Former Hugging Face Asia-Pacific lead Tiezhen Wang argued in a Rest of World interview that China's open-source AI strategy is reshaping the U.S.-China race, with Chinese labs releasing free, downloadable models while OpenAI and Anthropic keep their code closed. Wang framed the divide as collaborative rather than zero-sum, noting that DeepSeek's reinforcement-learning approach has become a default for many U.S. labs and that Chinese open weights often run on U.S. hardware. On the contested practice of distillation, training one model on another's outputs, he called it a neutral research technique and pointed out that Elon Musk recently admitted xAI distilled from OpenAI, questioning why firms that scraped the internet for training data now object to others reusing their outputs. He also flagged a monetization risk, arguing some Chinese labs may retreat from open-source unless capital markets and license changes let them profit.
