Jijia Vision completed two rounds of Pre-A and Pre-A+ financing totaling hundreds of millions of yuan, accelerating the arrival of the "ChatGPT moment in the physical world" with world models.
Recently, Physical AI company GigaVision announced the completion of two consecutive rounds of financing totaling hundreds of millions of yuan in the Pre - A and Pre - A+ rounds. The Pre - A round was led by Guozhong Capital, with Zifeng Capital and existing shareholder PKSHA Algorithm Fund participating in the follow - on investment. The Pre - A+ round was invested in by CICC Capital, Guangzhou Industrial Investment, Yicun Songling, and Huaqiang Capital.
In February 2025, GigaVision also completed a tens of millions of yuan angel++ round of financing, invested in by Puchao Capital, Hedinggong Capital, and Shanghai Angel Association.
GigaVision's completion of three consecutive rounds of financing within half a year fully demonstrates the capital market's recognition of its team's strength, technological roadmap, and product business progress.
GigaVision is a general intelligence company in the physical world, founded in June 2023. It focuses on the "world - model - driven basic model of the physical world" and is accelerating towards general intelligence in the physical world.
The company's products include the world model platform GigaWorld (driving + embodied), the embodied basic model GigaBrain, and general embodied bodies, etc., a full - stack series of Physical AI products. Previously, the company had completed several rounds of financing, including seed, angel, and angel+ rounds, totaling hundreds of millions of yuan.
In early 2023, the ChatGPT moment based on language models completely detonated general intelligence in the digital world, bringing about significant technological progress, an explosion of AI applications, and social impact. Currently, the "ChatGPT moment" in the physical world has become the next highly anticipated moment.
GigaVision believes that the paradigm of "world model + VLA + reinforcement learning" is gradually being perfected, and the ChatGPT moment in the physical world will arrive in 2 - 3 years, bringing about more profound technological breakthroughs in physical AI, an explosion of applications, and social impact.
Overseas, leading technology companies such as NVIDIA, Google DeepMind, and World Labs are all actively deploying and making rapid breakthroughs in the direction of world models.
Meanwhile, "VLA + reinforcement learning" based on real machines is rapidly iterating and expanding, but it faces a huge data bottleneck. The "world model" is the best way to solve the real - machine data bottleneck of general intelligence in the physical world.
"The ChatGPT moment in the physical world we define is achieving a 95% success rate in 90% of scenarios among 100 common tasks," said Dr. Huang Guan, the founder and CEO of GigaVision.
"The reason we think this moment will arrive in 2 to 3 years is a speculation based on the current speed of technological progress. In our paradigm, VLA solves the complexity of tasks, the world model solves generalization, and reinforcement learning solves accuracy and reliability. When the three work together, it will accelerate the explosion of physical AI applications," said Dr. Huang Guan.
In Huang Guan's view, after the arrival of the ChatGPT moment in the physical world, there will be an explosion of applications 10 to 100 times greater than that of language models, especially in the consumer market such as households and services. GigaVision's vision is to serve this huge - scale market.
As the first domestic technology company focusing on the "world model" direction, GigaVision is at the global leading level in both the "world model" and "VLA large model" directions and continues to lead the development of relevant technologies and industrial implementation.
In the direction of world models, GigaVision continues to promote technological innovation and industrial implementation and has reached signing and mass - production cooperation agreements with several leading enterprises such as robot and automobile OEMs.
In the direction of end - to - end VLA embodied basic models, GigaVision has industry - leading real - machine effects, the optimal model architecture design, and the highest data utilization efficiency, and has begun to move towards mass - production implementation.
Top - notch Team: A World - Class Full - Stack Algorithm Model Team Joins Hands with an Industrial Expert Executive Team
GigaVision closely relies on the Intelligent Vision Laboratory of the Department of Automation at Tsinghua University. The core team consists of top researchers from well - known institutions such as Tsinghua University and the Chinese Academy of Sciences, as well as executives and industrial experts from well - known companies such as Baidu, Microsoft, and Horizon Robotics.
The core team has published more than 200 papers in top - tier conferences and journals in artificial intelligence, led the team to win dozens of world championships in the world's most influential artificial intelligence competitions, and released several globally influential artificial intelligence technology achievements.
The team has world - class research capabilities, ultra - large - scale architecture engineering experience, and ultra - large - scale mass - production implementation experience in the Physical AI direction.
Dr. Huang Guan, the CEO is a doctor from the Department of Automation at Tsinghua University, a national - level innovation leading talent. He has led the team to win world championships in globally authoritative AI competitions multiple times and released several globally well - known AI achievements.
Dr. Huang Guan also has research work experience in well - known companies such as Microsoft, Samsung, and Horizon Robotics. He has led research and development teams of hundreds of people multiple times and has rich experience in continuous entrepreneurship in the Physical AI direction. As a core executive, he has cumulatively led or participated in financing of over one billion yuan.
Dr. Zhu Zheng, the Chief Scientist has more than 70 top - tier conference papers in the fields of vision and AI, with more than 16,000 citations. He is one of the young scientists with the highest citations in the domestic AI field.
Mao Jiming, the Vice President of Engineering has more than 16 years of experience in simulation/engineering/data/distributed architecture. He has served as the person - in - charge of simulation and engineering at Baidu Apollo and as an architect at T10 - level companies such as Baidu and Yingche.
Dr. Sun Shaoyan, the Vice President of Products has served as the director of Alibaba Cloud and the general manager of the data closed - loop product line at Horizon Robotics. He has industry - leading experience in ultra - large - scale data closed - loop products and architectures in the physical world.
The company's core team includes a world - class full - stack model team in algorithms, data, and infrastructure, a hardware team with rich experience in mass - producing humanoid robots, and an executive team of industrial experts with in - depth industry experience.
Technological Leadership: Driven by World Models, Leading a New Global Paradigm for Embodied Intelligence
The world model is Simulation 2.0 and the next generation of VLA: With the rapid development of generative AI, Simulation 2.0 based on generative world models is becoming the most important data source for general intelligence in the physical world.
By simulating and generating infinite parallel worlds with realistic vision, accurate geometry, and precise physics, and generating infinite interactive actions based on the interactive environment, Simulation 2.0 provides a new technological path and support for the training and testing of end - to - end VLA large models.
Meanwhile, as the generation ability of world models continues to increase, world models will gradually enter the robot end - side and become the next generation of VLA.
The globally leading VLA embodied basic model: In July 2025, GigaVision released the world's first world - model - driven embodied basic model, GigaBrain - 0 Preview, which solves the data bottleneck of embodied intelligence with a new paradigm.
GigaBrain - 0 is a dual - system end - to - end VLA embodied basic model with spatial reasoning ability, which can smoothly achieve long - range complex operation tasks such as flexible operation, two - arm collaboration, open interaction, and autonomous reasoning.
More importantly, GigaBrain - 0 has made breakthrough progress in data sources. More than 90% of the data comes from the data generated by GigaVision's self - developed "world model platform", and less than 10% comes from real - machine data collection. It achieves great generalization with extremely low cost.
Based on the massive data generated by the world model, GigaBrain - 0 has achieved "zero - shot" generalization of flexible long - range complex operations in the open world for the first time globally. The excellent performance of GigaBrain - 0 provides a new paradigm for solving the data bottleneck of embodied intelligence.
In September 2025, GigaVision will release the official version of GigaBrain - 0.
The globally leading driving and embodied world model platform: GigaVision's world model platform is applied in two directions: driving and embodiment.
In the direction of driving world models, GigaVision continues to lead technological breakthroughs and mass - production implementation in relevant fields.
It has published a series of representative achievements such as DriveDreamer, DriveDreamer - 2, DriveDreamer4D, ReconDreamer, and ReconDreamer++, all of which have been included in top - tier conferences in the fields of vision and AI.
In the direction of embodied world models, GigaVision has published representative leading academic works such as EmbodieDreamer, RoboTransfer, and EmbodiedGen.
GigaVision has industry - leading technological accumulation and model effects in both the video world model and 3D world model directions in the field of embodied intelligence and has begun large - scale application in actual implementation scenarios.
GigaVision will soon release the embodied world model platform GigaWorld - 0. Stay tuned.
Accelerating Mass - Production: Joining Hands with the Industry to Accelerate Mass - Production Implementation and Move Towards General Intelligence in the Physical World
Since its establishment, GigaVision has always prioritized commercial mass - production implementation.
It has not only been the first to achieve large - scale mass - production implementation of world models in the field of autonomous driving but has also achieved extensive cooperation and implementation in the directions of embodied world models and embodied large models. Driven by technology, it joins hands with the industry to accelerate mass - production implementation and move towards general intelligence in the physical world.
In the direction of driving world models, GigaVision has reached signing and mass - production cooperation agreements with several domestic leading automobile OEMs, overseas and joint - venture leading automobile OEMs, AI chip manufacturers, and Tier 1 giants.
In the direction of embodied world models, GigaVision has reached in - depth cooperation with several humanoid robot innovation centers and training fields to create full - stack products such as the world's leading virtual - real combined training field.
In the direction of embodied VLA large models, GigaVision has reached in - depth cooperation with several global automobile industry giants and leading embodied body companies. Multiple parties jointly explore the implementation of embodied intelligence in multiple scenarios such as industry, services, and households.
GigaVision will mainly use the financing funds for technological product R & D and market business expansion. In addition, it will also invest more in customer delivery.
In addition to its algorithm advantages, GigaVision also has advantages in data, infrastructure, and industrial implementation. It serves customer needs in a multi - dimensional way and promotes the industry to accelerate towards the "ChatGPT moment in the physical world".
Views from Investors: World Models are the Key to General Intelligence in the Physical World, and GigaVision Has Comprehensive Technological and Industrial Advantages
Shi Xin, the Executive General Manager of Guozhong Capital said: GigaVision is a leader in the domestic world model and embodied intelligence directions. It not only started the earliest technological layout and innovation leadership in relevant directions but also quickly achieved large - scale industrial implementation.
We highly value the company's team's technological advantages and industrial experience. We look forward to the ChatGPT moment in the physical world and hope that GigaVision can lead the ChatGPT moment in China's physical world.
Yang Xiaoning, the Chairman of Zifeng Capital said: Currently, the embodied intelligence and robot industry is at a critical moment of vigorous development. Embodied large models are the key to the industry's development. The GigaVision team has more than 10 years of experience in physical AI, has gone through multiple physical AI eras such as CV and autonomous driving, and has continuously produced globally leading technological achievements and achieved large - scale industrial implementation.
We are optimistic that the GigaVision team will continue to achieve greater success in the era of embodied intelligent robots.
The project leader of CICC Capital said: Data and models are the keys to the development of embodied intelligent robots. The GigaVision team has leading advantages in both data and models of embodied intelligence. We are very optimistic about using world models to more fundamentally solve the problems of data and models in embodied intelligence.
Meanwhile, the GigaVision team has both leading technological advantages and in - depth industrial accumulation. CICC Capital will continue to be optimistic about and support the team, helping the team to continuously innovate in technology and implement applications in the field of embodied intelligence and promoting the embodied intelligence industry to reach new heights.
The project leader of Guangzhou Industrial Investment said: Guangzhou Industrial Investment has long been optimistic about the broad market space of embodied intelligent robots in the future. The current development bottleneck of embodied intelligent robots is the embodied large model. As the capabilities of embodied large models continue to improve, embodied intelligent robots are expected to become the next super - terminal.
With the support of the world model, GigaVision's end - to - end VLA large model can not only smoothly achieve various long - range complex operation tasks such as handling flexible objects, two - arm collaboration, open interaction, and autonomous reasoning but also achieve "zero - shot" generalization of flexible complex long - range operations in the open world for the first time. Guangzhou Industrial Investment will rely on the technological and industrial scenario resources in Guangzhou to continuously support the development of GigaVision.
Zhang Chen, the Investment General Manager of Yicun Songling said: The world model is the next most fundamental thing after language models and is one of the ecological points that Yicun Songling plans to layout in its decision - making cycle. Physical AI will improve the mapping relationship between natural rules in the physical and digital worlds. Physical engines and world models may be the necessary technological paths.
The GigaVision team's joint optimization paradigm of generation + reconstruction based on Gaussian splashing is very impressive. The team is also one of the earliest domestic teams to systematically layout world models and has quickly achieved industrial implementation in multiple application directions.
Yicun Songling has long been optimistic about the directions of artificial intelligence and robots and will continue to strongly support GigaVision and other projects in the Songling AI ecosystem, promoting the artificial intelligence industry to reach new heights and creating greater industrial and social value.
Xie Liguang, the Senior Investment Director of Huaqiang Capital said: Huaqiang Capital is deeply involved in the revolution of the embodied intelligence and robot industry. GigaVision is at the leading level in China in both the world model and VLA large model directions.
We are optimistic about the team's comprehensive advantages in technological depth, breadth, and industrial experience in the Physical AI direction. We hope that GigaVision will become a representative domestic enterprise in embodied large models and expect the team to make greater breakthroughs in robot terminals and applications.
Li Liwei, the Founding Partner of Puchao Capital said: GigaVision is a leader in domestic spatial intelligence and world models. It not only started the earliest technological layout and achieved innovative breakthroughs but also achieved large - scale application implementation in multiple fields such as driving, embodiment, and content.
Puchao Capital is optimistic about GigaVision's team's leading technological accumulation and in - depth industrial implementation experience in the field of artificial intelligence and will continue to support GigaVision to become a new - generation representative enterprise in the field of artificial intelligence and world models.
Dr. Huang Guan