LLM expert interview: The trend of China’s large model industry is clear and the commercial implementation process is accelerating (Nomura)
After interviews with LLM experts, Nomura Securities pointed out that China’s large model industry is shifting from an “arms race” to “application implementation.” Experts say the gap in model capabilities is narrowing, with the popularity of open source model
After interviews with LLM experts, Nomura Securities pointed out that China’s large model industry is shifting from an “arms race” to “application implementation.” Experts say the gap in model capabilities is narrowing, with the popularity of open source models enabling more companies to quickly build applications. The key change is that the commercialization process is accelerating, especially in To B fields, such as finance, medical and education industries, where the demand for enterprise-level AI agents and knowledge base solutions has begun to explode. This is different from the "conceptual hype" generally expected by the market. Experts believe that real demand is taking shape. Investors should focus on application layer companies with vertical industry understanding and customer resources, rather than just underlying model developers. One sentence conclusion: The value of China's large model industry is migrating to the application layer. Companies with industry knowledge and technology implementation plans will be the first to achieve commercial breakthroughs, rather than simply pursuing model capabilities. Good/bad: Good for Chinese To B software and SaaS companies as well as AI solution providers applied in the industry. The current market hype for AI computing power may have priced in most expectations, but the commercial explosion on the application side has not yet been fully priced. Catalysts: 1) AI cloud revenue or large model application cases disclosed by leading Internet companies (such as Tencent and Alibaba) at their 2Q26 results conference; 2) Further clarification of industry regulatory policies to encourage the development of AI applications; 3) Benchmarking enterprise-level AI order landing announcements.