The United States is exploring technologies to track advanced artificial intelligence (AI) chips and prevent their unauthorized export, particularly to China. Despite restrictions introduced in 2022, sophisticated processors continue reaching prohibited destinations, raising concerns about enforcement and national security, tells IEEE Spectrum.
In October 2026, U.S. authorities arrested a man accused of illegally shipping more than $300 million worth of GPU-equipped servers to China using documentation intended for Malaysia and Singapore. The incident highlights the growing need for reliable chip-tracking mechanisms.
One proposed solution is ping-based location verification (PLV), which determines a chip’s approximate location by measuring communication delays between the processor and trusted servers worldwide. Researchers at Georgetown University’s Center for Security and Emerging Technology (CSET) suggest integrating this functionality into new chips or adding it to existing processors through firmware updates.
Another approach involves cryptographic sovereignty certificates. Developed by secure computing company Lucid Computing, these credentials use hardware-based security and digital signatures to verify a processor’s physical location at a particular time.
Researchers are also investigating GPU fingerprinting, which identifies individual processors through unique manufacturing characteristics. A proof-of-concept study demonstrated fingerprinting techniques using Nvidia H200 GPUs, potentially reducing reliance on cryptographic keys that attackers could compromise.
However, these technologies face limitations. PLV cannot locate inactive chips during storage or transportation, while sophisticated attackers might manipulate verification signals. Researchers therefore recommend combining electronic monitoring with physical inspections.
Implementation costs and potential security vulnerabilities present additional challenges. Moreover, location tracking cannot prevent Chinese companies from legally accessing advanced processors through overseas cloud services.
Although mandatory tracking legislation remains under consideration, researchers expect stricter verification requirements as AI computing becomes increasingly important to national security.
For semiconductor manufacturers and data center operators, these developments could introduce new hardware security requirements, compliance responsibilities, and verification technologies throughout the AI computing supply chain.
