2025-07-29
“Global Debuts” and “China Premieres” Highlight the New Momentum of AI Innovation
Source:Economic Information Daily
The 2025 World Artificial Intelligence Conference (WAIC) was held in Shanghai from July 26 to 28. For the first time, the exhibition area exceeded 70,000 square meters, drawing more than 800 participating companies and featuring over 3,000 cutting-edge exhibits, as well as over 100 new products making their “global debut” or “China premiere.” This marked the largest edition of the event to date… This global AI conference offered a window into the rapid pace of technological innovation—from embodied intelligence and multimodal large models to a constant stream of groundbreaking products such as humanoid robots and AI terminals. These advancements continue to inject fresh momentum into economic growth.
At this year’s conference, humanoid robots once again captured the spotlight. Compared to 18 companies last year, more than 80 related enterprises took part this time. Along with the launch of many new products, humanoid robots are rapidly expanding their range of applications from playing boxing and curling to hammering nails and delicately handling tofu, accelerating their shift from industrial settings to everyday life.
On the opening day of the conference, the National and Local Co-built Humanoid Robotics Innovation Center introduced three humanoid robots from its “Qinglong” series. Each model is tailored for specific application scenarios ranging from security and manufacturing to science education and tourism, industrial operations, and urban services. According to Liu Yufei, Deputy General Manager of the Innovation Center, each product is geared toward the practical, real-life use of humanoid robots. For example, the Qinglong Pro robot is equipped with a full-body array of multidimensional sensors, including visible light, infrared thermal imaging, and gas detection. It also features upgraded spatiotemporal intelligent positioning capabilities. This enables it to operate smoothly across both indoor and outdoor environments, making it well-adapted for use in parks, factories, and other outdoor conditions.
Fourier unveiled its first interactive companion humanoid robot, the GR-3, at the conference. A key innovation is its soft-skin covering, which gives the robot a more lifelike and warmer feel. Meanwhile, KEENON Robotics made the global debut of its bipedal embodied service robot, the XMAN-F1, which effortlessly carried out tasks such as popping popcorn and preparing personalized iced beverages.
While hardware products are being upgraded at an accelerated pace, software, especially models and platforms, is also advancing rapidly. At the conference, AgiBot unveiled its first open-source world model platform, “Genie Envisioner,” which enables robots to rehearse in a digital spatial-temporal environment before taking on real-world tasks.
“The core of the platform is GE-Base, a multi-view video diffusion model. By leveraging over one million synchronized video streams, recorded from head and dual-arm wrist views, spanning nearly 3,000 hours, it can effectively learn spatial layouts, action dynamics, and semantic intent involved in robot manipulation tasks,” explained Yao Maoqing, President of AgiBot’s Embodied Intelligence Division. Looking ahead, the platform will integrate a broader range of sensor modalities to enable full-body mobility and human-computer collaboration, accelerating the practical adoption of intelligent manufacturing and service robots.
During the conference, Du Guangda, Deputy Director of the Department of Science and Technology at the Ministry of Industry and Information Technology, noted that in recent years, China’s humanoid robot products have made steady strides in innovation, achieving the ability to stand firmly, walk steadily, and run swiftly. They are shifting from stage demonstrations to real-world factory applications, moving rapidly from the lab into a phase of large-scale production and rapid commercial rollout. Looking ahead, the Ministry of Industry and Information Technology will promote tiered applications within the humanoid robot industry, support pilot zones and state-owned enterprises in expanding practical use scenarios, and create model demonstration sites for humanoid robot applications.
Meanwhile, large AI models are being rapidly adopted across industries such as electronics and consumer goods, with AI-powered devices such as AI phones, AI computers, and AI glasses emerging one after another.
With the simple voice command “Xiaochen, Xiaochen, take a photo,” users can instantly capture stunning moments. At the China Telecom booth, the newly released “Tianyi AI Glasses” drew crowds eager to try them out. According to staff on site, powered by the self-developed “Xingchen Large Model” technology, these AI glasses are continuously expanding the boundaries of human-computer interaction, enabling features like first-person perspective content creation during outdoor adventures and real-time English-Chinese interpretation in meetings, greatly enhancing the smart living experience.
“AI glasses are the most significant product category in the wearable technology sector and are poised to become a new pair of ‘eyes’ and ‘ears’ for users,” said Song Gang, Head of Terminal Business at Alibaba’s Intelligent Information Business Group, during the conference. He also revealed that Alibaba is expected to officially launch its AI glasses later this year, with anticipated breakthroughs in AI interaction, wearing, display and imaging, and endurance.
Beneath the ever-growing range of smart applications, foundational computing power continues to evolve. “Building large-scale computing capacity is like constructing a ‘giant power plant,’ while AI applications require ‘plug-and-play outlets’ for seamless integration. AI computing infrastructure must be tailored to meet the agile computing power demands of small and diverse AI companies,” said Xia Lixue, co-founder and CEO of Infinigence AI. At the conference, the company unveiled its first full-scale AI performance enhancement solution. Designed to address a broad range of scenarios, including cross-regional intelligent computing networks, intelligent computing clusters, and diverse intelligent terminals, it delivers unified management of diverse computing resources, offering end-to-end support from model scheduling and performance optimization to AI application deployment.
In the eyes of participants, the rapidly evolving AI technology is steadily unleashing fresh momentum for industrial upgrades and overall productivity growth, fueling a new engine for the intelligent economy.
“Personalized AI agents, new business forms such as human-computer collaborative production, and distributed transactions are driving the formation of ultra-large-scale collaborative innovation networks, which are in turn enabling exponential improvements in resource allocation efficiency and innovation speed across society, thereby accelerating the robust growth of the intelligent economy,” said Yang Jie, Chairman of China Mobile. Looking ahead, the intelligent economy will evolve into a more advanced form of the digital economy. Forecasts suggest that it could raise the global GDP growth rate by one percentage point annually, injecting lasting momentum into the development of new quality productive forces.