Goldman Sachs stated that the humanoid robot performance at the 2026 CCTV Spring Festival Gala signifies a major advancement in China's robotic hardware engineering, but the true artificial intelligence capabilities still require further testing.
In a research report released on February 21st, Goldman Sachs noted that the Year of the Horse Spring Festival Gala served as a significant platform for showcasing Chinese humanoid robot technology. Performances by robots from private companies such as Unitree, Noetrix, Magic Lab, and Galbot demonstrated substantial progress in hardware and full-body control. However, the highly choreographed nature of the performances makes it difficult to fully assess their underlying AI capabilities.
Goldman Sachs maintains its forecast for global humanoid robot shipments of 51,000 units in 2026 and 76,000 units in 2027, representing a multi-fold increase from the 15,000-20,000 units expected in 2025. Stocks in the related supply chain may see short-term benefits, but investors should be cautious of first-quarter earnings pressure and risks from rising raw material costs. In the long term, the pace of development in general AI capabilities will determine the industry's upside potential, with the "world model" technical approach being a key area to watch.
**The Gala Stage: A Showcase of Hardware Capability**
Goldman Sachs views the Year of the Horse Spring Festival Gala as a concentrated display of China's robotic capabilities. Unitree's robots demonstrated multi-machine coordination and synchronized movements, exhibiting fluid motion capabilities and dynamic balance, along with failure recovery abilities resembling human agility in multiple scenarios. Robots from Magic Lab and Galbot displayed impressive dexterity, capable of precisely executing complex manipulations. Noetrix emphasized its bionic robot's seamless collaboration with humans and expressive movements.
Analyst Jacqueline Du pointed out that the coordination and stability demonstrated by these robots in highly difficult choreography proves rapid improvement in actuator technology, sensor integration, and advanced control algorithms. However, Gala performances are typically pre-programmed and meticulously rehearsed, designed to showcase the robots' maximum potential under controlled conditions. They often rely on predefined trajectories and reactive behaviors rather than real-time autonomous decision-making in unpredictable environments. Consequently, while the physical execution was flawless, their cognitive intelligence, adaptability, and true ability to learn and operate independently in unstructured environments remain unclear.
**The "Gala Effect" Boosts Social Awareness and Application Expansion**
Despite questions about AI capabilities, Goldman Sachs believes the Gala's massive viewership provided a valuable opportunity to enhance social awareness and broaden public understanding of humanoid robots. For many, this was likely their first close encounter with such advanced machines, helping to demystify the technology and potentially alleviate anxieties about robotics. By showcasing humanoid robots in an entertaining and engaging context, the Gala helps foster more positive perceptions and acceptance of these technologies. This increased visibility is crucial for expanding potential applications beyond traditional industrial settings. As the public becomes more familiar and comfortable with humanoid robots, it will open doors for their integration into service industries, entertainment, education, and even personal assistance.
Goldman Sachs predicts global humanoid robot shipments will reach 51,000 units in 2026 and 76,000 units in 2027, a multi-fold increase from the 15,000 to 20,000 units forecast for 2025. This growth is primarily driven by specialized commercial deployments, such as security patrols and customer service in public spaces (e.g., hotels, banks, museums, exhibition centers, car dealerships, and supermarkets). These applications effectively leverage existing task planning, mobility, and interaction capabilities while avoiding the complexities of highly dexterous manipulation.
**The Long-Term Key: General AI Capabilities and the "World Model"**
Goldman Sachs expects the market may react positively to major humanoid robot supply chain stocks in the coming trading sessions, anticipating a surge in applications in the coming years. However, Goldman Sachs cautions investors to be prudent, as the first-quarter 2026 earnings (to be reported in March-April) could be challenging for robotics industry stocks heavily reliant on the Chinese electric vehicle market. Furthermore, inflation in raw material costs, particularly for copper, silver, gold, and memory chips, poses a potential risk to near-term margin trends. Investors chasing the trend need to fully consider these fundamental pressures.
Goldman Sachs emphasizes that, long-term, the development pace of general AI capabilities will be crucial. Driven by effective data and model strategies, this will determine the potential upside relative to their baseline 2035 expectation of 1.38 million units. Goldman Sachs is particularly focused on the "world model" approach, which involves an internal learned representation or simulation of the environment. This enables AI agents to understand how the world works, predict future states, and plan actions without requiring constant physical interaction. Such models allow robots to reason about causality, paving the way for potentially more general, capable, and safer machines.