The US Department of Commerce has moved to close a significant loophole that may have allowed America’s most advanced AI chips to reach Chinese companies through their overseas subsidiaries for nearly a year. The unexpected guidance, posted on Sunday, suggests that sophisticated processors from Nvidia and AMD have been making their way to Chinese AI firms based in countries like Malaysia despite broader US efforts to restrict semiconductor access.
The development highlights a critical gap in America’s chip export controls that may have undermined months of policy designed to limit China’s AI development capabilities. Industry sources estimate that hundreds of thousands of advanced chips could have been exported during this period, representing a substantial flow of cutting-edge technology to Chinese entities.
The Commerce Department’s weekend guidance now enforces license requirements for advanced chips going to any entity headquartered in China, regardless of where those entities are physically located. This closes what experts describe as a major oversight in the export control framework that was inadvertently created when the Trump administration chose not to enforce certain AI chip regulations in May 2024.
The loophole emerged when the Commerce Department announced it would not enforce the AI Diffusion rule that had been issued in the final days of the Biden administration. That rule was designed to govern global access to AI chips, but its suspension left overseas subsidiaries of Chinese companies free to purchase advanced processors without licenses.
Chris McGuire, a technology expert and former State Department official, called the situation “a HUGE problem” in social media posts, noting that Chinese companies have likely been buying Nvidia’s latest Blackwell chips at scale through this regulatory gap. The affected chips include some of the world’s most sophisticated AI processors:
- Nvidia’s Rubin and Blackwell processors
- AMD’s MI350x chips
- Other advanced computing components used for AI development
The scale of potential exports remains unclear, though one chip industry source with supply chain knowledge estimated the number could reach hundreds of thousands of units. This represents a significant volume of advanced technology that was intended to be restricted under US export controls targeting Chinese AI development.
The timing of this guidance is particularly significant as it comes amid escalating technology competition between the US and China. American policymakers have been working to prevent Chinese companies from accessing the semiconductors needed to develop advanced AI capabilities, viewing this as a national security priority. The discovery of this loophole suggests that these efforts may have been less effective than intended.
In an unusual twist, the new guidance does not require existing users to stop using chips that have already been exported or to cut off servicing of advanced computing equipment like servers. This means that Chinese entities that obtained chips during the loophole period can continue using them, though new acquisitions will now face licensing requirements.
Neither Nvidia, AMD, nor the Commerce Department immediately responded to requests for comment about the guidance or the potential scale of chip exports during the regulatory gap. The situation underscores the complexity of implementing effective export controls in a global technology supply chain where companies often operate through multiple international subsidiaries.
The development also raises questions about oversight and coordination in US technology policy, particularly during transitions between administrations. The gap appears to have resulted from policy decisions made during the early days of the Trump administration’s return to office, highlighting how changes in regulatory enforcement can have significant unintended consequences in the fast-moving semiconductor industry.
