Exclusive Interview with Yan Qianhang, Vice President of FreeS Fund, and Guan Chaoyu, Founder of Qingmao Intelligence: Envisioning the Applications and Prospects of Embodied Intelligence
Amid the wave of artificial intelligence, embodied intelligence and humanoid robot technology are emerging as new focal points. This interview invites Mr. Yan Qianhang from FreeS Fund and Mr. Guan Chaoyu from Qingmao Intelligence to conduct an in-depth discussion on the connotation, extension of embodied intelligence, and the future of humanoid robots. Mr. Yan Qianhang has rich investment experience in cutting-edge technology and interdisciplinary fields, with a particular focus on and layout in the embodied intelligence and humanoid robot track. Mr. Guan Chaoyu will discuss the challenges and opportunities of artificial intelligence infrastructure in the era of large models. Through their dialogue, we will gain insight into how embodied intelligence endows robots with a "soul" and the application prospects of artificial intelligence in various fields in the future. This is not only a dialogue at the forefront of technology but also a profound insight into investment trends and industrial cooperation.
Yan Qianhang is Vice President of FreeS Fund. He focuses on early-stage investment opportunities in cutting-edge technology and pan-technology interdisciplinary fields. The projects he mainly invests in and participates in managing include Zhuji Dynamics, Hangkai Microelectronics, Interstellar Light-year, etc.
Guan Chaoyu graduated from Tsinghua University with a master’s degree in Computer Science and Technology. He is the Founder and CEO of Qingmao Intelligence Technology (Beijing) Co., Ltd. He has long been engaged in artificial intelligence research and achieved remarkable results in this field. During his postgraduate studies, he won the Tsinghua University Special Scholarship and the Siebel Scholarship, and was the youngest recipient of the Tsinghua University Postgraduate Special Award. He has published 7 papers as the first author, led the open-source development of AutoGL, the first automatic graph learning platform, during his postgraduate studies, and led his team to win two championships in top international deep learning competitions. He is the chief architect of AutoGL, the world’s first open-source automatic graph learning system (with over 800 stars on Github), and holds 1 invention patent. Currently, he has founded Qingmao Intelligence Technology (Beijing) Co., Ltd., focusing on AI Infra, and has secured over 100 million yuan in financing.
Hello everyone, I’m Yan Qianhang from FreeS Fund, mainly responsible for early-stage investments in cutting-edge technology and interdisciplinary fields. Founded in 2015, FreeS Fund is an early-stage venture capital institution that mainly focuses on early-stage investment layout in three directions: technology, consumption, and healthcare. In recent years, we have paid special attention to the fields of embodied intelligence and humanoid robots, and have invested in a number of enterprises. We have witnessed the transformation of this track from a niche area to a popular one, and will continue to keep a close eye on it. Our goal is to explore potential new tracks and support industry entrepreneurs in realizing their dreams.
Hello everyone, I’m Guan Chaoyu, Founder of Qingmao Intelligence. Founded at the end of 2022, Qingmao Intelligence focuses on artificial intelligence infrastructure in the era of large models, dedicated to solving the problem of deploying large models on various chips—including overseas chips such as NVIDIA and AMD, as well as domestic chips like those from Huawei, especially edge chips (e.g., chips for embodied intelligence, PCs, or in-vehicle embedded systems). We mainly address issues such as energy consumption and latency, conduct model optimization and deployment, and help enterprises build model capabilities on the edge side or data center side.
Today, our discussion focuses on embodied intelligence and humanoid robots. First, we would like to ask the two experts to explain the meaning, extension, and connotation of embodied intelligence.
Embodied AI, in essence, refers to AI with a physical entity. It is not limited to robots; any device equipped with a physical entity can be called embodied intelligence. Currently, embodied intelligence is closely associated with robots, especially humanoid robots. Humanoid robots need a "soul" or "brain," and the emergence of embodied intelligence just provides this element—integrating the upsurge of humanoid robots with the boom of AI, and spawning many new entrepreneurs.
The rapid development of AI has endowed robots with a "brain." We would like to ask how to gradually deploy large models to the edge side, such as robot brains, mobile phones, or PCs.
At present, many large companies and leading players are researching how to deploy AI to edge devices. The advantages of edge devices include timeliness, stability, and privacy protection. Deploying models to hardware usually involves three steps: model adjustment and optimization, framework optimization, and hardware optimization. Our goal is to enable models to be deployed on edge devices for fast computing, while meeting requirements for energy consumption and battery life.
Embodied intelligence and humanoid robots are the results of interdisciplinary development. Mr. Yan, what are the key technologies involved and the investment areas worthy of attention?
The selling points of robots have evolved from industrial robots, collaborative robots, etc. AI intelligence is divided into three levels: perception, decision-making, and execution. We expect robots to possess these capabilities. Core technologies include autonomous decision-making and execution capabilities on the intelligent side, as well as movement capabilities. We focus on investment opportunities, especially the technological improvements in intelligent and movement capabilities.
The biggest bottleneck of embodied intelligence lies in its generalization, which is also its greatest advantage. Mr. Guan, what is the current pattern of domestic GPU computing power? How can we make up for the gap through software means in the future?
The computing power sources for embodied intelligence are divided into two parts: cloud side and edge side. On the cloud side, NVIDIA occupies a dominant position; despite the U.S. ban, domestic demand for computing power remains strong. Domestic GPUs and NPUs are growing, but there are still gaps in hardware performance and software ecology. We hope to contribute to the development of the domestic ecology through our own efforts—especially on the edge side, where we see more opportunities.
China and the United States are frontier countries in the development of artificial intelligence, but the final competition may lie in applications. Mr. Yan, what are the potential killer application areas of artificial intelligence in the future?
From the development of ChatGPT, some proven applications have emerged, such as AI programming and AI search. In the future, the combination of AI assistants and AI hardware may become a new trend. AI models should be able to process multi-modal information and provide multi-modal output results. We look forward to the integration of AI with scenario-based knowledge, which will be a challenge for AI applications. For the B-end market, the application of AI will rely on data governance and model capabilities to form enterprise intelligence.
Nanjing is an important innovative city in the Yangtze River Delta and is vigorously laying out investment and investment promotion in the artificial intelligence field. Mr. Guan and Mr. Yan, what are your prospects for future cooperation with Nanjing Innovation Investment Group and Nanjing?
As a startup, we focus on financing, talent recruitment, and project development. We hope to conduct in-depth cooperation with Nanjing and Nanjing Innovation Investment Group in policy-based recruitment, talent base construction, and computing power equipment, to promote the operation of domestic AI large models on domestic chips.
FreeS Fund has a long history of cooperation with Nanjing, and Nanjing has provided us with strong support. Nanjing is a city with a high quality of life and abundant talent and scientific research resources. We hope to help more potential technologies and projects take root in Nanjing and go global.
Source: FreeS Fund, Qingmao Intelligence, Zhang Surong (Fourth Investment Department of Nanjing Innovation Investment Group), Zhou Xin (Fund Department of Nanjing Innovation Investment Group)
Reviewer: Xue Yao
Publisher: You Yi
