China has solidified its strategy for building an independent artificial intelligence (AI) ecosystem by showcasing AI accelerators, computing systems, open-source models, and humanoid robots at the 2026 World Artificial Intelligence Conference (WAIC). The country is emphasizing technological self-reliance while also seeking to expand its influence in global AI governance.
According to industry sources, more than 1,100 companies participated in WAIC 2026, held in Shanghai from July 17 to 21, presenting over 3,000 products and technologies. The exhibition covered approximately 100,000 square meters, with around 300 products unveiled for the first time.
This year's WAIC has also expanded to showcase China's vision for international AI cooperation. Chinese President Xi Jinping attended the opening ceremony for the first time since the conference's inception, highlighting the controllability of AI while opposing technology restrictions based on national security. On the eve of the opening, representatives from 29 countries signed an agreement to establish the World Artificial Intelligence Cooperation Organization (WAICO), signaling China's intent to enhance its influence in global AI governance.
From an industrial perspective, China's achievements in AI technological self-reliance were underscored. Companies like Huawei have unveiled products and services that span AI development to industrial applications, despite restrictions on access to advanced U.S. semiconductors.
Huawei introduced its Ascend-based AI computing system, the Atlas 950 Superpod, showcasing a configuration of 1,024 neural processing units (NPUs) that can be expanded to a maximum of 8,192. Other companies, including ZTE, also presented supernode-based AI infrastructure, integrating semiconductors, servers, connectivity technologies, and software to meet the demands of large-scale AI model training and inference. Industry analysts suggest that Chinese companies are strengthening their system capabilities in light of U.S. semiconductor access restrictions.
The competitive landscape in AI has shifted towards agent-based models that perform real-world tasks. Chinese companies are promoting open-source models and executable AI, aiming to secure an ecosystem that can be rapidly applied in industry and services.
Notably, Moonshot AI drew attention by unveiling its Kimi K3 open-weight model, which boasts 28 trillion parameters and has demonstrated performance on par with top U.S. models in independent evaluations. Following the development of the high-performance AI DeepSeek at a low cost last year, assessments indicate that the performance gap in AI technology between the U.S. and China is narrowing quickly. This strategy is interpreted as an effort to expand the global development ecosystem through open models.
Embodied AI and humanoids have also emerged as key topics. Over 200 companies participated in the fields of intelligent computing and embodied intelligence, with more than 300 robots demonstrating tasks such as parts assembly and logistics operations. China is accelerating the application of humanoids in industrial settings based on its manufacturing supply chain, which encompasses everything from motors and gear reducers to sensors, batteries, and final product assembly.
Additionally, WAIC introduced a dedicated exhibition area for one-person startups (OPC) for the first time, attracting 180 companies. This area highlighted individuals and micro-organizations utilizing AI tools for development, planning, design, and marketing tasks.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.