China's leading artificial intelligence companies have joined forces to launch two new industry alliances aimed at building a self-sufficient domestic AI ecosystem. The move comes as Chinese firms scramble to reduce reliance on foreign technology, particularly U.S.-made semiconductors, in the face of tightening export restrictions on advanced Nvidia chipsets.

The twin alliances were announced during the World Artificial Intelligence Conference (WAIC) in Shanghai, which concluded Monday, drawing significant attention to China’s ambitions in AI development.

Strategic Alliances for Domestic Resilience

The Model-Chip Ecosystem Innovation Alliance is the more technically focused of the two, bringing together large language model (LLM) developers and domestic chipmakers. Among its founding members are Enflame, Huawei, Biren, and Moore Threads — all manufacturers of graphics processing units (GPUs) affected by U.S. export controls. The alliance, spearheaded by LLM developer StepFun, aims to create a vertically integrated ecosystem encompassing chips, AI models, and computing infrastructure.

“This is an innovative ecosystem that connects the complete technology chain from chips to models to infrastructure,” said Zhao Lidong, CEO of Enflame, emphasizing the importance of internal collaboration under external pressure.

The second alliance, formed under the Shanghai General Chamber of Commerce AI Committee, aims to integrate AI more deeply into China's industrial landscape. Participants include prominent AI companies such as SenseTime, which has pivoted to LLMs after being sanctioned for its facial recognition tech, as well as StepFun, MiniMax, and chipmakers Metax and Iluvatar CoreX.

Huawei Takes the Spotlight

Huawei dominated the hardware conversation at WAIC with the unveiling of its CloudMatrix 384, a powerful AI computing system featuring 384 of its latest 910C chips. According to U.S. research firm SemiAnalysis, the system matches or even outperforms Nvidia’s cutting-edge GB200 NVL72 in some benchmarks, thanks to Huawei’s system-level innovations and chip clustering strategies.

While individual Chinese chips still lag behind Nvidia’s in raw performance, companies like Huawei are closing the gap through intelligent architecture and software optimization. Six other Chinese tech firms, including Metax, also showcased similar “clustering” approaches. Metax's demonstration included an AI supernode using 128 C550 chips, designed for large-scale, liquid-cooled data center environments.

Consumer-Facing Innovations

The conference wasn't just about backend breakthroughs. Tech giants like Tencent, Baidu, and Alibaba also revealed cutting-edge AI applications aimed at consumers and businesses.

  • Tencent introduced its open-source Hunyuan3D World Model 1.0, enabling users to generate interactive 3D environments using simple text or image prompts.
  • Baidu showcased next-gen “digital human” tech that can create virtual livestreamers. This system can clone a person’s voice, tone, and gestures from just 10 minutes of footage — a potential game-changer for content creators and marketing.
  • Alibaba unveiled Quark AI Glasses, set for release in China by late 2025. Powered by its proprietary Qwen AI model, the glasses will support navigation, QR code scanning, and Alipay integration via voice command.

A Unified Push for AI Sovereignty

These developments reflect a coordinated push by China to build a self-reliant AI ecosystem in the face of geopolitical and technological challenges. With Nvidia’s most powerful chips increasingly out of reach due to U.S. sanctions, Chinese tech firms are pivoting towards domestic solutions — both in hardware innovation and in industry-wide collaboration.

By tightening cooperation between chipmakers and AI developers, and by showcasing a growing portfolio of consumer and enterprise-grade AI products, China is signaling that it intends to stay competitive — even as global tech tensions reshape the landscape.