What I saw at CIIF 2026 in Shanghai: AI-powered robot arms, precision manipulation, humanoid technologies, and the possibility of a new open robotics ecosystem.
China’s robotics industry is attracting enormous attention, particularly around humanoid robots. But during my visit to the China International Industry Fair (CIIF 2026) in Shanghai, I found something more interesting: AI is beginning to change conventional industrial robotics.
This could affect not only how robots work, but also who develops them.
I recorded a 30-minute walkthrough of the exhibition, showing actual demonstrations of industrial robots, AI-related technologies, sensors, actuators, and manufacturing equipment.
Watch the full CIIF 2026 video tour:
https://youtu.be/rAR8wkpvtGAAI Is Entering Conventional Industrial RoboticsChina’s industrial robotics industry already has a large customer base, established manufacturers, and extensive supply chains.
Companies such as FANUC, ABB, KUKA, and Yaskawa have long supplied robots to factories. Chinese manufacturers, including DOBOT and JAKA, have expanded the market for collaborative robots and newer industrial automation solutions.
Traditionally, industrial robots have been programmed for specific operations, often through proprietary controllers and teaching systems. They excel at repetitive tasks requiring precision and reliability.
At CIIF, I saw demonstrations combining conventional robotic arms with computer vision, learning technologies, and new software environments.
Watch 02:43 — AI and robot learning in industrial robotics
The robot hardware sometimes looked familiar. The development environment was changing.
Watch 04:22 — Digital twins and robot control
One particularly interesting demonstration involved a robot inserting a USB connector.
Watch 08:21 — Precision manipulation: inserting a USB connector
USB insertion is easy for humans but challenging for robots. Small positioning errors can prevent insertion, while excessive force can damage the connector.
This type of task requires precise coordination between sensing and motion control.
Other exhibits demonstrated force control, torque sensing, and manipulation involving physical contact.
Watch 13:56 — Force control and torque sensors
These demonstrations do not necessarily prove that every movement is autonomously generated by AI. Exhibition robots may use predefined motions or carefully prepared conditions.
However, they illustrate the growing overlap between traditional industrial automation and technologies developed for AI robotics.Why AI Could Attract New Engineers to Industrial RoboticsI’ve spent many years working with makers, hardware startups, and open-source communities in Shenzhen.
Engineers in these communities often develop prototypes using shared software libraries, open-source hardware, and tools such as ROS. Many come from university robotics competitions, personal projects, or startup environments.
Traditional industrial robotics has a different development culture.
Large manufacturers supply robot arms, controllers, and specialized software. System integrators deploy them in factories, where safety, precision, and long-term reliability are essential.
This creates significant barriers for individual developers.
AI introduces a different set of technical requirements: computer vision, machine learning, simulation, robot learning, and data engineering.
These skills are common among engineers working on humanoids, autonomous vehicles, drones, and AI startups.
Watch 19:49 — Vision-Language-Action models and robotics research
As AI becomes part of industrial robotics, engineers from these fields may begin working on factory automation.
This is particularly interesting in China, where the industrial robotics market already has customers and established manufacturing infrastructure.
Young engineers may not need to create entirely new applications for humanoid robots. They could apply AI technologies to existing industrial problems.
Watch 21:52 — How new robot systems differ from conventional industrial robots
For me, this potential movement of engineers between industries is one of the most important developments to watch.Could Open-Source Development Create a New Category of Industrial Robots?Arduino, Raspberry Pi, and M5Stack have demonstrated how accessible development platforms can expand hardware ecosystems.
Their value goes beyond inexpensive hardware. Documentation, software libraries, examples, and developer communities make it easier for individuals to build prototypes and eventually develop commercial products.
Could something similar happen in industrial robotics?
Open robotics projects such as ROS and Hugging Face’s LeRobot already make parts of robotics development more accessible. The SO-101 robot arm, for example, allows developers to experiment with robot learning using relatively accessible hardware.
Watch 17:47 — Direct human control of a robotic arm
Imagine industrial robots where developers can combine robotic arms, grippers, cameras, and AI models from different providers.
Developers could share training datasets and task-specific models. Others could modify and improve them. Companies could build specialized industrial applications on top of these shared technologies.
Industrial robots still require rigorous safety engineering, reliability, and maintenance. Open development tools cannot replace those requirements.
But more of the development process could become accessible to companies and individual engineers outside the traditional industrial robotics manufacturers.
This could eventually create a new category of industrial robots: machines designed for industrial applications, but supported by more open software and developer ecosystems.
That is a possibility rather than something already established at CIIF. The exhibition showed some of the technologies that could help move industrial robotics in this direction.Humanoid Robots May Accelerate Industrial RoboticsThe commercial future of humanoid robots remains uncertain.
Many companies are developing humanoids, but their long-term applications and economics are still being tested.
However, humanoid development is driving improvements in actuators, reducers, motors, sensors, force control, and AI-based motion systems.
These technologies can also benefit conventional industrial robots.
Watch 16:25 — How humanoid development affects the robotics component industry
CIIF featured a wide range of component suppliers, manufacturing equipment companies, and automation providers.
Video highlights:05:28 — Robot drive systems and control components14:51 — Factory automation and manufacturing services15:41 — Heavy-duty handling equipmentChina’s manufacturing services are also becoming more accessible. Companies such as JLC have expanded their offerings from PCB production to CNC machining, sheet metal fabrication, and 3D printing.
These services help robotics startups prototype and iterate more quickly.
Even if some humanoid startups eventually fail, the engineers, components, and technologies developed during the current boom may find applications elsewhere.
Industrial robotics is a particularly promising destination because it already serves a substantial market.What CIIF Reveals About China’s Robotics EcosystemShenzhen is an exciting place to see new hardware startups, drones, and experimental products.
CIIF in Shanghai provides a different perspective.
Established industrial robotics manufacturers, newer Chinese companies, automation suppliers, universities, and research institutions exhibit together.
Watch 23:25 — The robotics market and supporting ecosystem
During a previous visit to Xiaomi’s automated factory, I saw many FANUC industrial robots. Even advanced Chinese factories continue to rely on established Japanese and European robotics technologies.
Watch 22:59 — Comparing CIIF exhibits with Xiaomi’s automated factory
At CIIF, university research and industrial applications were also displayed in close proximity.
Watch 26:14 — University research and industry collaboration
The exhibition provided an opportunity to see how established manufacturers, startups, researchers, and component suppliers are approaching the same technological changes.The Next Important Robot May Not Be a HumanoidWatch 28:04 — Final observations from CIIF 2026
My main interest in China’s hardware industry is how technologies move from prototypes to commercial products.
That process depends on more than the performance of an individual product. Development tools, components, documentation, manufacturing services, and communities all matter.
AI could bring new engineers and development practices into conventional industrial robotics, connecting an established manufacturing market with the rapidly evolving AI and open-source communities.
Whether this results in a genuinely open industrial robotics ecosystem remains to be seen.
But the possibility is worth watching.
The next major change in industrial robotics may come not only from better machines, but from a different generation of people building them.Watch the Complete CIIF 2026 Exhibition TourSee the industrial robots, AI demonstrations, manufacturing equipment, and component suppliers discussed in this article.
▶ Watch CIIF 2026 Shanghai — Full 30-Minute Video Walkthrough
Masakazu Takasu is based in Shenzhen, China, and works with hardware startups, makers, and open-source communities. He is a partner at Heroad Investments and co-founder of Nico-Tech Shenzhen.
Topics: Industrial Robotics, Physical AI, Open Source, China Manufacturing, Robotics Startups, Developer Ecosystems.