China Surpasses US in Open AI Advancement, Challenging Tech Hegemony
China is rapidly emerging as a dominant force in the open AI ecosystem, surpassing the United States for the first time in the share of new open-source AI models downloaded globally. This shift signals a potential realignment of power in the artificial intelligence landscape, as Chinese companies embrace open-source strategies while American tech giants increasingly focus on closed, proprietary systems.
The ascent of Chinese AI is driven by a surge in performance and accessibility. Recent reports indicate that models developed in China are achieving comparable, and in some cases superior, results to those from leading US companies. On May 1st, Bloomberg reported that DeepSeek’s V3.2 model demonstrated performance on par with OpenAI’s anticipated GPT-5 in several key inference benchmarks. Notably,the V3.2 model, specifically designed for complex problem-solving, achieved a score of 96 on AIME problems, exceeding Google’s Gemini 3.0 Pro’s score of 95. It also achieved top scores in the international Mathematics Olympiad (IMO) and the International Olympiad in Facts (IOI).
DeepSeek developers describe V3.2 as the first model engineered to “think on its own,” autonomously handling tasks like search and calculation – a crucial step toward more sophisticated agent-type AI. This evolution moves beyond simple language generation and chatbot functionality, enabling AI to perform real-world actions.
The growing influence of Chinese AI is quantifiable. According to a report released in late April by Hugging Face, a leading open-source AI model platform, Chinese-developed models accounted for 17% of new open-source model downloads this year, exceeding the 15.8% share held by American models from companies like Meta. This marks the first time China has held the lead in the open-source AI field in the past five years. “The United States’ dominance in the open source AI field is decreasing and China’s influence is growing,” researchers stated.
This trend is further exemplified by ‘Kimi K2 Thinking,’ released as open source by Moonshot AI last month, which outperformed American AI models in major benchmark evaluations. While companies like OpenAI and google continue to invest heavily in closed large language models, Chinese developers are gaining traction by offering freely downloadable models with comparable capabilities.
This strategy is reshaping the competitive landscape. Although closed models from US companies currently maintain a performance edge, China is gaining ground in a new metric: widespread adoption. As one analyst noted, the standard for evaluating AI is shifting from pure performance to how widely models are utilized by developers worldwide.This increasing usage is poised to redefine the technological landscape and potentially shift the center of AI influence eastward.
The economic factors driving this shift are also important. Eric Schmidt, former CEO of Google, highlighted at an AI Summit at the University of Southern California (USC) in late April that many startup developers are choosing Chinese AI models “simply because it is cheap, not for political reasons.” He further predicted that China will likely release its advancements in artificial general intelligence (AGI) as open source, accelerating its impact on the world.
Why is this happening? China’s rise in open AI is due to a strategic shift towards open-source development, driven by cost-effectiveness and accessibility for developers.American companies are largely focused on proprietary, closed models.
Who is involved? Key players include Chinese companies like deepseek and Moonshot AI,which are developing high-performing open-
