AINEWS
U.S. AI Chip Export Curbs Now Track Corporate Owners, Not Shipping Addresses
A May 31 Commerce Department rule requires export licenses for any China or Macau headquartered entity, no matter where its affiliates ship.
A May 31 Commerce Department rule requires export licenses for any China or Macau headquartered entity, no matter where its affiliates ship.
The U.S. export control regime for advanced semiconductors has spent several years learning a simple lesson: where a chip ships tells you nothing about who is really buying it. On May 31, 2026, the Commerce Department's Bureau of Industry and Security turned that lesson into a rule.
New enforcement guidance from BIS requires export licenses for any entity headquartered in China or Macau, regardless of where its affiliates are physically located or where the goods eventually cross a border. The change redefines who counts as a restricted buyer, not just which destinations count as restricted territory.
That is a deeper intervention than it first looks. Earlier rounds of chip controls operated on a destination model: chip goes to China, license required. Sophisticated buyers responded by routing orders through subsidiaries in Singapore, Malaysia, and elsewhere, using corporate shells as legal firewalls. The May 31 guidance rejects that pattern. If the parent is Chinese, the license is required. If the parent is in Macau, the license is required. The shipping address on the purchase order is no longer the controlling fact.
Reporting from Al Jazeera confirms the implementation, noting the rule covers advanced semiconductors usable for AI training and supercomputing, a broader scope than the leading-edge nodes earlier restrictions targeted.
Legal analysis from Hogan Lovells walks through the new license requirement: companies now have to audit the corporate structure of every customer they have been treating as a third-country buyer, not just the shipping label on each order. For a U.S. chipmaker like Nvidia, whose high-end AI processors have been the most-coveted target of subsidiary-driven procurement, this is the practical difference between chasing shipping manifests and auditing cap tables.
Commerce Secretary Howard Lutnick has publicly flagged his concern that some advanced chip-making equipment may have already reached China through these channels. The new guidance is, in part, an acknowledgment that destination-based enforcement had been leaking.
What the guidance is not, however, is a clean, unified position inside the U.S. government. The same TechTimes reporting notes that Trump administration officials are publicly disputing whether a real "loophole" ever existed, a posture that effectively contests the premise of the rule they just issued. Whether that is bureaucratic face-saving or a substantive policy disagreement will shape how aggressively BIS enforces the guidance in its early months.
The targets are not subtle. Chinese firms including Huawei and SMIC (Semiconductor Manufacturing International Corporation, China's largest contract chipmaker) sit inside the scope of the May 31 rule, and the structural shift reaches the entire Nvidia and AMD product line that touches AI training and supercomputing.
The mechanism matters because it changes what compliance actually means. U.S. chipmakers can no longer satisfy export controls by mapping geographies. They have to map corporate ownership. That is a quieter, harder discipline, and the regulatory regime that emerges from the May 31 rule will be the one that decides whether closing the subsidiary channel was a real fix or simply the opening move in the next round of corporate-structure evasion.