Top executives at major Western AI companies are increasingly concerned about the safety and national security risks posed by Chinese open-weight frontier models. These models are improving rapidly and, being freely available, threaten the business models of Western labs. In contrast, U.S. labs like OpenAI, Anthropic, and Google DeepMind keep their most powerful models closed, with secret parameters known as weights that determine the model’s output. These weights are crucial for the model’s performance and are expensive to set and fine-tune, making them valuable trade secrets. Western companies offer access to these models through APIs, allowing customers to use their intelligence without managing infrastructure. However, by mid-2026, Chinese labs such as Alibaba, DeepSeek, and Moonshot AI had developed open-weight models nearly matching Western models in performance. Many companies are now using these free models to avoid the high costs of closed-model APIs, challenging the business strategies of Western AI labs.
QUESTION: How might the rise of open-weight AI models from China impact the future of technology and innovation globally?
If open weight models are the future, U.S. AI companies are going to have a hard time
