{
  "id": 19144764,
  "title": "Beyond LLMs? China forges new path for AI with focus on industrial systems",
  "url": "https://urgent.news/2026/11/09/beyond-llms-china-forges-new-path-for-ai-with-focus-on-industrial",
  "topic": "ai",
  "section": "AI",
  "published": "2026-11-09T11:32:21.000Z",
  "source": {
    "name": "SCMP Tech",
    "slug": "scmp-tech",
    "url": "https://www.scmp.com/tech/tech-trends/article/3373822/beyond-llms-china-forges-new-path-ai-focus-industrial-systems"
  },
  "original_language": "en",
  "account": "China is shifting its focus in artificial intelligence development away from general large language models (LLMs) towards specialized systems tailored for addressing real-world industrial challenges, according to a prominent computer science researcher from Tsinghua University, Cui Peng. At the Hong Kong International Computer Conference, Cui emphasized that LLMs are ill-suited for industrial applications due to their difficulty in handling structured numerical data and patterns. While LLMs, developed by tech giants like OpenAI and Anthropic, have been instrumental in the global AI surge, they struggle to effectively interpret data from factory sensors and financial records, as highlighted by Cui.\n\nBeijing's \"AI plus\" initiative aims to embed AI technology across various sectors, including manufacturing, logistics, and healthcare. The State Council plan set a goal for AI agents and intelligent devices to permeate Chinese society with a penetration rate of over 70 percent by 2027. At the two-day conference themed “Hong Kong’s Next Five Years: From AI+ to Competitive Advantage,” Cui shared that his research team has successfully developed LimiX, a large data model trained exclusively on synthetic data. This model is adept at recognizing causal patterns and excelling in industrial and scientific tasks, such as controlled nuclear fusion, with a projected reduction in deployment costs and time by around 90 percent, alongside an accuracy boost of over 15 percent on average.\n\nChinese tech companies, such as Tencent Holdings and Huawei Technologies, are leveraging AI to advance scientific discovery. Tencent’s Chief AI Scientist, Vinces Yao Shunyu, announced plans to create \"autonomous science labs\" capable of conducting intricate experiments, including those in protein research. Additionally, Huawei’s Global Chief Architect of Digital Government, Nie Jun Yu, pointed out that China has transitioned from isolated AI applications to industry-wide systems, often overseen by local governments. This shift differs from the United States and Europe, where AI development often emphasizes competition based solely on model capabilities. Instead, China prioritizes wide accessibility and infrastructure development through local government coordination, particularly in regions like Zhejiang and Guangdong.\n\nHong Kong, under the leadership of Chief Executive John Lee Ka-chiu, is enhancing its AI computing infrastructure and bolstering supportive measures to accelerate AI adoption across industries. The city aims to achieve a 36-fold increase in compute capacity by 2032, positioning itself as a leading international innovation and technology hub. The ongoing discussions at the Hong Kong International Computer Conference reflect Hong Kong's commitment to aligning its development goals with Beijing’s vision, which places significant emphasis on technological innovation and the integration of AI into the economy.",
  "summary": "China’s artificial intelligence development is expanding beyond general large language models (LLMs) towards a new generation of specialised systems designed to solve real-world industrial problems, according to a leading computer science researcher. Cui Peng, an associate professor at Beijing’s elite Tsinghua University, said that specialist systems capable of analysing industrial data, such as…",
  "key_points": [
    "China pivots AI focus from LLMs to industrial-specific systems",
    "Beijing's AI Plus initiative targets 70% penetration by 2027",
    "LimiX model developed to recognize causal patterns in industrial tasks"
  ],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}