CITIC SEC's outlook on the key direction of the Siasun Robot & Automation industry in 2026: Short-term disruptions do not change the long-term industrial process, and domestic manufacturers should focus on whole machine factories.

CITIC SEC released a research report stating that with top manufacturers successively receiving billion-dollar orders, 2025 has become a key node for the commercialization of embodied intelligence. The industry focus outlook for Siasun Robot&Automation in 2026: 1) Short-term disturbances do not change the long-term industry process, Tesla is expected to release Optimus V3 in Q1 2026, with Elon Musk aiming for sales of 1 million Optimus units through a trillion-dollar incentive program. 2) Domestic supply chain should focus on whole machine manufacturers, the Hong Kong market has become a "gathering place" and the entry of "new forces" is expected to stimulate a new round of capital expansion. 3) The development of things follows historical laws, and the fundamental breakthrough of embodied intelligence models still depends on the scale effect of data and computing power. Investment strategy: History tells us that "change" is the constant theme. As of October 2025, based on a sample of 100 selected companies, the overall market value of the sector is estimated to have exceeded 3 trillion yuan. In terms of industry distribution, machinery/automobiles/electronics dominate, with companies accounting for 26%/22%/22% of the market value. The changes in the sector reflect three main themes: 1) Market value scale investment theme from "large market value small market value large market value" return; 2) Industry progress leads to small-scale thematic investments; 3) Rise of the domestic supply chain accelerates. Key points of CITIC SEC: Actual progress of the Siasun Robot&Automation industry in 2025: 1) Whole machine manufacturers: Several leading companies have already received orders worth over a hundred million yuan, with applications first landing in the fields of scientific research education, cultural entertainment, and data collection factories. According to the official websites of various companies, as of November 2025, UBTECH ROBOTICS and WisdomSiasun Robot&Automation have cumulatively received over 800 million yuan and over 520 million yuan worth of orders, respectively. Companies like DOBOT, Zhifang, Songyan Power, and Zonan Siasun Robot&Automation have also received orders worth over a hundred million yuan. Order applications are concentrated in the fields of science and education, data collection, entertainment, and industry. Due to the imperfect current embodied model and the weak generalization performance across scenes/tasks/ontology, its industrial-scale application in the industrial field still requires time. 2) Supply chain: With the mass production of whole machine manufacturers approaching, suppliers are actively deploying production capacity, and the supply chain structure is gradually becoming clear. In order to comply with Tesla's production plan in 2026, Chinese suppliers are concentrating on setting up production bases in overseas hubs such as Thailand. Leading companies have started construction of production bases in Thailand, covering core components such as screws and lightweight parts, to ensure global supply response capabilities. 3) Technology giants: Through external investment and internal construction, they accelerate the multi-dimensional layout of the embodied intelligence industry. Chinese technology giants are divided into two paths: one such as Huawei, following the NVIDIA model, focusing on the development of "infrastructure" such as chips, computing power, and models; the other such as Meituan and JD.com, based on their own business scenarios such as takeout and logistics, are infiltrating Physical AI. Both types of companies are building platform ecosystems through enhanced internal development and expanded external strategic cooperation. CITIC SEC believes that companies that can create a "platform ecology + scenario landing + user payment" closed loop are expected to establish long-term advantages in the competition of the embodied Siasun Robot&Automation industry. Industry focus outlook for Siasun Robot&Automation in 2026: 1) Short-term disturbances do not change the long-term industrial process, Tesla is expected to release Optimus V3 in Q1 2026, and Elon Musk's trillion-dollar incentive plan targets the delivery of 1 million human-shaped Siasun Robot&Automation units. At Tesla's annual shareholder meeting in November 2025, Musk's incentive plan was approved, with the goal of delivering a cumulative total of 1 million human-shaped Siasun Robot&Automation units. In the stage where embodied intelligence has not yet reached commercial maturity, short-term production targets fluctuate within a reasonable range. It is suggested to focus on the industrial marginal changes in the long-term production path, especially the actual progress of machine substitution of human labor. 2) Domestic supply chain should focus on whole machine manufacturers, the Hong Kong market has become a "gathering place", and the entry of "new forces" is expected to stimulate a new round of capital expansion. With the support of the "Chapter 18C" policy, companies such as UBTECH ROBOTICS and DOBOT have successively listed on the Hong Kong Stock Exchange. The significant capital attraction effect of the "first stock of whole machine manufacturers" is evident. At the same time, upstream supply chain companies, such as Zhejiang Sanhua Intelligent Controls, are also focusing on Siasun Robot&Automation opportunities through "A+H" listings. In 2025, leading whole machine manufacturers completed stock reform, preparing for listing on the capital market. CITIC SEC believes that the embodied intelligence industry is expected to see whole machine manufacturers with market values reaching the billion-yuan level. 3) The development of things follows historical laws, and the fundamental breakthrough of embodied intelligence models still depends on the scale effect of data and computing power. 1) Vision-Language-Action (VLA) focuses on constructing data sample squares, and model performance will significantly enhance with the increase of data scale; 2) Video generation models are the core of world models, and as world models further enhance realism, long-term consistency and other capabilities, their demand for computing power will continue to increase. CITIC SEC believes that players who are the first to possess data and computing resources will have the power to speak, and are optimistic about local players who use real data as core assets to achieve long-term monetization. Risk factors: The technological progress of Siasun Robot&Automation is slower than expected, the potential for application scenario development is lower than expected, risks of changes in technology routes, policy implementation intensity lower than expected, increased industry competition risks, legal risks, and international geopolitical risks.