The Trump administration announced tighter US AI chip export controls on November 16, targeting overseas data centers that provide advanced computing power to Chinese companies and other restricted customers.
The Commerce Department’s new rules would require licenses for certain semiconductor shipments and remote computing arrangements. The restrictions aim to prevent companies from bypassing export controls by renting powerful AI processors outside China.
US AI Chip Export Controls Target Remote Computing
Existing restrictions limit direct shipments of advanced American semiconductors to China. However, Chinese companies can still rent computing capacity from data centers operating in other countries.
The Commerce Department’s interim final rule extends licensing requirements to qualifying overseas data centers serving customers in restricted jurisdictions.
The requirements target facilities providing more than 12 million total processing performance units. That represents approximately the computing capacity of 750 Nvidia H100 or H200 processors.
Operators must examine customer ownership, including indirect control by companies headquartered in restricted countries. Commercial recommendation systems used by retailers and social platforms are exempt from the new remote-access requirement.
Five Countries Face Additional Chip Licensing Requirements
The rules also introduce additional licensing requirements for advanced chip shipments to Singapore, Malaysia, Indonesia, Thailand and Brazil.
U.S.-headquartered companies and qualifying American-owned subsidiaries receive exemptions when they meet specified reporting requirements.
The Commerce Department will review applications on a case-by-case basis. Existing data centers covered by the remote-access provision generally receive 90 days from the rule’s effective date to obtain authorization.
The published rule draft did not specify its final Federal Register effective date.
Megaspeed Investigation Adds Pressure on Export Enforcement
The announcement follows scrutiny of Singapore-based Megaspeed, a company linked to allegations involving former Nvidia CEO Jensen Huang.
Treasury Secretary Scott Bessent said several government departments supported restricting Megaspeed months earlier. However, the Commerce Department had not acted, according to The Wall Street Journal.
Megaspeed has denied illegally diverting restricted processors to China. The allegations remain under investigation.
TechAmerica previously examined reported Nvidia chip shipments linked to Chinese computing customers, including questions about intermediaries and export compliance.
Chinese AI Companies Seek Overseas Computing Capacity
Chinese technology companies had already begun seeking additional computing capacity abroad amid concerns about tighter American restrictions.
Reuters reported that several companies approached overseas data center operators about reserving Nvidia-powered infrastructure. Some smaller AI laboratories were considering providers in Southeast Asia.
The restrictions could complicate those arrangements by requiring additional authorization and customer verification.
The administration is also formally rescinding the previous AI Diffusion Rule, which was never enforced. The replacement focuses on chip destinations, end users and remote computing access rather than broad worldwide licensing.
How the requirements affect existing cloud contracts will depend on customer ownership, computing capacity and licensing decisions.










