China’s drive toward self-reliance in AI: from chips to large language models 

On Tuesday, July 22, MERICS held an online presentation and discussion of the upcoming report "China’s drive toward self-reliance in AI: from chips to large language models". 

China sees AI as critical for national and economic security and is pursuing self-reliance across the technology stack. Beijing has made significant efforts to develop home-grown solutions from large language models like DeepSeek to machine learning frameworks used to build them and the semiconductors powering AI workloads. This drive has taken on renewed urgency in the face of U.S. export controls on advanced AI chips. 

The new MERICS report examines the progress China has made across these different layers, and the state’s methods in helping development. It explores China’s domestic inputs as well as the external dependencies such as foreign manufacturing equipment or access to the international open-source software community. The authors also point to implications for Europe across several dimensions (economic security, competitiveness, national security, geopolitics, and values). 

The report was presented by one of the authors, MERICS Analyst Wendy Chang. This was followed by an input from Jimmy Goodrich, Senior Advisor for technology analysis to the RAND Corporation. After that, participants were invited to ask questions and engage in the discussion. The discussion was moderated by MERICS Lead Analyst Rebecca Arcesati.

The report is part of the China Tech Observatory (CTO) project funded by Federal Ministry of Research, Technology and Space.