Robots enter production lines and AI applications accelerate as trading contracts beneath the technology theme’s popularity
Heavy-duty robots are beginning to cover multiple production-line processes, while competition in AI applications is extending to products and commercialization. But market turnover contracted on 26 October, and the numbers of high-turnover robotics and AI companies fell at the same time. Industrial progress and trading activity need to be assessed separately.
Industrial advances in robotics and AI applications are moving toward production use and commercialization, but market turnover is contracting and views on the leading themes are diverging. Active themes have not been accompanied by a corresponding broadening of capital participation.
What has changed with this heavy-duty robot?
The Genie G2 Max released by AgiBot combines heavy-duty capacity and precision operations in one robot: a payload of 20 kilograms per arm, a peak dual-arm payload of 50 kilograms, and arm repeat positioning accuracy of no more than ±0.5 millimeters. Its intended tasks already cover depalletizing and palletizing, opening boxes and feeding materials, collecting empty boxes, and putting goods into warehouse storage, extending its use to continuous production-line processes.
The G2 Max has begun validation and deployment on the production lines of leading daily-use consumer goods, e-commerce and food companies. Citing advances in model planning, spatial perception and Real-to-Sim modeling, as well as faster mass-production ramp-ups for domestically produced complete robots and Optimus, CITIC Securities believes humanoid robots are entering a new phase. Tesla’s release of Optimus V3 in the fourth quarter remains a future milestone expected by the firm.
How is the supply chain meeting robot development and training needs?
The material identifies links in R&D delivery and core components: JLC provides AgiBot with EDA design, PCB and PCBA delivery, and mechanical component manufacturing support; Leaderdrive supplies high-torque-density harmonic reducers and integrated joint modules, and has already delivered them in volume to leading international robotics customers. The two operate at different points in the value chain, so progress on complete robots cannot be used as a direct substitute for their respective delivery progress.
Training data collection products have also been deployed. Orbbec released a hardware platform for data collection without a robot body in July 2026, covering first-person, handheld-operation and wrist-mounted near-field perspectives. Skyworth Digital used its XR expertise in spatial positioning, multi-camera coordination and spatiotemporal synchronization to launch data collection products, which have generated sales in China and overseas.
Why is competition in AI applications shifting toward commercialization?
Guosen Securities believes that improving model capabilities and falling inference costs are extending competition from model iteration to application scenarios, product experience and commercialization capabilities. Specific paths include agents reshaping office work, marketing and enterprise workflows, and video generation technology improving content production efficiency while expanding content formats such as AI film and television, motion comics and interactive entertainment.
Bank of America’s assessment of Microsoft provides an enterprise example: Azure’s expansion and accelerating Copilot commercialization further consolidate its advantages in enterprise AI, security and infrastructure. The bank expects Azure demand to continue exceeding supply in fiscal 2027 and has raised its earnings forecasts for the next three years. While demand is strong, supply still constrains the amount of business it can take on.
Domestic companies are at different stages: DZH is still exploring AI investment research products and industry-specific vertical models, while Hanwang Technology’s businesses cover pen-based intelligent interaction, multimodal big data and AI smart terminals. Existing business activities and the commercialization of new applications need to be understood in light of each company’s own progress.
Why look at trading breadth when themes are active?
On 26 October, market turnover fell by 347.5 billion yuan from the previous trading day, and the number of companies with turnover above 1 billion yuan fell by 70 to 327. Of these, 119 were robotics concept companies and 82 were AI concept companies, down by 7 and 21, respectively. This shows that the breadth of high-turnover companies in active themes was also narrowing.
Price strength remained concentrated in a small number of stocks: the number of companies hitting a one-year high rose by 1 to 13, including 5 in robotics and 4 in AI. Only 2 companies reached all-time highs, one each in robotics and physical AI. On this basis, the material describes a contracting and diverging market, with greater disagreement among capital focused on the leading themes, and expects the market to most likely remain range-bound in the short term until trading volume expands meaningfully.
