiangxing Intelligence completed a 30 - million - yuan Series A round of financing, led by Songhe Capital, with Sequoia China and BV Baidu Ventures following. In 2020, it completed nearly 100 million yuan in a Series A+ round of financing, led by Poly Capital, with Legend Capital, Lingang Science and Technology Innovation Investment, and Shuimu Capital following. Existing shareholder Sequoia China continued to increase its investment.
At that time, in just two years since its establishment, Jiangxing Intelligence had achieved a complete closed - loop from technology R & D to product implementation, completed the implementation of multiple IoT hardware and software products and solutions, and achieved large - scale replication. It won orders from industry leaders such as the State Grid, China Southern Power Grid, Datang Group, Huadian Group, Jingneng Group, PetroChina, and Beijing Enterprises Water Group.
Over the years, Jiangxing Intelligence has basically maintained a financing pace of one round per year. In 2021, it completed a tens - of - millions - of - yuan Series A++ round of financing, led by Zhongguancun Development Qihang Fund, with Fuqi Capital and Zhuoyuan Asia following. Existing shareholders Sequoia China and Poly Capital continued to increase their investments. In total, Sequoia China has invested in Jiangxing Intelligence for four consecutive rounds, and its recognition is self - evident.
In 2022, despite the coldness in the primary market, Jiangxing Intelligence still successfully completed a 150 - million - yuan Series B round of financing, jointly invested by institutions such as Langmafeng Venture Capital, Legend Capital, Zhuoyuan Asia, and Guangyue Capital. In 2024, the Series B+ round of financing emerged, with Langmafeng Venture Capital, Zhuoyuan Asia, Songhe Capital, Legend Capital, Baidu Ventures, etc. making a joint strategic investment.
After communicating with multiple investors, they almost all mentioned a key point when talking about investing in Jiangxing Intelligence — the product closed - loop and implementation. The investment community obtained a set of data: In the past eight years, Jiangxing Intelligence has served more than 400 substations, more than 600 new - energy power stations, dozens of coal mines, oil fields, steel plants, chemical plants, and sewage treatment plants, as well as hundreds of kilometers of natural gas pipelines. The compound annual growth rate in the past three years has been nearly 70%. In 2025, the company's annual orders reached the level of 500 million yuan and it achieved profitability.
In the latest development, Jiangxing Intelligence has completed a Series B++ round of financing. The investors in this round have a strong industrial background. Chaoxi Capital is an industrial investment institution deeply involved in the energy and technology fields. Its industrial ecosystem is closely linked with dozens of leading enterprises and listed companies, including Longi, Canadian Solar, Jinko Solar, and Chint. Chendao Capital is an industrial investment institution affiliated with CATL, mainly focusing on investments in the mid - and downstream applications of the CATL industrial chain and the incubation of emerging scenarios.
JinkoSolar is a globally well - known solar technology enterprise. Yankuang Energy is the listed platform of Shandong Energy Group, and it is the only super - large energy enterprise in China with six listed platforms globally. Xiamen C&D Emerging Investment is a professional equity - asset management institution under Xiamen C&D Group, a state - owned logistics enterprise. As a long - term partner of Jiangxing Intelligence, Zhuoyuan Asia has supported the growth of a number of Tsinghua - affiliated artificial - intelligence and integrated - circuit enterprises to date.
It is understood that there is a detail in this process: The investment team specifically visited several provinces and conducted on - site inspections of scenarios such as wind farms and substations, and directly felt the on - site effects of Jiangxing Intelligence.
When talking about this investment, Wang Bin, a partner at Chaoxi Capital, frankly said that Jiangxing Intelligence leads the physical AI that truly "lives" in industrial scenarios, enabling industrial robots to evolve from "Big White" to "masters." This is an opportunity for Chaoxi Capital to layout industrial intelligence based on energy scenarios and has completed the business closed - loop. It is also the window period for the large - scale implementation of industrial robots that is opening. It is hoped that Jiangxing will become the underlying infrastructure for the intelligent upgrading of all industries.
Du Wei, the general manager of investment at JinkoSolar Holding, also shared that the Jiangxing team's forward - looking technical path and excellent execution ability in the field of edge AI and physical intelligence — its leading edge - cloud collaborative architecture and the "AI brain + robot" integration solution have verified the large - scale implementation ability in real industrial scenarios such as power/new - energy inspections, demonstrating profound industry know - how and strong productization potential.
In the view of Song Cong, the chairman and general manager of Yankuang Capital, "Facing the era opportunity of the deep integration of AI and the real industry, Jiangxing Intelligence has built a solid technical foundation and demonstrated excellent large - scale implementation ability in the power and energy fields."
Xiamen C&D Emerging Investment added, "The emergence of Jiangxing Intelligence has allowed us to see the rapid progress of AI in industrial implementation. Based on its excellent practices in the energy field, we also look forward to it bringing us surprises in more fields such as mines, chemicals, and rail transit." This once again verifies a simple truth: Every wave of technological innovation must ultimately translate into real value and tangible benefits for enterprises.
As an investor who has witnessed the growth of Jiangxing for a long time, Lin Haizhuo, the founding partner and chairman of Zhuoyuan Asia, was quite emotional. "Just as the spinning jenny was to the first industrial revolution, we highly recognize that Jiangxing Intelligence's 'general physical AI foundation' will help China become a global leader in the fourth industrial revolution."
An industrial synergy effect is emerging. Currently, Jiangxing Intelligence is exploring cooperation in products and business with CATL, JinkoSolar, and Yankuang Energy, and is looking for cooperation opportunities in intelligent inspections and operation and maintenance in the overseas market. As a result, more investors are on the way. It is reported that Jiangxing Intelligence's Series C financing is also close to completion.
Starting from industrial scenarios, the era of physical AI begins
Remember in July 2025, Jensen Huang, the head of NVIDIA, first clearly stated that the next wave of artificial intelligence is "physical AI." At the latest CES 2026 exhibition, he once again announced that the "ChatGPT moment" of physical AI is coming. Jensen Huang extends physical AI from robots/cars to industrial design and manufacturing, regarding it as the intersection of physical AI and the world's largest real - world industries.
This coincides with Jiangxing Intelligence's years of practice. In Pang Haitian's view, the large - scale implementation of physical AI is the focus in 2026. "We believe that whether it is for performances or scenarios for scientific research institutions, it is difficult to truly achieve large - scale implementation. Industry, on the other hand, is the scenario where large - scale orders can be most easily generated."
It should be noted that the industrial track is a typical scenario direction where "demand comes first." There has been a long - standing and rigid demand for robot replacement in industrial scenarios such as energy and power. Manufacturers not only need robots to have "eyes" but also require them to have higher autonomy to complete defect detection, process adjustment, and safety assurance with minimal human intervention, which precisely points to the core capabilities of physical AI.
So we can see that in recent years, many startups have shifted from looking for scenarios with robots to rooting themselves in the industry. This has created a unique barrier for Jiangxing Intelligence — from the very beginning, the team has been starting from the industry and scenarios.
However, although the scenario path is gradually becoming clear, there is still a long way to go for physical AI to truly achieve large - scale implementation. Pang Haitian frankly said that the most important and difficult problem to overcome currently is the data problem based on real physical scenarios. Different from large - language models that can extract rules from the long - accumulated language data on the Internet, the experience in the physical world does not exist in any past databases and can only be generated through interaction with the real world. Therefore, who can directly access the front - line sites and master more data from the physical world becomes the key in this race.
Looking back at Jiangxing Intelligence, the company has currently built an industry - leading three - dimensional integrated data foundation of "real - scenario data + real - machine operation data + world - model simulation data": Real - scenario data strengthens industry adaptability, real - machine operation data supports Sim - to - Real migration, and simulation data shortens the model learning cycle.
For Jiangxing Intelligence, real - scenario data and real - machine operation data form the basic foundation of its data acquisition. These two types of data come from real operating scenarios, have higher physical authenticity and multi - modal detail restoration, but are difficult to obtain, with a small quantity and high value. And Jiangxing's commercial implementation has made these data easily accessible — its thousands of industrial projects, covering core tracks such as power and energy, have accumulated 100,000 hours of real - time production data per day.
More importantly, different scenarios often have different types of physical data, which to some extent causes the Scaling Law to encounter obstacles. Jiangxing Intelligence covers the capabilities of all three types of physical AI data, making data acquisition faster, more cost - effective, and of higher quality. Finally, a data flywheel of "data collection - model training - iterative distribution - data regeneration" is formed, the more it is implemented, the smarter it becomes; the more it expands, the more efficient it gets.
Witnessing the wave of physical AI, everyone is looking forward to general - purpose robots entering every household. However, Pang Haitian has a different view: "Although general - purpose artificial intelligence is a hot topic, we have found that the requirements for robots are extremely different in different scenarios." He believes that it is difficult and unnecessary to achieve generality in the robot field, and instead, robots should specialize in specific areas.
This is reflected in Jiangxing Intelligence. In the short term, it will consolidate its leading position in the power and energy fields, continuously optimize and iterate the flywheel, and make the technology adapt to more segmented scenarios to lay a solid foundation for subsequent expansion. In the next 3 - 5 years, it will quickly expand to high - value tracks such as mines, chemicals, and rail transit to build an "industrial brain" in the physical world.
In the era