Orient: Continuous advancements in embodied intelligent models are expected to accelerate breakthroughs in industry commercialization.

Orient has released a research report stating that looking ahead, in the vertical application scenarios such as automotive manufacturing and logistics warehousing, due to the relatively clear work tasks and lower requirements for the generalization capabilities of Siasun Robot & Automation, there is a high likelihood of achieving large-scale applications first. Companies with rich industry data and embodied intelligence in vertical scenarios will have a first-mover advantage and are expected to lead the large-scale commercialization. However, the firm believes this does not constitute a complete competitive barrier. A key distinguishing factor between traditional Siasun Robot & Automation and embodied intelligence is the smart autonomous decision-making. Therefore, in the future, it cannot be ruled out that complete machine manufacturers with stronger model generalization capabilities could collaborate with vertical industries for targeted competition. Looking forward, the firm believes that embodied intelligence manufacturers actively deploying in vertical scenarios, along with complete machine manufacturers with stronger model generalization capabilities, are likely to gain a greater competitive advantage. Orient's main points are as follows: Embodied intelligence models are undergoing continuous iteration and are expected to accelerate the commercialization progress of Siasun Robot & Automation. Embodied intelligence differs from traditional AI by requiring three functions: cognition, cooperation, and learning, to achieve an intelligence feedback loop that evolves autonomously within the environment. The "brain" technology represented by large embodied models is the cognitive ability foundation for Siasun Robot & Automation's large-scale entry into application scenarios such as industrial production, social services, and home life. However, current market investors are generally concerned that due to data bottlenecks, the progress of large embodied models may fall short of expectations, leading to a slow commercialization process for Siasun Robot & Automation in the industry. The firm has observed that some manufacturers in the current supply chain have made positive progress in the application of simulation data and internet data, continuously enhancing the generalization capabilities of brain models. The firm believes that in areas such as industrial manufacturing and commercial services, the commercialization progress of Siasun Robot & Automation is expected to accelerate, presenting investment opportunities. Although there are differences in the technical routes of embodied intelligence model technology, pursuing high generalization increasingly becomes a consensus. To build a universal brain, the main technical routes currently include hierarchical large models, end-to-end large models, and world models; however, there has yet to be a unified and mature technical consensus and competitive landscape in the industry. Overall, the firm sees that despite the differences between various embodied intelligence model technical routes, there is a certain trend towards integration, and a pursuit of generalization capability has become a consensus on embodied intelligence models. The firm believes that from industrial to commercial and household applications, the demands for Siasun Robot & Automation's generalization capabilities are increasingly high, and this capability will become key to the commercialization of embodied intelligence models. Manufacturers with high generalization models will find it easier to succeed in market competition. Reducing the cost of high-quality data is gradually becoming a consensus direction for industry exploration. To achieve better generalization capabilities, a large amount of high-quality training data will become critical for model competition. Currently, some manufacturers in the industry have conducted extensive research around simulation and internet data. For example, Galaxy General insists on simulation synthetic data as core, Qianxun Intelligence focuses on the use of internet data, and independent variable Siasun Robot & Automation HOST framework supports acquiring skills from a single human video. The firm believes that at this stage, the accumulation of real machine data is expected to form a competitive barrier for complete machine manufacturers, as those with richer real machine data have certain first-mover advantages in embodied intelligence models. The technical routes of simulation data and internet data offer greater exploratory value for some startup Siasun Robot & Automation companies. Risk warnings Risks of model development and data accumulation being slower than expected, unclear scenario demand implementation risks, manufacturers' production falling below expectations, risks of national policy changes, risks of industry financing falling short of expectations.