Could Chinese AI models increasingly reshape the global AI market?

—

© [2025] [𝘒𝘐-𝘎𝘦𝘯.] 𝘔𝘢𝘳𝘤𝘦𝘭 𝘖𝘩𝘳𝘦𝘯𝘴𝘤𝘩𝘢𝘭𝘭. 𝘉𝘢𝘴𝘪𝘦𝘳𝘦𝘯𝘥 𝘢𝘶𝘧 𝘞𝘦𝘳𝘬𝘦𝘯 𝘷𝘰𝘯 𝘈𝘯𝘥𝘳𝘦𝘢𝘴 𝘖𝘩𝘳𝘦𝘯𝘴𝘤𝘩𝘢𝘭𝘭.

While they are already being used in the United States, Chinese AI models remain largely absent from the European market. According to a BBC article, a growing number of US companies are turning to Chinese AI applications, which are seen as efficient, cost-effective and publicly accessible. Their open-source approach is particularly attractive: the models can be adapted freely and further trained for internal business applications. According to company claims, such customised solutions can perform up to 30% more accurately than standard models.

In Europe, by contrast, Chinese AI applications remain relatively uncommon. In Germany, the DeepSeek app was removed from app stores in June 2025 after several state authorities launched a review. Particular concerns were raised about the storage and transfer of personal data to servers in China, for which DeepSeek was unable to provide sufficient evidence of compliance with data protection requirements. EU regulations in force since August 2024 have also made the use of such models significantly more difficult.

These regulatory differences could contribute to a fragmentation of the global AI market. Developers in Asia and the United States may be able to work with models that German companies and consumers cannot access. At the same time, the resulting pressure could create an opportunity to strengthen European AI alternatives and build the necessary infrastructure within Europe.

What does this mean for the use of AI in professional research?

Responsible handling of data, particularly sensitive or personal information, is essential in this context. Clear standards and binding rules for the use of AI are therefore crucial. At the same time, growing market fragmentation also creates risks: different AI systems are based on different assumptions about how information is gathered and weighted. For professional research, it is therefore becoming increasingly important not only to assess an AI system’s outputs, but also to consider critically where the model comes from, what it was trained on and the regulatory framework in which it operates.

By Ida Kandler