The United States has intensified its export restrictions on semiconductors, specifically targeting overseas subsidiaries of Chinese companies. This move aims to close existing loopholes in the control framework governing advanced artificial intelligence (AI) chips, reflecting the escalating technological competition with China.
On Sunday, the US Department of Commerce issued a guidance notice clarifying that stringent licensing requirements for exporting advanced AI chips now apply universally to all companies headquartered in or owned by parent firms based in China. This directive was released by the Bureau of Industry and Security (BIS) in response to increasing inquiries about enforcement following policy shifts initiated during the Trump administration.
Notably, this clarification follows the abandonment of the “Framework for Artificial Intelligence Diffusion,” a Biden-era initiative designed to establish a global licensing system for advanced AI processors. That framework had proposed strict export limits for most countries except close US allies.
The earlier proposal faced strong resistance from leading technology companies such as Nvidia, which cautioned that it could hinder global innovation and international collaboration. Subsequently, the Trump administration withdrew the framework, citing regulatory complexities and potential diplomatic repercussions.
However, the withdrawal created uncertainty regarding the status of existing export controls. The BIS has now confirmed that restrictions on Chinese-owned entities remain fully effective regardless of their location worldwide.
Nvidia, whose high-end Blackwell graphics processing units (GPUs) are already prohibited from direct export to China, stated it is complying with the updated guidance. Other major semiconductor manufacturers, including AMD, Intel, and Taiwan Semiconductor Manufacturing Company (TSMC), have yet to issue comments.
This development highlights the persistent tensions between Washington and Beijing over advanced semiconductor technologies, which are widely regarded as crucial for future leadership in artificial intelligence and high-performance computing.
Former officials and industry experts have noted that the regulatory ambiguity following the policy rollback may have enabled indirect acquisition of advanced chips through overseas subsidiaries, prompting demands for clearer enforcement measures.
Despite the firmer stance, the US has recently demonstrated selective leniency by permitting limited exports of less advanced chip models to China under specific licensing arrangements.
