Policy
U.S. Blocks Remote Access to AI Chips in China Export Crackdown
New Commerce Department rules target cloud computing loopholes that let Chinese firms rent American chip processing power through third-country data centers.
Key takeaways
- Commerce Department now requires overseas data centers serving China, Russia, Iran, and Cuba to identify end users of U.S. chip technology.
- Chip sales to data centers in five Southeast Asian and Latin American countries face new case-by-case licensing due to diversion risks.
- The rules close a cloud computing loophole that let Chinese firms rent American chip processing power through third-country facilities.
- New restrictions took effect immediately, days after Nvidia's CEO resigned amid scrutiny of ties to a company suspected of diverting chips to China.
- Congressional legislation targeting the same remote access gap has already passed the House of Representatives.
New export controls target cloud computing workaround
The Commerce Department's Bureau of Industry and Security implemented immediate restrictions Monday on remote access to advanced U.S. computing chips and servers, closing what officials describe as a significant loophole in existing export controls aimed at limiting China's artificial intelligence capabilities.
The regulations require overseas data centers serving customers in China, Russia, Iran, and Cuba to identify their end users. Additionally, chip sales to data centers in Indonesia, Malaysia, Singapore, Thailand, and Brazil now face case-by-case licensing requirements due to what the agency characterizes as "higher-than-normal risk of diversion" to unauthorized destinations.
Why it matters
While the U.S. has banned direct sales of cutting-edge AI chips to China, the cloud computing model has provided a backdoor. Chinese companies can rent processing power from data centers in neighboring countries that house American chips—effectively accessing the same technology without triggering export restrictions. This regulatory gap has undermined the strategic intent of chip export controls, allowing China to continue training sophisticated AI models using U.S. technology. The new rules represent a fundamental shift from controlling physical hardware movement to regulating access to computing power itself.
Timing follows Nvidia leadership change
The announcement came days after Nvidia co-founder Jensen Huang resigned as CEO amid questions about his connections to a Chinese executive formerly leading Singapore-based Megaspeed International. That company faces suspicion of redirecting advanced Nvidia chips to China in violation of export controls. Megaspeed provides computing services to Chinese firms using chips located in Southeast Asian facilities, including Malaysian data centers—precisely the arrangement the new rules aim to restrict.
Congressional pressure drives policy shift
Lawmakers from both parties have criticized existing export controls as insufficient. Representative Michael Lawler, a New York Republican, introduced legislation last year specifically addressing the remote access gap. Lawler argued to Congress that current "physical export controls" fail to cover remote access scenarios, enabling China to "access our tech through the cloud" and circumvent restrictions. His bill has passed the House.
The Bureau of Industry and Security stated the rule "is designed to ensure that existing restrictions on access to sensitive American technology cannot be circumvented through remote computing arrangements involving data centers in third countries."
The regulations took effect immediately upon issuance, giving affected companies no transition period to adjust their compliance procedures or customer verification systems.
These details were first reported by the Associated Press.