2025 WAIC: A Comprehensive Observation from L4-Level AI to Robots That Can Deliver Packages
A prominent financial news app has learned that the 2025 World Artificial Intelligence Conference (WAIC) was grand in scale, with numerous participants and a focus on the latest technological advancements and commercial applications in artificial intelligence. This article provides an overview of the conference from the robotics industry, large models, AI applications, hardware innovation, and market and industrial ecology. In recent times, the AI sector has experienced a significant rebound, with most companies benefiting, reaching as high as 50%+ or as low as 10%-20%. The unexpected events at WAIC (difficult-to-buy tickets, stop reporting after July) and GPT5 have led to a sustained market trend, with institutions adopting both high- and low-cut strategies to maintain their positions. A-share computer/AI applications are currently in a bearish market, with hotspots changing quickly, lacking the continuity of alpha returns; corporate performance growth has been stable (with most revenue growth reaching 10%+), but AI revenue is still small and has not yet reached its explosion phase.
One: Conference Overview
The 2025 World Artificial Intelligence Conference (WAIC) was grand in scale, with numerous participants and a focus on the latest technological advancements and commercial applications in artificial intelligence. The robotics exhibition area stood out, with an unprecedented number of exhibitors, exhibition areas, and attendees. The total exhibition area increased by approximately 40% compared to last year, with over 60 robot manufacturers participating, more than double the number from last year.
Two: Exhibition Features and Technical Highlights
Zone 1: A prominent zone featuring large companies such as Baidu, Alibaba, etc., with expanded exhibition areas, diverse application showcases, and rapid progress in endpoint models. For example, a car-endpoint model achieved intelligent interaction optimization, while a PC-linearized endpoint model enhanced document processing efficiency.
Three: Robotics Industry: From Technical Breakthrough to Commercialization Landing
1. Product and Scenario Progress
Industrial scenarios: Zhiyuan Technology's wheeled robot, in cooperation with DeMa Technology, achieved autonomous sorting with an error rate of ≤5mm; Flyx Technology's robots utilized self-developed torque sensors to grasp fragile items such as glass and chips; KPL's bipedal and wheeled robots focused on material handling, while Yuexin's wheeled bottom plate moved efficiently in factories.
2. Challenges and Technical Directions
Currently, most robots still require human control or lack autonomy, requiring an increase in "brain" intelligence (multidimensional perception, autonomous decision-making, and complex task decomposition). Domestic companies are still lagging behind international companies such as Tesla and bigger.