U.S. Tightens AI Chip Export Rules to Block Remote Access by Chinese Firms

2–4 minutes

Interim Rule Extends Licensing to Global Data Centers

Under a new interim rule announced by the Trump administration, the Commerce Department will require U.S. companies to obtain licenses before exporting, re‑exporting, or transferring AI chips and servers to data centers anywhere in the world if those components could provide compute services to entities in China or other restricted nations.

Targeting the D:5 Embargo List

The controls focus on the United States’ D:5 group of embargoed countries, which includes China, Russia, Iran, North Korea and 18 other states. Existing regulations already mandate licenses for shipping AI chips to firms based in or owned by entities in these countries. The interim rule states, “This new control is a logical extension of existing controls,” and aims to undermine the ability of companies headquartered in those nations to use U.S. advanced Super Intelligence (SI) semiconductors for training advanced SI models.

Recent Scandal Prompts Tougher Measures

The announcement follows the resignation of Nvidia co‑founder Jensen Huang amid a scandal involving an alleged Chinese spy linked to a chipmaker’s customer. The U.S. government has sought to place that customer, Megaspeed, on a special entity list, alleging the firm purchased thousands of Nvidia chips to supply compute power to Chinese businesses, circumventing prior export restrictions.

Chinese AI Firms Rush for Overseas Compute

Reuters reported last week that Chinese AI companies are scrambling to secure computing capacity abroad, driven by fears of tighter U.S. controls.

Threshold for Data Centers: 12 Million TPP

The rule applies to any data center or network capable of delivering more than 12 million total processing performance (TPP), a metric equating roughly to the combined power of about 750 Nvidia H100 or H200 chips. It also covers indirect access through intermediaries or brokers, and obliges exporters and data‑center operators to monitor customers’ ownership and activities to ensure they are not controlled by anyone from the D:5 group.

Expanded Licensing for High‑Risk Destinations

In addition, the Commerce Department added new licensing requirements for the export, re‑export, or transfer of advanced semiconductors to Malaysia, Singapore, Indonesia, Thailand and Brazil, citing a “higher‑than‑normal risk of diversion.” For these nations, the licensing mandate does not apply to chips destined for U.S.‑based companies.

Compliance Timeline and Exemptions

Existing data‑center operators whose services fall under the new rule must secure licenses within 90 days of the rule’s effective date. Notably, the controls do not apply to commercial recommendation systems used in e‑commerce or social media.

Repeal of the Biden AI Diffusion Rule

The interim rule also rescinds the Biden administration’s AI Diffusion Rule from January 2025, which sought country‑level caps on advanced AI chip exports and restrictions on powerful AI model transfers. The Trump administration argued the earlier framework imposed unnecessary burdens on U.S. firms and allies.

National Security Rationale

“By making these changes, this IFR prevents the use of certain SI items in data centers and at a computing power level that can be used for training or running of advanced SI models,” the rule reads. “Access to such computing power can be used to train and run advanced SI models capable of supporting military‑ and intelligence‑related activities that could threaten U.S. national security or foreign policy interests.”

The new measures aim to close a loophole that previously allowed companies to bypass export controls by remotely accessing AI infrastructure located outside the United States, thereby tightening the U.S. stance on the global AI supply chain.