第一财经

2026 WAIC Top 10 Highlights: Complete Lineup of Domestic Chips Debuted, AI Advances to Industrial Level

原文:2026WAIC十大看点:国产芯片全阵容亮相,AI步入产业层

Summary of Key Highlights

The most significant change at the 2026 World Artificial Intelligence Conference (WAIC) is the shift of AI from a focus on technological demonstrations to practical industrial applications, encompassing the entire industry chain from chip computing power, intelligent agents, humanoid robots, to consumer-grade devices. The conference showcased this transition through three interconnected areas: Zhangjiang, Expo, and Xuhui. It highlighted super-node clusters, “digital employees” capable of performing tasks, robots that can integrate into factories and homes, user-friendly AI products, as well as innovations from startups and new developments in global AI governance, outlining a clear path for AI to move from the laboratory to the real world.

Detailed Analysis

1. Advancing Computing Power: From Single Chips to Super-Node Clusters

In the past, the competition revolved around the computing power of individual chips; this year, it has shifted to super-node clusters—combining multiple chips to form a powerful computing unit, akin to transforming a single soldier into an entire army capable of handling more complex AI tasks (such as training large models).

  • Zhangjiang Chip Showcase: Domestic GPU manufacturers gathered for the first time. The DF1000 chip from Dongfang Suanxin achieved 4nm-level performance using 14nm technology, while companies like Muxi and TianShenZhiXin demonstrated a range of chips for various applications. Moore Threads even completed the training of a model with 236 billion parameters.
  • Expo Super-Node Exhibition: Huawei displayed the Atlas 950 SuperPoD, which can be expanded to 500,000 cards, claiming to be the most powerful in the world. Companies like Inspur, Alibaba, and ZTE also presented super-node solutions, serving as the core driving forces behind AI models and industrial applications.
  • Green Computing Power Becomes Standard: Liquid cooling technology is no longer just a concept; Huawei and Inspur showcased commercially viable liquid-cooled clusters of thousands of cards, addressing the high energy consumption issue associated with AI (liquid cooling is more energy-efficient than air cooling).

2. Intelligent Agents Evolving into “Workers”: From Chatbots to Digital Employees

Previously, intelligent agents were merely chat tools; this year, they have evolved into digital employees capable of independently completing complex tasks.

  • Baidu’s Innovations: Baidu’s systems can assist with writing competitive analysis reports and remember user habits, automatically transitioning between different applications (e.g., generating PPTs from documents).
  • Siemens’ Engineering Intelligence Agents: These agents can write industrial code and configure systems, increasing efficiency by 50%, equivalent to the work of a senior engineer.
  • Intelligent Operating Systems: The STEPX Neo smartphone features the world’s first AI-native operating system. It is designed from the ground up for intelligent agents, allowing the phone to make decisions and take actions on its own (e.g., scheduling appointments and managing emails).

3. Humanoid Robots Entering Everyday Life: From Concepts to Practical Use

Humanoid robots are no longer just decorative exhibits; they can now be used in real-world scenarios.

  • Industrial Applications: Zhiyuan demonstrated a robot that can perform material handling tasks on an actual production line, while Shizhihang’s robot can assemble car wiring.
  • Consumer Applications: Yutree’s GD01 is a humanoid transformable robot that can switch between human form and four-legged mode; Fourier’s desktop robots can engage in conversations, and Ruierman’s robots can make coffee. Qianxun Intelligence’s robots can handle complex tasks like organizing the living room.
  • Core Technology Breakthroughs: Jizhijia released the Gravity framework, enabling robots to understand the three-dimensional world and improve their performance over time (they learn from mistakes).

4. Consumer-grade AI Becoming Accessible: From Concepts to Everyday Use

The Xuhui exhibition area was a “fun park” for ordinary people, featuring 64 consumer-grade products that made AI more accessible.

  • Smartphones: The Honor Robot Phone is a humanoid smartphone that can interact with users and assist them with tasks.
  • AI Glasses: Technologies like Xinmou Technology’s glasses are lightweight and record meeting summaries automatically, while Liangliang Vision’s glasses provide real-time translations to overcome language barriers.
  • Other Devices: Companion robots, traditional Chinese medicine intelligent devices, and 3D generation tools cover various daily scenarios such as entertainment, health, and learning.

5. Accelerating the Growth of the AI Ecosystem: Both Startups and Global Governance Play a Role

This year’s WAIC saw not only large companies but also many startups presenting new ideas. There were also significant developments in global AI governance.

  • Startups with Potential: The H4 hall featured 153 startups, including XingjieZhiwei (real-time video generation) and Liangkun Technology (quantum + AI intelligent agents that raised tens of millions within five months).
  • New Directions in Global Governance: Nine Turing Award/Nobel Prize winners attended the conference, with Sutton, the “father of reinforcement learning,” giving a speech. An international AI cooperation organization was established to shape global AI development guidelines.
  • Educational Transformations: The “Human-AI Co-Education” forum discussed how AI will impact education and what kind of talents universities should cultivate for the AI era.

Conclusion

The most impressive aspect of this year’s WAIC is that AI is no longer a novelty confined to laboratories; it has truly entered factories, offices, and homes. From the infrastructure of computing power to consumer products, from digital employees to robots, a complete AI industry chain is now evident. Ordinary people can use AI glasses, companies can purchase super-node computing solutions, and developers can find new opportunities for innovation. Artificial intelligence has finally moved beyond the stage to become a reality in our lives.