U.S. Commerce closes Nvidia remote access loophole

fortune.com3 min read

The author argues that the global expansion of U.S. cloud and compute infrastructure creates a paradox by accelerating the adoption and revenue of Chinese AI models, particularly open-weight large language models. While Washington seeks to dominate global AI with U.S.-designed chips, U.S.-owned data centers and inference providers routinely serve Chinese LLMs to international users. The author contends that instead of restricting allied access, the U.S. should counter this trend by deliberately supporting and deploying competitive American open-weight models. Open article →