On December 5, John Thornhill, who serves as the Innovation Content Editor at the Financial Times, penned an article suggesting that, in contrast to China's focus on developing lightweight and cost-effective open models, the United States' pursuit of large-scale, closed-source models might be steering down the wrong path.

The Trump administration initiated the 'Genesis Project' with the aim of fostering the growth of the private AI sector by facilitating the sharing of public datasets and computing resources. However, major tech firms in the U.S. are inclined to develop sizable, proprietary 'closed-weight' models. Examples include ChatGPT and Gemini. On the other hand, Chinese AI companies show a preference for smaller, more affordable 'open-weight' models, such as DeepSeek and Alibaba's QianWen, which developers can more readily adapt for their needs.

Michael Power, a former global strategist at the investment firm Ninety One, posits that the U.S.'s substantial investment in massive closed AI models represents a 'catastrophic strategic blunder.' The latest research conducted by MIT and Hugging Face reveals that China's open-weight models have already eclipsed their U.S. equivalents in terms of global adoption and popularity.