According to TechCrunch, the rule requires a license to export, reexport or transfer AI chips and servers to data centers anywhere in the world if they may provide compute to companies based in the D:5 group of countries, which includes China, Russia, Iran, North Korea and 18 others. Until now, a license was needed only to ship chips to companies based in, or owned by companies in, those countries. The rule covers data centers and networks with more than 12 million TPP, the computing-power metric used in export controls, roughly the capacity of about 750 Nvidia H100 or H200 chips. It also covers indirect access through brokers, and requires exporters and data center operators to check their customers' ownership.
The rule also adds license requirements for advanced chip exports to Malaysia, Singapore, Indonesia, Thailand and Brazil, which the Commerce Department says "present a higher-than-normal risk of diversion." The requirement does not apply to chips destined for US companies in those countries. Existing data center operators covered by the rule must obtain licenses within 90 days of it taking effect. Commercial recommendation systems, such as those used in e-commerce and social media, are excluded.
The move comes days after Jensen Huang resigned from Nvidia in an affair linked to the customer Megaspeed. According to the report, the administration is considering placing it on a special list, on suspicion that it bought thousands of chips to provide compute to Chinese companies. It has not yet been published exactly when the rule takes effect.