Key Takeaways:
The 2026 World Artificial Intelligence Conference marked a turning point for the industry, with AI moving beyond chatbots into factories, pharmacies and daily life.
The 2026 World Artificial Intelligence Conference showed AI has moved beyond digital dialog boxes into physical production lines, with over 300 global debuts across a record 100,000-square-meter exhibition floor in Shanghai.
"The core value of industrial AI lies not in novel concepts, but in its ability to penetrate R&D, production and operation procedures, solve visible problems in real scenarios and deliver quantifiable returns," Xiao Song, executive vice president of Siemens AG and chairman of Siemens China, said at the event.
CMBI identified three core trends from the conference: continued optimization of AI infrastructure, the shift from digital to physical world applications, and expanding commercialization from applications to terminals. Huawei showcased the Ascend 950 super node, while Alibaba's T-Head unit displayed the Panjiu AL128 super node integrating 128 M890 chips in a single cabinet. Moonshot AI open-sourced the Kimi K3 model with 2.8 trillion parameters, and Ant Group's LingBot-VLA 2.0 adapted to more than 20 robot configurations from 17 manufacturers including Leju, AgiBot and Unitree.
CMBI remains positive on Tencent, Alibaba, Z.AI and MINIMAX under the AI theme. Tencent's WorkBuddy launched a standalone app, Alibaba released Meoo Team Edition, and SenseTime introduced Seko 3.0 for AI video generation — product launches that show monetization of AI tools is accelerating across China's technology sector.
From Model Parameters to Real-World Deployment
At previous WAIC editions, companies competed on large model parameter counts and benchmark scores. This year, the focus shifted to practical deployment. Siemens unveiled Eigen, described as the world's first AI agent for industrial automation engineering, capable of independently completing end-to-end tasks from interpreting project requirements to developing control software. The product has been deployed at hundreds of enterprises across multiple countries and won the WAIC 2026 SAIL Star Award.
Schneider Electric presented end-to-end AI-enabled solutions covering the full life cycle of energy and industrial operations, including a native power management agent that connects equipment access, data collection, knowledge accumulation, intelligent diagnosis and execution. ABB Robotics co-released an industry white paper with Nvidia establishing a methodology for rapid physical AI deployment through digital twins and synthetic data training.
On the embodied intelligence front, robots at this year's exhibition were deployed in genuine working environments rather than executing preset movements. Multiple real-world scenarios were reproduced at the venue, including automotive production lines and offline pharmacies, where heterogeneous robots completed full workflows such as handling, assembly, inspection and drug sorting without manually set scripts.
Investment Implications and the Commercialization Race
The shift from model competition to real-world deployment has changed how venture capital evaluates AI companies. Among more than 1,000 exhibitors, 130 enterprises were in the large model and generative AI track, with 83 focused on launching AI agents covering 17 practical application scenarios, according to technology think tank Jiazi Guangnian.
JD.com's JoyInside has partnered with nearly 200 brands and plans to connect to more than 10 million terminal devices in 2026. Ant Group's Afu, an AI health application, delivers a full-chain service system covering health consultation, report interpretation, online medical consultation and appointment registration, leveraging digital AI avatars created from authoritative medical experts.
For investors, the key question is which companies can translate AI capabilities into sustainable revenue. CMBI's positive stance on Tencent, Alibaba, Z.AI and MINIMAX reflects the view that these companies have the distribution, data and product breadth to monetize AI across multiple use cases. Tencent trades at about 22 times forward earnings, while Alibaba trades at roughly 11 times — valuations that leave room for upside if AI products drive meaningful revenue growth.
This article is for informational purposes only and does not constitute investment advice.
