According to a research report, the 2026 World Artificial Intelligence Conference (WAIC) held in Shanghai signifies a key shift in the evaluation criteria for the embodied AI industry from "ability to perform actions" to "ability to consistently generate production value." Industrial and warehousing scenarios are likely to become crucial testing grounds.

Moving forward, the industry's competitive focus is expected to gradually shift towards metrics such as task success rates, human intervention rates, failure recovery capabilities, and customer payback periods, rather than simply comparing degrees of freedom, computing power, or motion performance. From a supply chain perspective, investment opportunities are anticipated to expand from the concept of robot bodies to operational capabilities and production line integration, though component suppliers' performance will still depend on securing mass production contracts.

Key Observations from the Event

From July 17-20, on-site research was conducted at the WAIC in Shanghai, where over 1,100 companies showcased more than 3,000 products. For the advanced manufacturing sector, notable developments at this conference included embodied AI companies increasingly focusing their demonstrations on practical tasks such as warehouse handling, loading/unloading, quality inspection, precision operations, and public services.

For instance, Agibot unveiled new products including the Expedition A3Ultra, Elf G2Max, Lingxi X2EDU, and dexterous hands. The G2Max, developed in collaboration with JD Logistics, is planned for deployment in JD's Smart Wolf Warehouses for handling and palletizing tasks. NIO's subsidiary, Shenji, launched the "Ruidong" embodied intelligent development platform based on the NX9031U chip, extending the company's automotive expertise in perception, control, and software toolchains to the robotics field.

Shift in Industry Evaluation Standards

A primary significance of this WAIC is the shift in the embodied AI industry's evaluation standards from "whether actions can be completed" to "whether sustainable production value can be created." Previous exhibitions focused more on walking, running, jumping, and single grasping actions. This year, companies began actively demonstrating robots' continuous operation capabilities in warehousing, semiconductor loading/unloading, assembly, quality inspection, and public services.

Agibot disclosed that the G2Max dual-arm robot has a peak load capacity of 50 kilograms and can support 24/7 palletizing and handling operations. Multiple robots were also deployed during the conference for tour guidance and public services. It is anticipated that future industry competition will increasingly center on task success rates, human take-over rates, fault recovery capabilities, and customer investment payback periods, rather than simply comparing degrees of freedom, computing power, or motion performance.

Industrial and Warehousing as Key Validation Arenas

Industrial and warehousing scenarios are poised to become important validation windows for the commercialization of embodied AI in the second half of 2026. Compared to full-size bipedal humanoid robots, wheeled dual-arm and specialized robots may be deployed in customer scenarios and generate revenue more quickly.

Tasks like warehouse handling, loading/unloading, and quality inspection have relatively clear boundaries and more controllable working environments, making it easier to calculate economic returns based on labor savings, increased equipment utilization, and extended operation times. The entry of G2Max into JD Logistics' Smart Wolf Warehouses indicates that downstream customers are moving from trial uses at exhibitions to deployment in real-world scenarios.

In the near term, priority should be given to robotics companies that already possess customer scenarios and system integration capabilities, rather than assuming a single general-purpose humanoid platform can quickly cover all manufacturing segments.

Expanding Investment Opportunities in the Supply Chain

From an industry chain perspective, investment opportunities are expected to gradually expand from the concept of robot bodies to operational capabilities and production line integration. However, the financial performance of component suppliers will still need to be grounded in securing mass production contracts.

This conference saw a notable increase in demonstrations of dexterous hands, vision, tactile, force control, and multi-robot collaboration solutions. Companies like Mech-Mind showcased "one-brain-multi-form" and coordinated eye-brain-hand operations, while Agibot also introduced a new generation of dexterous hands. As robots evolve from handling to tasks like insertion, assembly, and fine manipulation, the importance of components such as joint modules, reducers, servo motors, ball screws, machine vision, and force/torque sensors increases correspondingly.

Relevant companies include Sanhua Intelligent Control, Tuopu Group, Leader Harmonious Drive Systems, Shuanghuan Driveline Co., Ltd., MOONS', Kinco Automation, and Inovance Technology. Currently, most collaborations remain at stages like sample delivery, joint development, or small-batch production. Supplier market share and value per unit may still change as robot body designs are adjusted.

Valuation Catalysts and Future Drivers

In terms of valuation, the WAIC served as a thematic catalyst for the robotics and automotive physical AI sectors. Shenji's launch of the "Ruidong" embodied intelligence platform based on the 5nm automotive-grade NX9031U chip reflects the potential spillover of the automotive industry's accumulated expertise in multi-sensor perception, real-time control, functional safety, and mass-production toolchains into robotics. Its high-end NX9031X chip is already deployed in NIO and Onvo vehicle models.

For robotics body and component companies, the shift from thematic to fundamentals-driven valuation will depend on factors in the second half of 2026, including paid orders, repeat purchases, customer acceptance, unit pricing, and project gross margins. The view is reiterated to prioritize companies with mature manufacturing capabilities, clear customer projects, and defined revenue conversion pathways.

Risks to Consider

Potential risks include slower-than-expected customer adoption and acceptance of embodied AI; changes in supply chain share due to product design adjustments; robot stability and economic returns failing to meet customer requirements; and discrepancies between operational data demonstrated at exhibitions and performance in real production environments.