Citi International: The generalization ability of large models is the main commercialization bottleneck for Siasun Robot & Automation, given an industry rating of stronger than the market.

Citi International released a research report stating that humanoid Siasun Robot & Automation, as the ideal carrier of embodied intelligence, has crossed the threshold for mass production by 2025. The main bottleneck currently hindering its large-scale commercialization is the insufficient generalization ability of large models. However, with the rapid development of models, computing power, and data in the future, it is expected to gradually break through. The domestic industrial chain, leveraging unique policy support, strong in-depth industrial capabilities, and cost advantages, is currently in a globally leading position and is expected to fully benefit from the large-scale rollout of humanoid Siasun Robot & Automation. It is recommended to focus on the core components that can first realize orders and revenue, rating the industry as outperforming the market. Citi International's main points are as follows: Humanoid Siasun Robot & Automation is the ideal carrier for embodied intelligence, crossing the mass production threshold in 2025. With the gradual maturity of the hardware technology of humanoid Siasun Robot & Automation and the rapid development of AI large models, driven by multiple factors such as supportive policies and increased investments in various countries, the humanoid Siasun Robot & Automation industry has reached a historic turning point. Multiple domestic and foreign manufacturers are expected to achieve scaled mass production by 2025. According to statistics from Omdia and Citi's research, the global shipment of humanoid Siasun Robot & Automation is expected to reach approximately 14,800 units in 2025, representing significant growth year-on-year from 2024. Meanwhile, the order scale of domestic Siasun Robot & Automation is expected to achieve a substantial breakthrough. From the planning of various domestic and international manufacturers, the next two years will be key for humanoid Siasun Robot & Automation to advance towards industrialization. The rapid improvement of hardware mobility is evident, but the generalization capability of large models is the main commercialization bottleneck at present. In the past two years, domestic and international humanoid Siasun Robot & Automation manufacturers have made significant progress in the mobility of their platforms. Based on the current hardware level demonstrated by humanoid Siasun Robot & Automation, Citi believes it can meet basic deployment needs. The bottleneck hindering large-scale commercialization of humanoid Siasun Robot & Automation lies mainly in the low level of intelligence centered around the "brain." On one hand, currently relatively mature VLA large models face certain bottlenecks in generalization capacity, while world models with stronger generalization capabilities are still in the early stages. On the other hand, the high-density, high-quality data required for the deployment of embodied intelligence models is severely lacking, resulting in the current downstream application scenarios of humanoid Siasun Robot & Automation still being confined to structured scenes such as industrial and vertical applications, making it a long way to go for non-structured scene applications like home service. Rapid development in algorithms, computing power, and data is expected to help break through the bottleneck of model generalization capability. In terms of models, world models are developing rapidly; in terms of computing power, with the optimization of algorithms for large models, improvements in inference engineering, and the expansion of computing infrastructure, the supply of computing power is rapidly increasing. In the past two years, inference costs have dropped sharply, and the training and inference costs for the "brain" of humanoid Siasun Robot & Automation are expected to decline in tandem. Regarding data, governments and manufacturers are accelerating their layouts to address data shortages. The rapid development of algorithms, computing power, and data, which are the three essential elements of physical AI, is expected to assist humanoid Siasun Robot & Automation in overcoming the bottleneck of insufficient model generalization capacity. Policy, funding, and technology resonating together gives the domestic industrial chain a significant advantage, which is expected to fully benefit. With the unique support system of policies and funding in China, strong manufacturing foundation, depth of the supply chain, and scaled cost advantages, the industrialization and cost reduction of components for Chinese humanoid Siasun Robot & Automation are currently at a globally leading position. By 2025, the top six global shipments of humanoid Siasun Robot & Automation are expected to be from Chinese companies, collectively accounting for 87.87% of the global shipment share. In the future, as humanoid Siasun Robot & Automation moves towards large-scale commercial applications, the domestic industrial chain is likely to fully benefit, particularly in the core component segment, which will be the first to realize relevant orders and revenue. Motion schemes are gradually converging, but areas such as dexterous hands, electronic skin, and thermal management are still undergoing iterations. Analyzing the various components of humanoid Siasun Robot & Automation, a typical humanoid Siasun Robot & Automation architecture includes perception systems, decision-making and control systems, execution systems, and other components, where the rotary actuators, linear actuators, and dexterous hands as execution systems account for a major portion of the cost composition. Currently, the electric drive execution system, mainly composed of motors, reducers, and screws, has become the choice for most manufacturers, demonstrating a high degree of convergence and certainty in the technological route for motion schemes. However, areas like dexterous hands, electronic skin, and thermal management remain in the exploratory stage due to current technical levels and cost factors. While there are certain uncertainties, the development potential remains considerable. Valuation The humanoid Siasun Robot & Automation industry is currently in a critical transition phase from technological validation to scaled mass production, with a primary focus on core components that are expected to realize orders and revenue first. Companies to watch include Jiangsu Hengli Hydraulic, Leader Harmonious Drive Systems, Zhejiang Rongtai Electric Material, Shanghai Beite Technology Group, Ningbo Zhenyu Technology, Ningbo Tuopu Group, Zhejiang Sanhua Intelligent Controls, Zhejiang XCC Group, Shanghai Moons'Electric, Kinco Automation (Shanghai) Co., Ltd, Jiangsu Riying Electronics, Hanwei Electronics Group Corporation, and Keli Sensing Technology. Major risks facing the rating Risks that the commercialization process of humanoid Siasun Robot & Automation may not meet expectations; risks of technological iteration; risks of declining profitability.