Chinese AI powers half of Korea's new models: study

The rapid adoption of Chinese AI models, particularly Alibaba's Qwen, among South Korean developers highlights a significant shift in their foundational technology choices. The study, which examined developer activity on Hugging Face, found that Chinese models accounted for 49.3 percent of new derivatives between January and August, while US models dropped to 35.9 percent and Korean models remained at 14.9 percent. This suggests developers are prioritizing practical considerations like cost and versatility.

Qwen's appeal extends to its wide range of model sizes, support for 119 languages, and readily available development tools, making customization easier for developers. This contrasts with South Korea's government-backed sovereign AI program, which has invested 200 billion won (USD 151 million) to foster homegrown alternatives. The program's independence requirements have already led to the elimination of Naver Cloud for using components from Alibaba's Qwen, highlighting a tension between policy goals and developer preferences.

Despite high developer adoption, Chinese AI models are not prominently featured in conventional usage figures; no Chinese chatbot ranked among Korea's top 10 most-used AI apps. This discrepancy suggests developers are integrating these models into their services without necessarily relying on Chinese companies' online platforms. However, concerns about data leaks were cited by 80.2 percent of surveyed businesses, indicating that security remains a significant consideration for broader adoption.

The report also warns that relying too heavily on a single model could leave developers vulnerable if its provider changes licensing terms or stops releasing new versions. For Korean open models to gain traction among developers, performance alone may be insufficient; they may also need commercially viable licenses, diverse model sizes, and robust customization tools to compete effectively.