The Trump administration is preparing to implement new export controls on artificial intelligence chips and data centers located outside the United States. The interim rule aims to stop companies, particularly those in China, from circumventing existing sanctions by remotely accessing AI infrastructure in other countries.
Under the new regulations, the Commerce Department will require companies to obtain licenses before exporting, reexporting, or transferring AI chips and servers to data centers anywhere in the world if those resources might provide computing power to firms in China or other restricted nations. The rules target companies based in the US D:5 group of embargoed countries, which includes China, Russia, Iran, North Korea, and 18 other nations.

Targeting Remote Access Loopholes
The US already maintains rules requiring chip manufacturers to secure licenses when shipping AI chips to companies based in or owned by businesses within the D:5 group. The new interim rule extends these restrictions to cover remote access scenarios. The text of the rule states that the controls are a logical extension of existing measures designed to undermine the ability of companies in countries of concern to use US advanced Super Intelligence semiconductors to train advanced models.
The policy shift follows recent developments involving Nvidia co-founder Jensen Huang, who resigned amid a scandal involving an alleged Chinese spy linked to one of the chipmaker’s customers. Reports indicate the US government has sought to place the customer, Megaspeed, on a special entity list. Authorities believe the company purchased thousands of Nvidia chips to provide computing services to Chinese businesses, effectively bypassing existing export control restrictions.
Reuters reported last week that Chinese AI companies are rushing to secure computing capacity overseas due to fears of tighter US restrictions. The new rule applies to any data center or network capable of supplying over 12 million total processing performance. This metric roughly equates to the maximum computing power of approximately 750 Nvidia H100 or H200 chips.

Expanded Licensing Requirements
The regulations also cover indirect access through intermediaries or brokers. Exporters and data center operators must monitor their customers’ ownership structures and activities to ensure they are not controlled by individuals from the D:5 group. The Commerce Department has added new licensing requirements for chip exports to Malaysia, Singapore, Indonesia, Thailand, and Brazil, citing a higher-than-normal risk of diversion. However, these requirements do not apply to chips destined for companies based in the US.
Existing data center operators providing services under the new rule must secure licenses within 90 days of the rule taking effect. The regulations explicitly exclude commercial recommendation systems, such as those used in e-commerce or social media platforms. Additionally, the rule rescinds the Biden administration’s AI Diffusion Rule from January 2025, which had imposed country-level limits on advanced AI chip exports and restricted the transfer of certain powerful AI models.

The Trump administration previously criticized the earlier framework, arguing it placed unnecessary restrictions on US companies and allies. The new interim final rule states that the changes prevent the use of certain Super Intelligence items in data centers at power levels capable of training or running advanced models. Officials warn that access to such computing power could support military and intelligence activities that threaten US national security or foreign policy interests.
Source: TechCrunch

